commit 1dc26557697ec990ef141e2f8eab033198f5a343 Author: ModelHub XC Date: Thu Jul 30 06:14:18 2026 +0800 初始化项目，由ModelHub XC社区提供模型 Model: Neelectric/Llama-3.1-8B-Instruct_SFT_sciencev00.06 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..73bb4ce --- /dev/null +++ b/README.md @@ -0,0 +1,59 @@ +--- +base_model: meta-llama/Llama-3.1-8B-Instruct +datasets: Neelectric/Replay_0.02.MoT_science.wildguardmix_reasoning.Llama3_4096toks +library_name: transformers +model_name: Llama-3.1-8B-Instruct_SFT_sciencev00.06 +tags: +- generated_from_trainer +- trl +- open-r1 +- sft +licence: license +--- + +# Model Card for Llama-3.1-8B-Instruct_SFT_sciencev00.06 + +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/Replay_0.02.MoT_science.wildguardmix_reasoning.Llama3_4096toks](https://huggingface.co/datasets/Neelectric/Replay_0.02.MoT_science.wildguardmix_reasoning.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_sciencev00.06", 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/07r7qk01) + + +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..0e618e9 --- /dev/null +++ b/all_results.json @@ -0,0 +1,8 @@ +{ + "total_flos": 2.091827002923077e+19, + "train_loss": 0.9512690712848968, + "train_runtime": 21456.929, + "train_samples": 145693, + "train_samples_per_second": 13.58, + "train_steps_per_second": 0.849 +} \ 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..9e3245c --- /dev/null +++ b/model-00001-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8a236d6a0ca982b932e175ebac95b7cb8db1136fa346431f044ba79579085245 +size 4976698672 diff --git a/model-00002-of-00004.safetensors b/model-00002-of-00004.safetensors new file mode 100644 index 0000000..52a8041 --- /dev/null +++ b/model-00002-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dec35f710dab975cec1c8d3b6bf3411c0e1c299eddd129801979aa0eccd8dfee +size 4999802720 diff --git a/model-00003-of-00004.safetensors b/model-00003-of-00004.safetensors new file mode 100644 index 0000000..4296b92 --- /dev/null +++ b/model-00003-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5f7a8defd99c06d6e7a96262214ba3f1388af3e7f77c1cf40e2a87e021105e6f +size 4915916176 diff --git a/model-00004-of-00004.safetensors b/model-00004-of-00004.safetensors new file mode 100644 index 0000000..f08ef30 --- /dev/null +++ b/model-00004-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a036be36364be35edb133499cdb1e7b9745550181f6858ca13efa4d9a2eba0eb +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..0e618e9 --- /dev/null +++ b/train_results.json @@ -0,0 +1,8 @@ +{ + "total_flos": 2.091827002923077e+19, + "train_loss": 0.9512690712848968, + "train_runtime": 21456.929, + "train_samples": 145693, + "train_samples_per_second": 13.58, + "train_steps_per_second": 0.849 +} \ No newline at end of file diff --git a/trainer_state.json b/trainer_state.json new file mode 100644 index 0000000..9702229 --- /dev/null +++ b/trainer_state.json @@ -0,0 +1,163951 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 2.0, + "eval_steps": 500, + "global_step": 18212, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.00010981770261366133, + "grad_norm": 7.300184726715088, + "learning_rate": 0.0, + "loss": 1.4841, + "mean_token_accuracy": 0.6132287979125977, + "num_tokens": 25099.0, + "step": 1 + }, + { + "epoch": 0.00021963540522732265, + "grad_norm": 6.59010124206543, + "learning_rate": 5.48847420417124e-10, + "loss": 1.4857, + "mean_token_accuracy": 0.6123285293579102, + "num_tokens": 54691.0, + "step": 2 + }, + { + "epoch": 0.000329453107840984, + "grad_norm": 6.737997531890869, + "learning_rate": 1.097694840834248e-09, + "loss": 1.4471, + "mean_token_accuracy": 0.617257833480835, + "num_tokens": 81754.0, + "step": 3 + }, + { + "epoch": 0.0004392708104546453, + "grad_norm": 7.583641052246094, + "learning_rate": 1.646542261251372e-09, + "loss": 1.5597, + "mean_token_accuracy": 0.6015791893005371, + "num_tokens": 106869.0, + "step": 4 + }, + { + "epoch": 0.0005490885130683066, + "grad_norm": 7.799304485321045, + "learning_rate": 2.195389681668496e-09, + "loss": 1.5389, + "mean_token_accuracy": 0.6119979619979858, + "num_tokens": 128674.0, + "step": 5 + }, + { + "epoch": 0.000658906215681968, + "grad_norm": 7.749614238739014, + "learning_rate": 2.7442371020856205e-09, + "loss": 1.4745, + "mean_token_accuracy": 0.618381917476654, + "num_tokens": 151198.0, + "step": 6 + }, + { + "epoch": 0.0007687239182956292, + "grad_norm": 7.0824995040893555, + "learning_rate": 3.293084522502744e-09, + "loss": 1.4695, + "mean_token_accuracy": 0.6155315637588501, + "num_tokens": 177995.0, + "step": 7 + }, + { + "epoch": 0.0008785416209092906, + "grad_norm": 7.947245121002197, + "learning_rate": 3.841931942919868e-09, + "loss": 1.5504, + "mean_token_accuracy": 0.5963025093078613, + "num_tokens": 200991.0, + "step": 8 + }, + { + "epoch": 0.000988359323522952, + "grad_norm": 6.7032294273376465, + "learning_rate": 4.390779363336992e-09, + "loss": 1.3837, + "mean_token_accuracy": 0.6347914934158325, + "num_tokens": 228184.0, + "step": 9 + }, + { + "epoch": 0.0010981770261366132, + "grad_norm": 8.360169410705566, + "learning_rate": 4.939626783754116e-09, + "loss": 1.6381, + "mean_token_accuracy": 0.5832401514053345, + "num_tokens": 250279.0, + "step": 10 + }, + { + "epoch": 0.0012079947287502745, + "grad_norm": 6.226801872253418, + "learning_rate": 5.488474204171241e-09, + "loss": 1.5231, + "mean_token_accuracy": 0.6058598756790161, + "num_tokens": 286022.0, + "step": 11 + }, + { + "epoch": 0.001317812431363936, + "grad_norm": 7.2726006507873535, + "learning_rate": 6.0373216245883644e-09, + "loss": 1.5151, + "mean_token_accuracy": 0.6121420860290527, + "num_tokens": 309816.0, + "step": 12 + }, + { + "epoch": 0.0014276301339775973, + "grad_norm": 6.859964847564697, + "learning_rate": 6.586169045005488e-09, + "loss": 1.4952, + "mean_token_accuracy": 0.6140187978744507, + "num_tokens": 336700.0, + "step": 13 + }, + { + "epoch": 0.0015374478365912585, + "grad_norm": 7.316954612731934, + "learning_rate": 7.135016465422612e-09, + "loss": 1.5261, + "mean_token_accuracy": 0.6069676876068115, + "num_tokens": 362302.0, + "step": 14 + }, + { + "epoch": 0.0016472655392049198, + "grad_norm": 6.271126747131348, + "learning_rate": 7.683863885839736e-09, + "loss": 1.5064, + "mean_token_accuracy": 0.6066415309906006, + "num_tokens": 397378.0, + "step": 15 + }, + { + "epoch": 0.0017570832418185812, + "grad_norm": 7.505338191986084, + "learning_rate": 8.23271130625686e-09, + "loss": 1.4103, + "mean_token_accuracy": 0.6231832504272461, + "num_tokens": 421430.0, + "step": 16 + }, + { + "epoch": 0.0018669009444322424, + "grad_norm": 6.962411880493164, + "learning_rate": 8.781558726673984e-09, + "loss": 1.4518, + "mean_token_accuracy": 0.6165722608566284, + "num_tokens": 446882.0, + "step": 17 + }, + { + "epoch": 0.001976718647045904, + "grad_norm": 8.64083194732666, + "learning_rate": 9.330406147091108e-09, + "loss": 1.5179, + "mean_token_accuracy": 0.6098174452781677, + "num_tokens": 467017.0, + "step": 18 + }, + { + "epoch": 0.002086536349659565, + "grad_norm": 7.880673408508301, + "learning_rate": 9.879253567508231e-09, + "loss": 1.4998, + "mean_token_accuracy": 0.6075156927108765, + "num_tokens": 488601.0, + "step": 19 + }, + { + "epoch": 0.0021963540522732263, + "grad_norm": 7.994058609008789, + "learning_rate": 1.0428100987925357e-08, + "loss": 1.5382, + "mean_token_accuracy": 0.6067255735397339, + "num_tokens": 511408.0, + "step": 20 + }, + { + "epoch": 0.002306171754886888, + "grad_norm": 7.758239269256592, + "learning_rate": 1.0976948408342482e-08, + "loss": 1.5908, + "mean_token_accuracy": 0.5939858555793762, + "num_tokens": 534586.0, + "step": 21 + }, + { + "epoch": 0.002415989457500549, + "grad_norm": 7.017293453216553, + "learning_rate": 1.1525795828759604e-08, + "loss": 1.4032, + "mean_token_accuracy": 0.636762797832489, + "num_tokens": 559709.0, + "step": 22 + }, + { + "epoch": 0.0025258071601142102, + "grad_norm": 6.615419864654541, + "learning_rate": 1.2074643249176729e-08, + "loss": 1.5045, + "mean_token_accuracy": 0.612634539604187, + "num_tokens": 588324.0, + "step": 23 + }, + { + "epoch": 0.002635624862727872, + "grad_norm": 7.1192193031311035, + "learning_rate": 1.2623490669593852e-08, + "loss": 1.5105, + "mean_token_accuracy": 0.6047998666763306, + "num_tokens": 614067.0, + "step": 24 + }, + { + "epoch": 0.002745442565341533, + "grad_norm": 6.206730842590332, + "learning_rate": 1.3172338090010976e-08, + "loss": 1.3443, + "mean_token_accuracy": 0.6477615237236023, + "num_tokens": 643025.0, + "step": 25 + }, + { + "epoch": 0.0028552602679551946, + "grad_norm": 6.302875995635986, + "learning_rate": 1.3721185510428101e-08, + "loss": 1.5174, + "mean_token_accuracy": 0.6025481224060059, + "num_tokens": 675607.0, + "step": 26 + }, + { + "epoch": 0.0029650779705688557, + "grad_norm": 6.24977970123291, + "learning_rate": 1.4270032930845225e-08, + "loss": 1.5201, + "mean_token_accuracy": 0.6012755632400513, + "num_tokens": 707103.0, + "step": 27 + }, + { + "epoch": 0.003074895673182517, + "grad_norm": 7.035017013549805, + "learning_rate": 1.4818880351262348e-08, + "loss": 1.5015, + "mean_token_accuracy": 0.6110948324203491, + "num_tokens": 732869.0, + "step": 28 + }, + { + "epoch": 0.0031847133757961785, + "grad_norm": 7.842037200927734, + "learning_rate": 1.536772777167947e-08, + "loss": 1.5211, + "mean_token_accuracy": 0.610758900642395, + "num_tokens": 752292.0, + "step": 29 + }, + { + "epoch": 0.0032945310784098397, + "grad_norm": 7.430971145629883, + "learning_rate": 1.5916575192096597e-08, + "loss": 1.5086, + "mean_token_accuracy": 0.6177800893783569, + "num_tokens": 775820.0, + "step": 30 + }, + { + "epoch": 0.003404348781023501, + "grad_norm": 6.97247838973999, + "learning_rate": 1.646542261251372e-08, + "loss": 1.5114, + "mean_token_accuracy": 0.6115334033966064, + "num_tokens": 803838.0, + "step": 31 + }, + { + "epoch": 0.0035141664836371624, + "grad_norm": 6.8859639167785645, + "learning_rate": 1.7014270032930844e-08, + "loss": 1.498, + "mean_token_accuracy": 0.6090949773788452, + "num_tokens": 832838.0, + "step": 32 + }, + { + "epoch": 0.0036239841862508236, + "grad_norm": 7.348848342895508, + "learning_rate": 1.756311745334797e-08, + "loss": 1.4519, + "mean_token_accuracy": 0.6218241453170776, + "num_tokens": 857795.0, + "step": 33 + }, + { + "epoch": 0.0037338018888644848, + "grad_norm": 6.980248928070068, + "learning_rate": 1.811196487376509e-08, + "loss": 1.4439, + "mean_token_accuracy": 0.6187339425086975, + "num_tokens": 883809.0, + "step": 34 + }, + { + "epoch": 0.0038436195914781464, + "grad_norm": 6.404088020324707, + "learning_rate": 1.8660812294182216e-08, + "loss": 1.4576, + "mean_token_accuracy": 0.6154692769050598, + "num_tokens": 914129.0, + "step": 35 + }, + { + "epoch": 0.003953437294091808, + "grad_norm": 6.166694641113281, + "learning_rate": 1.920965971459934e-08, + "loss": 1.4271, + "mean_token_accuracy": 0.63253253698349, + "num_tokens": 943768.0, + "step": 36 + }, + { + "epoch": 0.004063254996705469, + "grad_norm": 6.842498302459717, + "learning_rate": 1.9758507135016463e-08, + "loss": 1.5078, + "mean_token_accuracy": 0.6083175539970398, + "num_tokens": 971842.0, + "step": 37 + }, + { + "epoch": 0.00417307269931913, + "grad_norm": 8.340742111206055, + "learning_rate": 2.030735455543359e-08, + "loss": 1.5403, + "mean_token_accuracy": 0.5988523960113525, + "num_tokens": 993069.0, + "step": 38 + }, + { + "epoch": 0.004282890401932792, + "grad_norm": 8.014483451843262, + "learning_rate": 2.0856201975850713e-08, + "loss": 1.5021, + "mean_token_accuracy": 0.6167664527893066, + "num_tokens": 1015112.0, + "step": 39 + }, + { + "epoch": 0.004392708104546453, + "grad_norm": 7.2254319190979, + "learning_rate": 2.1405049396267835e-08, + "loss": 1.4846, + "mean_token_accuracy": 0.61235111951828, + "num_tokens": 1041206.0, + "step": 40 + }, + { + "epoch": 0.004502525807160114, + "grad_norm": 6.4620537757873535, + "learning_rate": 2.1953896816684964e-08, + "loss": 1.5877, + "mean_token_accuracy": 0.5889983177185059, + "num_tokens": 1073406.0, + "step": 41 + }, + { + "epoch": 0.004612343509773776, + "grad_norm": 8.003437042236328, + "learning_rate": 2.2502744237102085e-08, + "loss": 1.5001, + "mean_token_accuracy": 0.6205312609672546, + "num_tokens": 1095307.0, + "step": 42 + }, + { + "epoch": 0.0047221612123874365, + "grad_norm": 7.14776086807251, + "learning_rate": 2.3051591657519207e-08, + "loss": 1.3972, + "mean_token_accuracy": 0.6407195329666138, + "num_tokens": 1119581.0, + "step": 43 + }, + { + "epoch": 0.004831978915001098, + "grad_norm": 8.040427207946777, + "learning_rate": 2.3600439077936336e-08, + "loss": 1.4997, + "mean_token_accuracy": 0.6153391599655151, + "num_tokens": 1141343.0, + "step": 44 + }, + { + "epoch": 0.00494179661761476, + "grad_norm": 8.163596153259277, + "learning_rate": 2.4149286498353458e-08, + "loss": 1.5909, + "mean_token_accuracy": 0.5888681411743164, + "num_tokens": 1162773.0, + "step": 45 + }, + { + "epoch": 0.0050516143202284204, + "grad_norm": 6.99888277053833, + "learning_rate": 2.469813391877058e-08, + "loss": 1.4903, + "mean_token_accuracy": 0.607560932636261, + "num_tokens": 1189992.0, + "step": 46 + }, + { + "epoch": 0.005161432022842082, + "grad_norm": 7.509421348571777, + "learning_rate": 2.5246981339187705e-08, + "loss": 1.3737, + "mean_token_accuracy": 0.6380921602249146, + "num_tokens": 1212166.0, + "step": 47 + }, + { + "epoch": 0.005271249725455744, + "grad_norm": 7.036658763885498, + "learning_rate": 2.579582875960483e-08, + "loss": 1.5161, + "mean_token_accuracy": 0.604224681854248, + "num_tokens": 1239976.0, + "step": 48 + }, + { + "epoch": 0.005381067428069404, + "grad_norm": 6.288334369659424, + "learning_rate": 2.6344676180021952e-08, + "loss": 1.4031, + "mean_token_accuracy": 0.629793643951416, + "num_tokens": 1270172.0, + "step": 49 + }, + { + "epoch": 0.005490885130683066, + "grad_norm": 6.973037242889404, + "learning_rate": 2.6893523600439077e-08, + "loss": 1.4181, + "mean_token_accuracy": 0.6310900449752808, + "num_tokens": 1297315.0, + "step": 50 + }, + { + "epoch": 0.0056007028332967276, + "grad_norm": 8.28620433807373, + "learning_rate": 2.7442371020856202e-08, + "loss": 1.4736, + "mean_token_accuracy": 0.6133103370666504, + "num_tokens": 1316820.0, + "step": 51 + }, + { + "epoch": 0.005710520535910389, + "grad_norm": 6.957880973815918, + "learning_rate": 2.7991218441273324e-08, + "loss": 1.3918, + "mean_token_accuracy": 0.638521134853363, + "num_tokens": 1341097.0, + "step": 52 + }, + { + "epoch": 0.00582033823852405, + "grad_norm": 7.497064113616943, + "learning_rate": 2.854006586169045e-08, + "loss": 1.4696, + "mean_token_accuracy": 0.6229697465896606, + "num_tokens": 1363682.0, + "step": 53 + }, + { + "epoch": 0.0059301559411377115, + "grad_norm": 7.375792980194092, + "learning_rate": 2.9088913282107574e-08, + "loss": 1.5114, + "mean_token_accuracy": 0.6082385778427124, + "num_tokens": 1388075.0, + "step": 54 + }, + { + "epoch": 0.006039973643751373, + "grad_norm": 6.480703830718994, + "learning_rate": 2.9637760702524696e-08, + "loss": 1.4627, + "mean_token_accuracy": 0.6162403225898743, + "num_tokens": 1419426.0, + "step": 55 + }, + { + "epoch": 0.006149791346365034, + "grad_norm": 5.99519681930542, + "learning_rate": 3.018660812294182e-08, + "loss": 1.5734, + "mean_token_accuracy": 0.595029354095459, + "num_tokens": 1457895.0, + "step": 56 + }, + { + "epoch": 0.006259609048978695, + "grad_norm": 7.867841720581055, + "learning_rate": 3.073545554335894e-08, + "loss": 1.5336, + "mean_token_accuracy": 0.5995015501976013, + "num_tokens": 1479785.0, + "step": 57 + }, + { + "epoch": 0.006369426751592357, + "grad_norm": 7.9879841804504395, + "learning_rate": 3.128430296377607e-08, + "loss": 1.4886, + "mean_token_accuracy": 0.6127592325210571, + "num_tokens": 1501506.0, + "step": 58 + }, + { + "epoch": 0.006479244454206018, + "grad_norm": 7.547282695770264, + "learning_rate": 3.1833150384193193e-08, + "loss": 1.4892, + "mean_token_accuracy": 0.6175497174263, + "num_tokens": 1525905.0, + "step": 59 + }, + { + "epoch": 0.006589062156819679, + "grad_norm": 6.555757522583008, + "learning_rate": 3.2381997804610315e-08, + "loss": 1.4592, + "mean_token_accuracy": 0.6229491233825684, + "num_tokens": 1555029.0, + "step": 60 + }, + { + "epoch": 0.006698879859433341, + "grad_norm": 6.867852210998535, + "learning_rate": 3.293084522502744e-08, + "loss": 1.4876, + "mean_token_accuracy": 0.6074279546737671, + "num_tokens": 1582640.0, + "step": 61 + }, + { + "epoch": 0.006808697562047002, + "grad_norm": 6.894315719604492, + "learning_rate": 3.3479692645444566e-08, + "loss": 1.511, + "mean_token_accuracy": 0.6172817945480347, + "num_tokens": 1610382.0, + "step": 62 + }, + { + "epoch": 0.006918515264660663, + "grad_norm": 8.738565444946289, + "learning_rate": 3.402854006586169e-08, + "loss": 1.4632, + "mean_token_accuracy": 0.6241319179534912, + "num_tokens": 1629910.0, + "step": 63 + }, + { + "epoch": 0.007028332967274325, + "grad_norm": 7.844120025634766, + "learning_rate": 3.4577387486278816e-08, + "loss": 1.5114, + "mean_token_accuracy": 0.6114199161529541, + "num_tokens": 1651147.0, + "step": 64 + }, + { + "epoch": 0.007138150669887986, + "grad_norm": 7.497685432434082, + "learning_rate": 3.512623490669594e-08, + "loss": 1.5883, + "mean_token_accuracy": 0.6030220985412598, + "num_tokens": 1675273.0, + "step": 65 + }, + { + "epoch": 0.007247968372501647, + "grad_norm": 6.963746070861816, + "learning_rate": 3.567508232711306e-08, + "loss": 1.481, + "mean_token_accuracy": 0.6078290939331055, + "num_tokens": 1699307.0, + "step": 66 + }, + { + "epoch": 0.007357786075115309, + "grad_norm": 6.741507053375244, + "learning_rate": 3.622392974753018e-08, + "loss": 1.4928, + "mean_token_accuracy": 0.6109813451766968, + "num_tokens": 1725536.0, + "step": 67 + }, + { + "epoch": 0.0074676037777289695, + "grad_norm": 7.7036309242248535, + "learning_rate": 3.677277716794731e-08, + "loss": 1.5555, + "mean_token_accuracy": 0.6022288799285889, + "num_tokens": 1748837.0, + "step": 68 + }, + { + "epoch": 0.007577421480342631, + "grad_norm": 6.824525356292725, + "learning_rate": 3.732162458836443e-08, + "loss": 1.5121, + "mean_token_accuracy": 0.6029000878334045, + "num_tokens": 1776152.0, + "step": 69 + }, + { + "epoch": 0.007687239182956293, + "grad_norm": 6.101449012756348, + "learning_rate": 3.787047200878156e-08, + "loss": 1.3882, + "mean_token_accuracy": 0.636203408241272, + "num_tokens": 1807692.0, + "step": 70 + }, + { + "epoch": 0.007797056885569954, + "grad_norm": 7.119207859039307, + "learning_rate": 3.841931942919868e-08, + "loss": 1.5624, + "mean_token_accuracy": 0.6105334758758545, + "num_tokens": 1830767.0, + "step": 71 + }, + { + "epoch": 0.007906874588183616, + "grad_norm": 7.204656600952148, + "learning_rate": 3.8968166849615804e-08, + "loss": 1.5975, + "mean_token_accuracy": 0.5902471542358398, + "num_tokens": 1856798.0, + "step": 72 + }, + { + "epoch": 0.008016692290797276, + "grad_norm": 7.217982292175293, + "learning_rate": 3.9517014270032926e-08, + "loss": 1.5595, + "mean_token_accuracy": 0.5946428179740906, + "num_tokens": 1881105.0, + "step": 73 + }, + { + "epoch": 0.008126509993410937, + "grad_norm": 7.262866020202637, + "learning_rate": 4.006586169045005e-08, + "loss": 1.4361, + "mean_token_accuracy": 0.6177399158477783, + "num_tokens": 1904702.0, + "step": 74 + }, + { + "epoch": 0.008236327696024599, + "grad_norm": 7.47563362121582, + "learning_rate": 4.061470911086718e-08, + "loss": 1.537, + "mean_token_accuracy": 0.599556565284729, + "num_tokens": 1927596.0, + "step": 75 + }, + { + "epoch": 0.00834614539863826, + "grad_norm": 6.728052139282227, + "learning_rate": 4.1163556531284305e-08, + "loss": 1.4623, + "mean_token_accuracy": 0.6149227023124695, + "num_tokens": 1955364.0, + "step": 76 + }, + { + "epoch": 0.008455963101251922, + "grad_norm": 6.2657599449157715, + "learning_rate": 4.1712403951701427e-08, + "loss": 1.4507, + "mean_token_accuracy": 0.6279863119125366, + "num_tokens": 1984201.0, + "step": 77 + }, + { + "epoch": 0.008565780803865584, + "grad_norm": 6.946040153503418, + "learning_rate": 4.226125137211855e-08, + "loss": 1.4884, + "mean_token_accuracy": 0.6144826412200928, + "num_tokens": 2008184.0, + "step": 78 + }, + { + "epoch": 0.008675598506479244, + "grad_norm": 6.395391941070557, + "learning_rate": 4.281009879253567e-08, + "loss": 1.5004, + "mean_token_accuracy": 0.6030008792877197, + "num_tokens": 2038259.0, + "step": 79 + }, + { + "epoch": 0.008785416209092905, + "grad_norm": 7.018989562988281, + "learning_rate": 4.335894621295279e-08, + "loss": 1.4296, + "mean_token_accuracy": 0.6258906722068787, + "num_tokens": 2061994.0, + "step": 80 + }, + { + "epoch": 0.008895233911706567, + "grad_norm": 6.306015491485596, + "learning_rate": 4.390779363336993e-08, + "loss": 1.5164, + "mean_token_accuracy": 0.6062215566635132, + "num_tokens": 2092926.0, + "step": 81 + }, + { + "epoch": 0.009005051614320228, + "grad_norm": 6.942089080810547, + "learning_rate": 4.445664105378705e-08, + "loss": 1.5408, + "mean_token_accuracy": 0.5946286916732788, + "num_tokens": 2118184.0, + "step": 82 + }, + { + "epoch": 0.00911486931693389, + "grad_norm": 7.8949408531188965, + "learning_rate": 4.500548847420417e-08, + "loss": 1.454, + "mean_token_accuracy": 0.623281717300415, + "num_tokens": 2138755.0, + "step": 83 + }, + { + "epoch": 0.009224687019547552, + "grad_norm": 6.078876972198486, + "learning_rate": 4.555433589462129e-08, + "loss": 1.408, + "mean_token_accuracy": 0.6341299414634705, + "num_tokens": 2166213.0, + "step": 84 + }, + { + "epoch": 0.009334504722161213, + "grad_norm": 7.041683197021484, + "learning_rate": 4.6103183315038415e-08, + "loss": 1.5678, + "mean_token_accuracy": 0.5897043943405151, + "num_tokens": 2191695.0, + "step": 85 + }, + { + "epoch": 0.009444322424774873, + "grad_norm": 7.376568794250488, + "learning_rate": 4.6652030735455537e-08, + "loss": 1.4365, + "mean_token_accuracy": 0.6254812479019165, + "num_tokens": 2213223.0, + "step": 86 + }, + { + "epoch": 0.009554140127388535, + "grad_norm": 6.581801891326904, + "learning_rate": 4.720087815587267e-08, + "loss": 1.4781, + "mean_token_accuracy": 0.6094871759414673, + "num_tokens": 2240955.0, + "step": 87 + }, + { + "epoch": 0.009663957830002196, + "grad_norm": 6.573953151702881, + "learning_rate": 4.7749725576289793e-08, + "loss": 1.4703, + "mean_token_accuracy": 0.607469379901886, + "num_tokens": 2267344.0, + "step": 88 + }, + { + "epoch": 0.009773775532615858, + "grad_norm": 7.737906455993652, + "learning_rate": 4.8298572996706915e-08, + "loss": 1.4186, + "mean_token_accuracy": 0.6236586570739746, + "num_tokens": 2287800.0, + "step": 89 + }, + { + "epoch": 0.00988359323522952, + "grad_norm": 5.817032814025879, + "learning_rate": 4.884742041712404e-08, + "loss": 1.3856, + "mean_token_accuracy": 0.6333210468292236, + "num_tokens": 2315854.0, + "step": 90 + }, + { + "epoch": 0.009993410937843181, + "grad_norm": 6.377894878387451, + "learning_rate": 4.939626783754116e-08, + "loss": 1.4655, + "mean_token_accuracy": 0.6204627752304077, + "num_tokens": 2342685.0, + "step": 91 + }, + { + "epoch": 0.010103228640456841, + "grad_norm": 5.613321781158447, + "learning_rate": 4.994511525795828e-08, + "loss": 1.3722, + "mean_token_accuracy": 0.6356606483459473, + "num_tokens": 2370881.0, + "step": 92 + }, + { + "epoch": 0.010213046343070502, + "grad_norm": 5.827258110046387, + "learning_rate": 5.049396267837541e-08, + "loss": 1.372, + "mean_token_accuracy": 0.6378515958786011, + "num_tokens": 2399821.0, + "step": 93 + }, + { + "epoch": 0.010322864045684164, + "grad_norm": 6.29862642288208, + "learning_rate": 5.104281009879254e-08, + "loss": 1.4606, + "mean_token_accuracy": 0.6106088757514954, + "num_tokens": 2426518.0, + "step": 94 + }, + { + "epoch": 0.010432681748297826, + "grad_norm": 6.47166109085083, + "learning_rate": 5.159165751920966e-08, + "loss": 1.436, + "mean_token_accuracy": 0.6174803376197815, + "num_tokens": 2454361.0, + "step": 95 + }, + { + "epoch": 0.010542499450911487, + "grad_norm": 7.056508541107178, + "learning_rate": 5.214050493962678e-08, + "loss": 1.4769, + "mean_token_accuracy": 0.6099861264228821, + "num_tokens": 2477069.0, + "step": 96 + }, + { + "epoch": 0.010652317153525149, + "grad_norm": 6.033366680145264, + "learning_rate": 5.2689352360043903e-08, + "loss": 1.4633, + "mean_token_accuracy": 0.610264003276825, + "num_tokens": 2506503.0, + "step": 97 + }, + { + "epoch": 0.010762134856138809, + "grad_norm": 6.873673439025879, + "learning_rate": 5.3238199780461025e-08, + "loss": 1.4758, + "mean_token_accuracy": 0.6197937726974487, + "num_tokens": 2528890.0, + "step": 98 + }, + { + "epoch": 0.01087195255875247, + "grad_norm": 6.953697681427002, + "learning_rate": 5.3787047200878154e-08, + "loss": 1.5018, + "mean_token_accuracy": 0.6087633967399597, + "num_tokens": 2552711.0, + "step": 99 + }, + { + "epoch": 0.010981770261366132, + "grad_norm": 6.6121087074279785, + "learning_rate": 5.433589462129528e-08, + "loss": 1.5005, + "mean_token_accuracy": 0.6040401458740234, + "num_tokens": 2578243.0, + "step": 100 + }, + { + "epoch": 0.011091587963979794, + "grad_norm": 6.62105131149292, + "learning_rate": 5.4884742041712404e-08, + "loss": 1.4806, + "mean_token_accuracy": 0.6158633232116699, + "num_tokens": 2604021.0, + "step": 101 + }, + { + "epoch": 0.011201405666593455, + "grad_norm": 6.636417388916016, + "learning_rate": 5.5433589462129526e-08, + "loss": 1.4753, + "mean_token_accuracy": 0.6141564249992371, + "num_tokens": 2629199.0, + "step": 102 + }, + { + "epoch": 0.011311223369207117, + "grad_norm": 7.092854976654053, + "learning_rate": 5.598243688254665e-08, + "loss": 1.5731, + "mean_token_accuracy": 0.5987552404403687, + "num_tokens": 2652233.0, + "step": 103 + }, + { + "epoch": 0.011421041071820778, + "grad_norm": 8.07008171081543, + "learning_rate": 5.653128430296377e-08, + "loss": 1.5067, + "mean_token_accuracy": 0.6080474853515625, + "num_tokens": 2671384.0, + "step": 104 + }, + { + "epoch": 0.011530858774434438, + "grad_norm": 6.9622721672058105, + "learning_rate": 5.70801317233809e-08, + "loss": 1.5254, + "mean_token_accuracy": 0.6035105586051941, + "num_tokens": 2695792.0, + "step": 105 + }, + { + "epoch": 0.0116406764770481, + "grad_norm": 6.628222465515137, + "learning_rate": 5.762897914379802e-08, + "loss": 1.3858, + "mean_token_accuracy": 0.6289123296737671, + "num_tokens": 2720150.0, + "step": 106 + }, + { + "epoch": 0.011750494179661761, + "grad_norm": 7.643450736999512, + "learning_rate": 5.817782656421515e-08, + "loss": 1.3795, + "mean_token_accuracy": 0.6352161765098572, + "num_tokens": 2739885.0, + "step": 107 + }, + { + "epoch": 0.011860311882275423, + "grad_norm": 8.139312744140625, + "learning_rate": 5.872667398463227e-08, + "loss": 1.4228, + "mean_token_accuracy": 0.6313039064407349, + "num_tokens": 2758648.0, + "step": 108 + }, + { + "epoch": 0.011970129584889085, + "grad_norm": 6.87192964553833, + "learning_rate": 5.927552140504939e-08, + "loss": 1.4304, + "mean_token_accuracy": 0.6221833825111389, + "num_tokens": 2783616.0, + "step": 109 + }, + { + "epoch": 0.012079947287502746, + "grad_norm": 7.545658588409424, + "learning_rate": 5.982436882546651e-08, + "loss": 1.4831, + "mean_token_accuracy": 0.6154478788375854, + "num_tokens": 2805721.0, + "step": 110 + }, + { + "epoch": 0.012189764990116406, + "grad_norm": 6.513121128082275, + "learning_rate": 6.037321624588364e-08, + "loss": 1.5071, + "mean_token_accuracy": 0.6078747510910034, + "num_tokens": 2834606.0, + "step": 111 + }, + { + "epoch": 0.012299582692730068, + "grad_norm": 6.430140018463135, + "learning_rate": 6.092206366630077e-08, + "loss": 1.4379, + "mean_token_accuracy": 0.6138849258422852, + "num_tokens": 2862019.0, + "step": 112 + }, + { + "epoch": 0.01240940039534373, + "grad_norm": 7.044624328613281, + "learning_rate": 6.147091108671789e-08, + "loss": 1.4124, + "mean_token_accuracy": 0.622858464717865, + "num_tokens": 2885326.0, + "step": 113 + }, + { + "epoch": 0.01251921809795739, + "grad_norm": 6.968350410461426, + "learning_rate": 6.201975850713501e-08, + "loss": 1.3674, + "mean_token_accuracy": 0.6380969285964966, + "num_tokens": 2908991.0, + "step": 114 + }, + { + "epoch": 0.012629035800571052, + "grad_norm": 6.221890449523926, + "learning_rate": 6.256860592755214e-08, + "loss": 1.414, + "mean_token_accuracy": 0.6249101758003235, + "num_tokens": 2935899.0, + "step": 115 + }, + { + "epoch": 0.012738853503184714, + "grad_norm": 6.053843021392822, + "learning_rate": 6.311745334796927e-08, + "loss": 1.3774, + "mean_token_accuracy": 0.6296281814575195, + "num_tokens": 2964705.0, + "step": 116 + }, + { + "epoch": 0.012848671205798374, + "grad_norm": 7.4735212326049805, + "learning_rate": 6.366630076838639e-08, + "loss": 1.4273, + "mean_token_accuracy": 0.6170215606689453, + "num_tokens": 2986407.0, + "step": 117 + }, + { + "epoch": 0.012958488908412035, + "grad_norm": 6.7352375984191895, + "learning_rate": 6.421514818880352e-08, + "loss": 1.435, + "mean_token_accuracy": 0.6137486696243286, + "num_tokens": 3010790.0, + "step": 118 + }, + { + "epoch": 0.013068306611025697, + "grad_norm": 6.870153427124023, + "learning_rate": 6.476399560922063e-08, + "loss": 1.3857, + "mean_token_accuracy": 0.6365429162979126, + "num_tokens": 3034036.0, + "step": 119 + }, + { + "epoch": 0.013178124313639359, + "grad_norm": 6.209249019622803, + "learning_rate": 6.531284302963776e-08, + "loss": 1.4531, + "mean_token_accuracy": 0.6109081506729126, + "num_tokens": 3062614.0, + "step": 120 + }, + { + "epoch": 0.01328794201625302, + "grad_norm": 5.581733226776123, + "learning_rate": 6.586169045005487e-08, + "loss": 1.2684, + "mean_token_accuracy": 0.6540181040763855, + "num_tokens": 3089903.0, + "step": 121 + }, + { + "epoch": 0.013397759718866682, + "grad_norm": 5.601731777191162, + "learning_rate": 6.6410537870472e-08, + "loss": 1.4869, + "mean_token_accuracy": 0.6060202121734619, + "num_tokens": 3119997.0, + "step": 122 + }, + { + "epoch": 0.013507577421480343, + "grad_norm": 5.953408718109131, + "learning_rate": 6.695938529088913e-08, + "loss": 1.3922, + "mean_token_accuracy": 0.6284226179122925, + "num_tokens": 3144944.0, + "step": 123 + }, + { + "epoch": 0.013617395124094003, + "grad_norm": 6.0721306800842285, + "learning_rate": 6.750823271130625e-08, + "loss": 1.3765, + "mean_token_accuracy": 0.6293390393257141, + "num_tokens": 3168661.0, + "step": 124 + }, + { + "epoch": 0.013727212826707665, + "grad_norm": 5.561980724334717, + "learning_rate": 6.805708013172338e-08, + "loss": 1.4077, + "mean_token_accuracy": 0.6164328455924988, + "num_tokens": 3197108.0, + "step": 125 + }, + { + "epoch": 0.013837030529321327, + "grad_norm": 5.212100982666016, + "learning_rate": 6.86059275521405e-08, + "loss": 1.3583, + "mean_token_accuracy": 0.6298313140869141, + "num_tokens": 3224176.0, + "step": 126 + }, + { + "epoch": 0.013946848231934988, + "grad_norm": 5.6212615966796875, + "learning_rate": 6.915477497255763e-08, + "loss": 1.2923, + "mean_token_accuracy": 0.6539316773414612, + "num_tokens": 3246547.0, + "step": 127 + }, + { + "epoch": 0.01405666593454865, + "grad_norm": 5.686472415924072, + "learning_rate": 6.970362239297475e-08, + "loss": 1.4132, + "mean_token_accuracy": 0.6188251972198486, + "num_tokens": 3271521.0, + "step": 128 + }, + { + "epoch": 0.014166483637162311, + "grad_norm": 5.632158279418945, + "learning_rate": 7.025246981339188e-08, + "loss": 1.4773, + "mean_token_accuracy": 0.6069074869155884, + "num_tokens": 3295440.0, + "step": 129 + }, + { + "epoch": 0.014276301339775971, + "grad_norm": 5.308712005615234, + "learning_rate": 7.0801317233809e-08, + "loss": 1.3387, + "mean_token_accuracy": 0.6469953656196594, + "num_tokens": 3321844.0, + "step": 130 + }, + { + "epoch": 0.014386119042389633, + "grad_norm": 4.765414237976074, + "learning_rate": 7.135016465422612e-08, + "loss": 1.3961, + "mean_token_accuracy": 0.6279177665710449, + "num_tokens": 3353172.0, + "step": 131 + }, + { + "epoch": 0.014495936745003294, + "grad_norm": 5.251784801483154, + "learning_rate": 7.189901207464325e-08, + "loss": 1.4649, + "mean_token_accuracy": 0.6107134819030762, + "num_tokens": 3379760.0, + "step": 132 + }, + { + "epoch": 0.014605754447616956, + "grad_norm": 5.269290447235107, + "learning_rate": 7.244785949506036e-08, + "loss": 1.4409, + "mean_token_accuracy": 0.6005305051803589, + "num_tokens": 3403762.0, + "step": 133 + }, + { + "epoch": 0.014715572150230618, + "grad_norm": 4.574446678161621, + "learning_rate": 7.299670691547749e-08, + "loss": 1.4286, + "mean_token_accuracy": 0.616790771484375, + "num_tokens": 3437490.0, + "step": 134 + }, + { + "epoch": 0.01482538985284428, + "grad_norm": 5.093397617340088, + "learning_rate": 7.354555433589462e-08, + "loss": 1.3726, + "mean_token_accuracy": 0.625381350517273, + "num_tokens": 3464129.0, + "step": 135 + }, + { + "epoch": 0.014935207555457939, + "grad_norm": 5.562347888946533, + "learning_rate": 7.409440175631174e-08, + "loss": 1.4255, + "mean_token_accuracy": 0.6106982231140137, + "num_tokens": 3487272.0, + "step": 136 + }, + { + "epoch": 0.0150450252580716, + "grad_norm": 5.5066962242126465, + "learning_rate": 7.464324917672886e-08, + "loss": 1.4552, + "mean_token_accuracy": 0.6099640130996704, + "num_tokens": 3511402.0, + "step": 137 + }, + { + "epoch": 0.015154842960685262, + "grad_norm": 4.9704742431640625, + "learning_rate": 7.519209659714599e-08, + "loss": 1.4161, + "mean_token_accuracy": 0.6098848581314087, + "num_tokens": 3538895.0, + "step": 138 + }, + { + "epoch": 0.015264660663298924, + "grad_norm": 4.555685520172119, + "learning_rate": 7.574094401756312e-08, + "loss": 1.4195, + "mean_token_accuracy": 0.621460497379303, + "num_tokens": 3569243.0, + "step": 139 + }, + { + "epoch": 0.015374478365912585, + "grad_norm": 4.9964094161987305, + "learning_rate": 7.628979143798024e-08, + "loss": 1.3576, + "mean_token_accuracy": 0.6325840353965759, + "num_tokens": 3595017.0, + "step": 140 + }, + { + "epoch": 0.015484296068526247, + "grad_norm": 5.335128307342529, + "learning_rate": 7.683863885839736e-08, + "loss": 1.317, + "mean_token_accuracy": 0.6360108852386475, + "num_tokens": 3617473.0, + "step": 141 + }, + { + "epoch": 0.015594113771139909, + "grad_norm": 5.377992153167725, + "learning_rate": 7.738748627881449e-08, + "loss": 1.3279, + "mean_token_accuracy": 0.630800724029541, + "num_tokens": 3640391.0, + "step": 142 + }, + { + "epoch": 0.01570393147375357, + "grad_norm": 5.108697414398193, + "learning_rate": 7.793633369923161e-08, + "loss": 1.3613, + "mean_token_accuracy": 0.6416325569152832, + "num_tokens": 3663846.0, + "step": 143 + }, + { + "epoch": 0.015813749176367232, + "grad_norm": 4.999919891357422, + "learning_rate": 7.848518111964874e-08, + "loss": 1.301, + "mean_token_accuracy": 0.6352896690368652, + "num_tokens": 3687353.0, + "step": 144 + }, + { + "epoch": 0.01592356687898089, + "grad_norm": 5.142717361450195, + "learning_rate": 7.903402854006585e-08, + "loss": 1.4031, + "mean_token_accuracy": 0.6207288503646851, + "num_tokens": 3711019.0, + "step": 145 + }, + { + "epoch": 0.01603338458159455, + "grad_norm": 5.12539005279541, + "learning_rate": 7.958287596048298e-08, + "loss": 1.4255, + "mean_token_accuracy": 0.6086678504943848, + "num_tokens": 3732093.0, + "step": 146 + }, + { + "epoch": 0.016143202284208215, + "grad_norm": 5.508090019226074, + "learning_rate": 8.01317233809001e-08, + "loss": 1.3369, + "mean_token_accuracy": 0.6386147737503052, + "num_tokens": 3753304.0, + "step": 147 + }, + { + "epoch": 0.016253019986821875, + "grad_norm": 4.175239562988281, + "learning_rate": 8.068057080131722e-08, + "loss": 1.3591, + "mean_token_accuracy": 0.625747799873352, + "num_tokens": 3784719.0, + "step": 148 + }, + { + "epoch": 0.016362837689435538, + "grad_norm": 5.125462532043457, + "learning_rate": 8.122941822173437e-08, + "loss": 1.3414, + "mean_token_accuracy": 0.6382431983947754, + "num_tokens": 3806930.0, + "step": 149 + }, + { + "epoch": 0.016472655392049198, + "grad_norm": 4.4893598556518555, + "learning_rate": 8.177826564215148e-08, + "loss": 1.3652, + "mean_token_accuracy": 0.633797287940979, + "num_tokens": 3834318.0, + "step": 150 + }, + { + "epoch": 0.01658247309466286, + "grad_norm": 4.58860445022583, + "learning_rate": 8.232711306256861e-08, + "loss": 1.43, + "mean_token_accuracy": 0.6069995164871216, + "num_tokens": 3863058.0, + "step": 151 + }, + { + "epoch": 0.01669229079727652, + "grad_norm": 4.776368618011475, + "learning_rate": 8.287596048298572e-08, + "loss": 1.2881, + "mean_token_accuracy": 0.6490373611450195, + "num_tokens": 3887396.0, + "step": 152 + }, + { + "epoch": 0.01680210849989018, + "grad_norm": 4.752491474151611, + "learning_rate": 8.342480790340285e-08, + "loss": 1.3429, + "mean_token_accuracy": 0.6378241777420044, + "num_tokens": 3912219.0, + "step": 153 + }, + { + "epoch": 0.016911926202503844, + "grad_norm": 4.466789245605469, + "learning_rate": 8.397365532381998e-08, + "loss": 1.3539, + "mean_token_accuracy": 0.6312689185142517, + "num_tokens": 3939545.0, + "step": 154 + }, + { + "epoch": 0.017021743905117504, + "grad_norm": 4.080368995666504, + "learning_rate": 8.45225027442371e-08, + "loss": 1.4001, + "mean_token_accuracy": 0.6207311153411865, + "num_tokens": 3973679.0, + "step": 155 + }, + { + "epoch": 0.017131561607731167, + "grad_norm": 4.398621559143066, + "learning_rate": 8.507135016465423e-08, + "loss": 1.3, + "mean_token_accuracy": 0.6415899991989136, + "num_tokens": 3998766.0, + "step": 156 + }, + { + "epoch": 0.017241379310344827, + "grad_norm": 4.707389831542969, + "learning_rate": 8.562019758507134e-08, + "loss": 1.3901, + "mean_token_accuracy": 0.6252470016479492, + "num_tokens": 4026565.0, + "step": 157 + }, + { + "epoch": 0.017351197012958487, + "grad_norm": 4.449059009552002, + "learning_rate": 8.616904500548847e-08, + "loss": 1.3131, + "mean_token_accuracy": 0.6381980776786804, + "num_tokens": 4050281.0, + "step": 158 + }, + { + "epoch": 0.01746101471557215, + "grad_norm": 4.702731609344482, + "learning_rate": 8.671789242590558e-08, + "loss": 1.38, + "mean_token_accuracy": 0.6232395172119141, + "num_tokens": 4075261.0, + "step": 159 + }, + { + "epoch": 0.01757083241818581, + "grad_norm": 5.142807960510254, + "learning_rate": 8.726673984632271e-08, + "loss": 1.4044, + "mean_token_accuracy": 0.6202396154403687, + "num_tokens": 4098468.0, + "step": 160 + }, + { + "epoch": 0.017680650120799474, + "grad_norm": 4.17048454284668, + "learning_rate": 8.781558726673985e-08, + "loss": 1.2923, + "mean_token_accuracy": 0.6482641100883484, + "num_tokens": 4127103.0, + "step": 161 + }, + { + "epoch": 0.017790467823413134, + "grad_norm": 4.270692348480225, + "learning_rate": 8.836443468715697e-08, + "loss": 1.3624, + "mean_token_accuracy": 0.630677342414856, + "num_tokens": 4153834.0, + "step": 162 + }, + { + "epoch": 0.017900285526026797, + "grad_norm": 4.59856653213501, + "learning_rate": 8.89132821075741e-08, + "loss": 1.3746, + "mean_token_accuracy": 0.6222448348999023, + "num_tokens": 4178367.0, + "step": 163 + }, + { + "epoch": 0.018010103228640457, + "grad_norm": 4.402196407318115, + "learning_rate": 8.946212952799121e-08, + "loss": 1.3122, + "mean_token_accuracy": 0.649791419506073, + "num_tokens": 4205623.0, + "step": 164 + }, + { + "epoch": 0.018119920931254117, + "grad_norm": 4.191536903381348, + "learning_rate": 9.001097694840834e-08, + "loss": 1.3215, + "mean_token_accuracy": 0.6406217217445374, + "num_tokens": 4235213.0, + "step": 165 + }, + { + "epoch": 0.01822973863386778, + "grad_norm": 4.014502048492432, + "learning_rate": 9.055982436882546e-08, + "loss": 1.4118, + "mean_token_accuracy": 0.6163216829299927, + "num_tokens": 4264987.0, + "step": 166 + }, + { + "epoch": 0.01833955633648144, + "grad_norm": 4.611756801605225, + "learning_rate": 9.110867178924259e-08, + "loss": 1.4257, + "mean_token_accuracy": 0.6123114824295044, + "num_tokens": 4292937.0, + "step": 167 + }, + { + "epoch": 0.018449374039095103, + "grad_norm": 5.502588748931885, + "learning_rate": 9.165751920965971e-08, + "loss": 1.2803, + "mean_token_accuracy": 0.640020489692688, + "num_tokens": 4312735.0, + "step": 168 + }, + { + "epoch": 0.018559191741708763, + "grad_norm": 4.982048034667969, + "learning_rate": 9.220636663007683e-08, + "loss": 1.3601, + "mean_token_accuracy": 0.6240072846412659, + "num_tokens": 4334667.0, + "step": 169 + }, + { + "epoch": 0.018669009444322426, + "grad_norm": 5.62700891494751, + "learning_rate": 9.275521405049396e-08, + "loss": 1.2646, + "mean_token_accuracy": 0.6510325074195862, + "num_tokens": 4352210.0, + "step": 170 + }, + { + "epoch": 0.018778827146936086, + "grad_norm": 5.838393211364746, + "learning_rate": 9.330406147091107e-08, + "loss": 1.334, + "mean_token_accuracy": 0.6355873942375183, + "num_tokens": 4370916.0, + "step": 171 + }, + { + "epoch": 0.018888644849549746, + "grad_norm": 4.5211076736450195, + "learning_rate": 9.38529088913282e-08, + "loss": 1.3718, + "mean_token_accuracy": 0.6184235215187073, + "num_tokens": 4399973.0, + "step": 172 + }, + { + "epoch": 0.01899846255216341, + "grad_norm": 4.207301616668701, + "learning_rate": 9.440175631174534e-08, + "loss": 1.3849, + "mean_token_accuracy": 0.6204959750175476, + "num_tokens": 4425616.0, + "step": 173 + }, + { + "epoch": 0.01910828025477707, + "grad_norm": 3.991732120513916, + "learning_rate": 9.495060373216246e-08, + "loss": 1.211, + "mean_token_accuracy": 0.6588553190231323, + "num_tokens": 4451964.0, + "step": 174 + }, + { + "epoch": 0.019218097957390733, + "grad_norm": 4.936097145080566, + "learning_rate": 9.549945115257959e-08, + "loss": 1.3038, + "mean_token_accuracy": 0.6438684463500977, + "num_tokens": 4472434.0, + "step": 175 + }, + { + "epoch": 0.019327915660004392, + "grad_norm": 3.815870523452759, + "learning_rate": 9.60482985729967e-08, + "loss": 1.3355, + "mean_token_accuracy": 0.6372572183609009, + "num_tokens": 4501645.0, + "step": 176 + }, + { + "epoch": 0.019437733362618052, + "grad_norm": 4.520309925079346, + "learning_rate": 9.659714599341383e-08, + "loss": 1.3307, + "mean_token_accuracy": 0.6366158127784729, + "num_tokens": 4524870.0, + "step": 177 + }, + { + "epoch": 0.019547551065231716, + "grad_norm": 3.7993950843811035, + "learning_rate": 9.714599341383095e-08, + "loss": 1.3543, + "mean_token_accuracy": 0.6282926797866821, + "num_tokens": 4553665.0, + "step": 178 + }, + { + "epoch": 0.019657368767845376, + "grad_norm": 3.801699638366699, + "learning_rate": 9.769484083424807e-08, + "loss": 1.2958, + "mean_token_accuracy": 0.6457051634788513, + "num_tokens": 4582433.0, + "step": 179 + }, + { + "epoch": 0.01976718647045904, + "grad_norm": 4.6142778396606445, + "learning_rate": 9.82436882546652e-08, + "loss": 1.359, + "mean_token_accuracy": 0.6246019005775452, + "num_tokens": 4604879.0, + "step": 180 + }, + { + "epoch": 0.0198770041730727, + "grad_norm": 3.667733669281006, + "learning_rate": 9.879253567508232e-08, + "loss": 1.3001, + "mean_token_accuracy": 0.6465510129928589, + "num_tokens": 4636491.0, + "step": 181 + }, + { + "epoch": 0.019986821875686362, + "grad_norm": 4.10603666305542, + "learning_rate": 9.934138309549945e-08, + "loss": 1.2858, + "mean_token_accuracy": 0.6434091925621033, + "num_tokens": 4667431.0, + "step": 182 + }, + { + "epoch": 0.020096639578300022, + "grad_norm": 3.927851676940918, + "learning_rate": 9.989023051591656e-08, + "loss": 1.3165, + "mean_token_accuracy": 0.6338636875152588, + "num_tokens": 4697010.0, + "step": 183 + }, + { + "epoch": 0.020206457280913682, + "grad_norm": 3.8883936405181885, + "learning_rate": 1.0043907793633369e-07, + "loss": 1.3247, + "mean_token_accuracy": 0.6361231803894043, + "num_tokens": 4727807.0, + "step": 184 + }, + { + "epoch": 0.020316274983527345, + "grad_norm": 4.070976734161377, + "learning_rate": 1.0098792535675082e-07, + "loss": 1.2847, + "mean_token_accuracy": 0.6487807631492615, + "num_tokens": 4750366.0, + "step": 185 + }, + { + "epoch": 0.020426092686141005, + "grad_norm": 5.554687023162842, + "learning_rate": 1.0153677277716795e-07, + "loss": 1.295, + "mean_token_accuracy": 0.6452535390853882, + "num_tokens": 4767516.0, + "step": 186 + }, + { + "epoch": 0.02053591038875467, + "grad_norm": 4.027348041534424, + "learning_rate": 1.0208562019758508e-07, + "loss": 1.2966, + "mean_token_accuracy": 0.6425517797470093, + "num_tokens": 4791772.0, + "step": 187 + }, + { + "epoch": 0.020645728091368328, + "grad_norm": 4.457411766052246, + "learning_rate": 1.0263446761800219e-07, + "loss": 1.3133, + "mean_token_accuracy": 0.639520525932312, + "num_tokens": 4816224.0, + "step": 188 + }, + { + "epoch": 0.02075554579398199, + "grad_norm": 4.357842922210693, + "learning_rate": 1.0318331503841932e-07, + "loss": 1.3067, + "mean_token_accuracy": 0.6417659521102905, + "num_tokens": 4842421.0, + "step": 189 + }, + { + "epoch": 0.02086536349659565, + "grad_norm": 4.8794331550598145, + "learning_rate": 1.0373216245883643e-07, + "loss": 1.4585, + "mean_token_accuracy": 0.6195328235626221, + "num_tokens": 4864519.0, + "step": 190 + }, + { + "epoch": 0.02097518119920931, + "grad_norm": 4.138430118560791, + "learning_rate": 1.0428100987925356e-07, + "loss": 1.3144, + "mean_token_accuracy": 0.6352958083152771, + "num_tokens": 4894869.0, + "step": 191 + }, + { + "epoch": 0.021084998901822975, + "grad_norm": 4.257168292999268, + "learning_rate": 1.0482985729967068e-07, + "loss": 1.3312, + "mean_token_accuracy": 0.6326555609703064, + "num_tokens": 4923179.0, + "step": 192 + }, + { + "epoch": 0.021194816604436634, + "grad_norm": 3.669776678085327, + "learning_rate": 1.0537870472008781e-07, + "loss": 1.3458, + "mean_token_accuracy": 0.6317281126976013, + "num_tokens": 4955491.0, + "step": 193 + }, + { + "epoch": 0.021304634307050298, + "grad_norm": 5.042634963989258, + "learning_rate": 1.0592755214050494e-07, + "loss": 1.4154, + "mean_token_accuracy": 0.6146949529647827, + "num_tokens": 4978901.0, + "step": 194 + }, + { + "epoch": 0.021414452009663958, + "grad_norm": 5.174006938934326, + "learning_rate": 1.0647639956092205e-07, + "loss": 1.2821, + "mean_token_accuracy": 0.6448178291320801, + "num_tokens": 5000632.0, + "step": 195 + }, + { + "epoch": 0.021524269712277617, + "grad_norm": 5.172635555267334, + "learning_rate": 1.0702524698133918e-07, + "loss": 1.3094, + "mean_token_accuracy": 0.6444629430770874, + "num_tokens": 5023601.0, + "step": 196 + }, + { + "epoch": 0.02163408741489128, + "grad_norm": 4.7726616859436035, + "learning_rate": 1.0757409440175631e-07, + "loss": 1.2865, + "mean_token_accuracy": 0.6437050700187683, + "num_tokens": 5050909.0, + "step": 197 + }, + { + "epoch": 0.02174390511750494, + "grad_norm": 5.9151530265808105, + "learning_rate": 1.0812294182217344e-07, + "loss": 1.2921, + "mean_token_accuracy": 0.6426801681518555, + "num_tokens": 5075265.0, + "step": 198 + }, + { + "epoch": 0.021853722820118604, + "grad_norm": 7.064569473266602, + "learning_rate": 1.0867178924259056e-07, + "loss": 1.2227, + "mean_token_accuracy": 0.6534507870674133, + "num_tokens": 5091827.0, + "step": 199 + }, + { + "epoch": 0.021963540522732264, + "grad_norm": 4.182703495025635, + "learning_rate": 1.0922063666300768e-07, + "loss": 1.339, + "mean_token_accuracy": 0.6269482970237732, + "num_tokens": 5119046.0, + "step": 200 + }, + { + "epoch": 0.022073358225345927, + "grad_norm": 4.053332328796387, + "learning_rate": 1.0976948408342481e-07, + "loss": 1.3291, + "mean_token_accuracy": 0.6271547079086304, + "num_tokens": 5145532.0, + "step": 201 + }, + { + "epoch": 0.022183175927959587, + "grad_norm": 4.51522159576416, + "learning_rate": 1.1031833150384192e-07, + "loss": 1.3732, + "mean_token_accuracy": 0.6176618337631226, + "num_tokens": 5174935.0, + "step": 202 + }, + { + "epoch": 0.022292993630573247, + "grad_norm": 4.410999298095703, + "learning_rate": 1.1086717892425905e-07, + "loss": 1.146, + "mean_token_accuracy": 0.6694151163101196, + "num_tokens": 5199211.0, + "step": 203 + }, + { + "epoch": 0.02240281133318691, + "grad_norm": 4.905502796173096, + "learning_rate": 1.1141602634467617e-07, + "loss": 1.2839, + "mean_token_accuracy": 0.641507089138031, + "num_tokens": 5219605.0, + "step": 204 + }, + { + "epoch": 0.02251262903580057, + "grad_norm": 4.5065083503723145, + "learning_rate": 1.119648737650933e-07, + "loss": 1.3309, + "mean_token_accuracy": 0.6371282339096069, + "num_tokens": 5247258.0, + "step": 205 + }, + { + "epoch": 0.022622446738414233, + "grad_norm": 5.392463684082031, + "learning_rate": 1.1251372118551042e-07, + "loss": 1.3102, + "mean_token_accuracy": 0.6342374086380005, + "num_tokens": 5274190.0, + "step": 206 + }, + { + "epoch": 0.022732264441027893, + "grad_norm": 4.395215034484863, + "learning_rate": 1.1306256860592754e-07, + "loss": 1.3259, + "mean_token_accuracy": 0.6361116170883179, + "num_tokens": 5300146.0, + "step": 207 + }, + { + "epoch": 0.022842082143641557, + "grad_norm": 5.067382335662842, + "learning_rate": 1.1361141602634467e-07, + "loss": 1.3212, + "mean_token_accuracy": 0.6367398500442505, + "num_tokens": 5327531.0, + "step": 208 + }, + { + "epoch": 0.022951899846255217, + "grad_norm": 4.517926216125488, + "learning_rate": 1.141602634467618e-07, + "loss": 1.3143, + "mean_token_accuracy": 0.6379848718643188, + "num_tokens": 5355779.0, + "step": 209 + }, + { + "epoch": 0.023061717548868876, + "grad_norm": 4.501303672790527, + "learning_rate": 1.1470911086717892e-07, + "loss": 1.374, + "mean_token_accuracy": 0.6261417865753174, + "num_tokens": 5386732.0, + "step": 210 + }, + { + "epoch": 0.02317153525148254, + "grad_norm": 5.366635322570801, + "learning_rate": 1.1525795828759604e-07, + "loss": 1.2377, + "mean_token_accuracy": 0.6544817090034485, + "num_tokens": 5412942.0, + "step": 211 + }, + { + "epoch": 0.0232813529540962, + "grad_norm": 4.997899055480957, + "learning_rate": 1.1580680570801317e-07, + "loss": 1.3553, + "mean_token_accuracy": 0.6231484413146973, + "num_tokens": 5443987.0, + "step": 212 + }, + { + "epoch": 0.023391170656709863, + "grad_norm": 4.606481552124023, + "learning_rate": 1.163556531284303e-07, + "loss": 1.3575, + "mean_token_accuracy": 0.6385343670845032, + "num_tokens": 5471974.0, + "step": 213 + }, + { + "epoch": 0.023500988359323523, + "grad_norm": 7.5633673667907715, + "learning_rate": 1.1690450054884741e-07, + "loss": 1.2444, + "mean_token_accuracy": 0.6549309492111206, + "num_tokens": 5494178.0, + "step": 214 + }, + { + "epoch": 0.023610806061937183, + "grad_norm": 4.344141006469727, + "learning_rate": 1.1745334796926454e-07, + "loss": 1.2842, + "mean_token_accuracy": 0.6449506282806396, + "num_tokens": 5524848.0, + "step": 215 + }, + { + "epoch": 0.023720623764550846, + "grad_norm": 6.455509185791016, + "learning_rate": 1.1800219538968166e-07, + "loss": 1.2323, + "mean_token_accuracy": 0.6537482738494873, + "num_tokens": 5544691.0, + "step": 216 + }, + { + "epoch": 0.023830441467164506, + "grad_norm": 4.403587818145752, + "learning_rate": 1.1855104281009878e-07, + "loss": 1.35, + "mean_token_accuracy": 0.6317843198776245, + "num_tokens": 5576544.0, + "step": 217 + }, + { + "epoch": 0.02394025916977817, + "grad_norm": 3.803525686264038, + "learning_rate": 1.1909989023051591e-07, + "loss": 1.3027, + "mean_token_accuracy": 0.6439926624298096, + "num_tokens": 5605617.0, + "step": 218 + }, + { + "epoch": 0.02405007687239183, + "grad_norm": 5.268388748168945, + "learning_rate": 1.1964873765093303e-07, + "loss": 1.2862, + "mean_token_accuracy": 0.6422793865203857, + "num_tokens": 5633593.0, + "step": 219 + }, + { + "epoch": 0.024159894575005492, + "grad_norm": 6.4832258224487305, + "learning_rate": 1.2019758507135017e-07, + "loss": 1.2971, + "mean_token_accuracy": 0.6367346048355103, + "num_tokens": 5655095.0, + "step": 220 + }, + { + "epoch": 0.024269712277619152, + "grad_norm": 4.717020034790039, + "learning_rate": 1.2074643249176729e-07, + "loss": 1.2478, + "mean_token_accuracy": 0.6529148817062378, + "num_tokens": 5686474.0, + "step": 221 + }, + { + "epoch": 0.024379529980232812, + "grad_norm": 4.873488903045654, + "learning_rate": 1.212952799121844e-07, + "loss": 1.2717, + "mean_token_accuracy": 0.6535371541976929, + "num_tokens": 5712087.0, + "step": 222 + }, + { + "epoch": 0.024489347682846475, + "grad_norm": 6.070002555847168, + "learning_rate": 1.2184412733260154e-07, + "loss": 1.2286, + "mean_token_accuracy": 0.6618334054946899, + "num_tokens": 5731576.0, + "step": 223 + }, + { + "epoch": 0.024599165385460135, + "grad_norm": 7.745193004608154, + "learning_rate": 1.2239297475301866e-07, + "loss": 1.16, + "mean_token_accuracy": 0.6649696826934814, + "num_tokens": 5753436.0, + "step": 224 + }, + { + "epoch": 0.0247089830880738, + "grad_norm": 6.015007972717285, + "learning_rate": 1.2294182217343577e-07, + "loss": 1.2425, + "mean_token_accuracy": 0.6438331007957458, + "num_tokens": 5772769.0, + "step": 225 + }, + { + "epoch": 0.02481880079068746, + "grad_norm": 6.39506721496582, + "learning_rate": 1.2349066959385291e-07, + "loss": 1.2336, + "mean_token_accuracy": 0.6521508693695068, + "num_tokens": 5800594.0, + "step": 226 + }, + { + "epoch": 0.024928618493301122, + "grad_norm": 6.148627281188965, + "learning_rate": 1.2403951701427003e-07, + "loss": 1.3162, + "mean_token_accuracy": 0.6203601956367493, + "num_tokens": 5829124.0, + "step": 227 + }, + { + "epoch": 0.02503843619591478, + "grad_norm": 6.565359115600586, + "learning_rate": 1.2458836443468714e-07, + "loss": 1.2169, + "mean_token_accuracy": 0.6563264727592468, + "num_tokens": 5852383.0, + "step": 228 + }, + { + "epoch": 0.02514825389852844, + "grad_norm": 5.77878475189209, + "learning_rate": 1.2513721185510429e-07, + "loss": 1.1745, + "mean_token_accuracy": 0.672430694103241, + "num_tokens": 5878884.0, + "step": 229 + }, + { + "epoch": 0.025258071601142105, + "grad_norm": 4.216459274291992, + "learning_rate": 1.256860592755214e-07, + "loss": 1.218, + "mean_token_accuracy": 0.6616355180740356, + "num_tokens": 5907538.0, + "step": 230 + }, + { + "epoch": 0.025367889303755765, + "grad_norm": 5.9494805335998535, + "learning_rate": 1.2623490669593854e-07, + "loss": 1.259, + "mean_token_accuracy": 0.6501343846321106, + "num_tokens": 5936460.0, + "step": 231 + }, + { + "epoch": 0.025477707006369428, + "grad_norm": 6.331883430480957, + "learning_rate": 1.2678375411635563e-07, + "loss": 1.3376, + "mean_token_accuracy": 0.6419795751571655, + "num_tokens": 5958287.0, + "step": 232 + }, + { + "epoch": 0.025587524708983088, + "grad_norm": 6.569954872131348, + "learning_rate": 1.2733260153677277e-07, + "loss": 1.2926, + "mean_token_accuracy": 0.6349687576293945, + "num_tokens": 5980655.0, + "step": 233 + }, + { + "epoch": 0.025697342411596748, + "grad_norm": 7.149877071380615, + "learning_rate": 1.278814489571899e-07, + "loss": 1.2959, + "mean_token_accuracy": 0.6353533864021301, + "num_tokens": 6003652.0, + "step": 234 + }, + { + "epoch": 0.02580716011421041, + "grad_norm": 5.130307674407959, + "learning_rate": 1.2843029637760703e-07, + "loss": 1.1869, + "mean_token_accuracy": 0.6757994890213013, + "num_tokens": 6033595.0, + "step": 235 + }, + { + "epoch": 0.02591697781682407, + "grad_norm": 6.259466648101807, + "learning_rate": 1.2897914379802412e-07, + "loss": 1.2316, + "mean_token_accuracy": 0.6529790759086609, + "num_tokens": 6059704.0, + "step": 236 + }, + { + "epoch": 0.026026795519437734, + "grad_norm": 5.514070987701416, + "learning_rate": 1.2952799121844126e-07, + "loss": 1.1488, + "mean_token_accuracy": 0.6762130856513977, + "num_tokens": 6085340.0, + "step": 237 + }, + { + "epoch": 0.026136613222051394, + "grad_norm": 6.137386322021484, + "learning_rate": 1.300768386388584e-07, + "loss": 1.1975, + "mean_token_accuracy": 0.6641693711280823, + "num_tokens": 6110940.0, + "step": 238 + }, + { + "epoch": 0.026246430924665057, + "grad_norm": 7.948629379272461, + "learning_rate": 1.3062568605927552e-07, + "loss": 1.2767, + "mean_token_accuracy": 0.6447526216506958, + "num_tokens": 6137687.0, + "step": 239 + }, + { + "epoch": 0.026356248627278717, + "grad_norm": 7.334742069244385, + "learning_rate": 1.3117453347969266e-07, + "loss": 1.2416, + "mean_token_accuracy": 0.648277223110199, + "num_tokens": 6160394.0, + "step": 240 + }, + { + "epoch": 0.026466066329892377, + "grad_norm": 5.88040828704834, + "learning_rate": 1.3172338090010975e-07, + "loss": 1.2235, + "mean_token_accuracy": 0.657954752445221, + "num_tokens": 6181500.0, + "step": 241 + }, + { + "epoch": 0.02657588403250604, + "grad_norm": 6.056039333343506, + "learning_rate": 1.322722283205269e-07, + "loss": 1.3241, + "mean_token_accuracy": 0.6281425952911377, + "num_tokens": 6207253.0, + "step": 242 + }, + { + "epoch": 0.0266857017351197, + "grad_norm": 6.597836971282959, + "learning_rate": 1.32821075740944e-07, + "loss": 1.2276, + "mean_token_accuracy": 0.6631567478179932, + "num_tokens": 6236986.0, + "step": 243 + }, + { + "epoch": 0.026795519437733364, + "grad_norm": 5.723628520965576, + "learning_rate": 1.3336992316136115e-07, + "loss": 1.3836, + "mean_token_accuracy": 0.6137393712997437, + "num_tokens": 6263320.0, + "step": 244 + }, + { + "epoch": 0.026905337140347024, + "grad_norm": 5.138948440551758, + "learning_rate": 1.3391877058177826e-07, + "loss": 1.2098, + "mean_token_accuracy": 0.6782844662666321, + "num_tokens": 6287798.0, + "step": 245 + }, + { + "epoch": 0.027015154842960687, + "grad_norm": 5.99278450012207, + "learning_rate": 1.3446761800219538e-07, + "loss": 1.3296, + "mean_token_accuracy": 0.6377227902412415, + "num_tokens": 6312762.0, + "step": 246 + }, + { + "epoch": 0.027124972545574347, + "grad_norm": 6.62746000289917, + "learning_rate": 1.350164654226125e-07, + "loss": 1.223, + "mean_token_accuracy": 0.652114748954773, + "num_tokens": 6337939.0, + "step": 247 + }, + { + "epoch": 0.027234790248188007, + "grad_norm": 7.105957508087158, + "learning_rate": 1.3556531284302963e-07, + "loss": 1.1968, + "mean_token_accuracy": 0.6599587202072144, + "num_tokens": 6358725.0, + "step": 248 + }, + { + "epoch": 0.02734460795080167, + "grad_norm": 5.689815044403076, + "learning_rate": 1.3611416026344675e-07, + "loss": 1.2067, + "mean_token_accuracy": 0.6565775871276855, + "num_tokens": 6381892.0, + "step": 249 + }, + { + "epoch": 0.02745442565341533, + "grad_norm": 6.045871734619141, + "learning_rate": 1.366630076838639e-07, + "loss": 1.302, + "mean_token_accuracy": 0.6408628225326538, + "num_tokens": 6410119.0, + "step": 250 + }, + { + "epoch": 0.027564243356028993, + "grad_norm": 5.67733097076416, + "learning_rate": 1.37211855104281e-07, + "loss": 1.2021, + "mean_token_accuracy": 0.6561005115509033, + "num_tokens": 6433824.0, + "step": 251 + }, + { + "epoch": 0.027674061058642653, + "grad_norm": 4.913093566894531, + "learning_rate": 1.3776070252469812e-07, + "loss": 1.2237, + "mean_token_accuracy": 0.6488772034645081, + "num_tokens": 6457044.0, + "step": 252 + }, + { + "epoch": 0.027783878761256313, + "grad_norm": 5.3536601066589355, + "learning_rate": 1.3830954994511526e-07, + "loss": 1.2738, + "mean_token_accuracy": 0.6392724514007568, + "num_tokens": 6478731.0, + "step": 253 + }, + { + "epoch": 0.027893696463869976, + "grad_norm": 5.897751808166504, + "learning_rate": 1.3885839736553238e-07, + "loss": 1.1202, + "mean_token_accuracy": 0.6790419816970825, + "num_tokens": 6499628.0, + "step": 254 + }, + { + "epoch": 0.028003514166483636, + "grad_norm": 4.741534233093262, + "learning_rate": 1.394072447859495e-07, + "loss": 1.2219, + "mean_token_accuracy": 0.6522753238677979, + "num_tokens": 6527028.0, + "step": 255 + }, + { + "epoch": 0.0281133318690973, + "grad_norm": 6.115477561950684, + "learning_rate": 1.399560922063666e-07, + "loss": 1.1948, + "mean_token_accuracy": 0.6639425754547119, + "num_tokens": 6549683.0, + "step": 256 + }, + { + "epoch": 0.02822314957171096, + "grad_norm": 7.621041297912598, + "learning_rate": 1.4050493962678375e-07, + "loss": 1.3007, + "mean_token_accuracy": 0.6364548802375793, + "num_tokens": 6575824.0, + "step": 257 + }, + { + "epoch": 0.028332967274324623, + "grad_norm": 4.963761806488037, + "learning_rate": 1.4105378704720087e-07, + "loss": 1.2073, + "mean_token_accuracy": 0.6678869724273682, + "num_tokens": 6601780.0, + "step": 258 + }, + { + "epoch": 0.028442784976938282, + "grad_norm": 6.196525573730469, + "learning_rate": 1.41602634467618e-07, + "loss": 1.2588, + "mean_token_accuracy": 0.663587749004364, + "num_tokens": 6624146.0, + "step": 259 + }, + { + "epoch": 0.028552602679551942, + "grad_norm": 8.272608757019043, + "learning_rate": 1.421514818880351e-07, + "loss": 1.269, + "mean_token_accuracy": 0.6455587148666382, + "num_tokens": 6645901.0, + "step": 260 + }, + { + "epoch": 0.028662420382165606, + "grad_norm": 5.2638139724731445, + "learning_rate": 1.4270032930845224e-07, + "loss": 1.2104, + "mean_token_accuracy": 0.6588470935821533, + "num_tokens": 6676249.0, + "step": 261 + }, + { + "epoch": 0.028772238084779266, + "grad_norm": 6.407500267028809, + "learning_rate": 1.4324917672886938e-07, + "loss": 1.2805, + "mean_token_accuracy": 0.6384953260421753, + "num_tokens": 6701377.0, + "step": 262 + }, + { + "epoch": 0.02888205578739293, + "grad_norm": 4.943818092346191, + "learning_rate": 1.437980241492865e-07, + "loss": 1.1894, + "mean_token_accuracy": 0.6640825867652893, + "num_tokens": 6726482.0, + "step": 263 + }, + { + "epoch": 0.02899187349000659, + "grad_norm": 5.275237083435059, + "learning_rate": 1.4434687156970364e-07, + "loss": 1.2711, + "mean_token_accuracy": 0.6425172686576843, + "num_tokens": 6751702.0, + "step": 264 + }, + { + "epoch": 0.029101691192620252, + "grad_norm": 6.066036701202393, + "learning_rate": 1.4489571899012073e-07, + "loss": 1.3125, + "mean_token_accuracy": 0.636096715927124, + "num_tokens": 6773615.0, + "step": 265 + }, + { + "epoch": 0.029211508895233912, + "grad_norm": 7.0488600730896, + "learning_rate": 1.4544456641053787e-07, + "loss": 1.1862, + "mean_token_accuracy": 0.6665740609169006, + "num_tokens": 6799907.0, + "step": 266 + }, + { + "epoch": 0.029321326597847572, + "grad_norm": 5.597392559051514, + "learning_rate": 1.4599341383095498e-07, + "loss": 1.3272, + "mean_token_accuracy": 0.6390376687049866, + "num_tokens": 6825317.0, + "step": 267 + }, + { + "epoch": 0.029431144300461235, + "grad_norm": 4.804421424865723, + "learning_rate": 1.4654226125137212e-07, + "loss": 1.2479, + "mean_token_accuracy": 0.6529965400695801, + "num_tokens": 6852437.0, + "step": 268 + }, + { + "epoch": 0.029540962003074895, + "grad_norm": 5.6291961669921875, + "learning_rate": 1.4709110867178924e-07, + "loss": 1.2302, + "mean_token_accuracy": 0.6538423895835876, + "num_tokens": 6879565.0, + "step": 269 + }, + { + "epoch": 0.02965077970568856, + "grad_norm": 5.672983646392822, + "learning_rate": 1.4763995609220636e-07, + "loss": 1.2679, + "mean_token_accuracy": 0.6472458839416504, + "num_tokens": 6909783.0, + "step": 270 + }, + { + "epoch": 0.029760597408302218, + "grad_norm": 4.852105140686035, + "learning_rate": 1.4818880351262347e-07, + "loss": 1.2912, + "mean_token_accuracy": 0.6415196657180786, + "num_tokens": 6936662.0, + "step": 271 + }, + { + "epoch": 0.029870415110915878, + "grad_norm": 5.15963077545166, + "learning_rate": 1.487376509330406e-07, + "loss": 1.1744, + "mean_token_accuracy": 0.6735693216323853, + "num_tokens": 6960884.0, + "step": 272 + }, + { + "epoch": 0.02998023281352954, + "grad_norm": 5.472066402435303, + "learning_rate": 1.4928649835345773e-07, + "loss": 1.3139, + "mean_token_accuracy": 0.6438592672348022, + "num_tokens": 6987241.0, + "step": 273 + }, + { + "epoch": 0.0300900505161432, + "grad_norm": 4.593018054962158, + "learning_rate": 1.4983534577387484e-07, + "loss": 1.1892, + "mean_token_accuracy": 0.6668792366981506, + "num_tokens": 7013978.0, + "step": 274 + }, + { + "epoch": 0.030199868218756865, + "grad_norm": 5.402882099151611, + "learning_rate": 1.5038419319429198e-07, + "loss": 1.2171, + "mean_token_accuracy": 0.6456851363182068, + "num_tokens": 7039689.0, + "step": 275 + }, + { + "epoch": 0.030309685921370524, + "grad_norm": 5.7603840827941895, + "learning_rate": 1.509330406147091e-07, + "loss": 1.11, + "mean_token_accuracy": 0.6784617900848389, + "num_tokens": 7063731.0, + "step": 276 + }, + { + "epoch": 0.030419503623984188, + "grad_norm": 5.451739311218262, + "learning_rate": 1.5148188803512624e-07, + "loss": 1.2663, + "mean_token_accuracy": 0.6449663639068604, + "num_tokens": 7088742.0, + "step": 277 + }, + { + "epoch": 0.030529321326597848, + "grad_norm": 5.38316535949707, + "learning_rate": 1.5203073545554336e-07, + "loss": 1.2515, + "mean_token_accuracy": 0.6505165100097656, + "num_tokens": 7114974.0, + "step": 278 + }, + { + "epoch": 0.030639139029211507, + "grad_norm": 6.21360445022583, + "learning_rate": 1.5257958287596047e-07, + "loss": 1.1974, + "mean_token_accuracy": 0.6630162000656128, + "num_tokens": 7138626.0, + "step": 279 + }, + { + "epoch": 0.03074895673182517, + "grad_norm": 5.983727931976318, + "learning_rate": 1.531284302963776e-07, + "loss": 1.1889, + "mean_token_accuracy": 0.6665114164352417, + "num_tokens": 7163830.0, + "step": 280 + }, + { + "epoch": 0.03085877443443883, + "grad_norm": 4.669240951538086, + "learning_rate": 1.5367727771679473e-07, + "loss": 1.3193, + "mean_token_accuracy": 0.6308317184448242, + "num_tokens": 7194745.0, + "step": 281 + }, + { + "epoch": 0.030968592137052494, + "grad_norm": 4.8747878074646, + "learning_rate": 1.5422612513721184e-07, + "loss": 1.2651, + "mean_token_accuracy": 0.6468274593353271, + "num_tokens": 7224378.0, + "step": 282 + }, + { + "epoch": 0.031078409839666154, + "grad_norm": 4.9163384437561035, + "learning_rate": 1.5477497255762899e-07, + "loss": 1.2239, + "mean_token_accuracy": 0.6548401713371277, + "num_tokens": 7253002.0, + "step": 283 + }, + { + "epoch": 0.031188227542279817, + "grad_norm": 4.7044830322265625, + "learning_rate": 1.5532381997804607e-07, + "loss": 1.2443, + "mean_token_accuracy": 0.6527295112609863, + "num_tokens": 7279824.0, + "step": 284 + }, + { + "epoch": 0.03129804524489348, + "grad_norm": 5.839854717254639, + "learning_rate": 1.5587266739846322e-07, + "loss": 1.31, + "mean_token_accuracy": 0.645553469657898, + "num_tokens": 7303306.0, + "step": 285 + }, + { + "epoch": 0.03140786294750714, + "grad_norm": 5.962801933288574, + "learning_rate": 1.5642151481888036e-07, + "loss": 1.1361, + "mean_token_accuracy": 0.6725028157234192, + "num_tokens": 7323139.0, + "step": 286 + }, + { + "epoch": 0.0315176806501208, + "grad_norm": 5.518036842346191, + "learning_rate": 1.5697036223929747e-07, + "loss": 1.1898, + "mean_token_accuracy": 0.663238525390625, + "num_tokens": 7343597.0, + "step": 287 + }, + { + "epoch": 0.031627498352734464, + "grad_norm": 4.337423324584961, + "learning_rate": 1.5751920965971461e-07, + "loss": 1.2463, + "mean_token_accuracy": 0.6524689197540283, + "num_tokens": 7376652.0, + "step": 288 + }, + { + "epoch": 0.03173731605534812, + "grad_norm": 4.732696056365967, + "learning_rate": 1.580680570801317e-07, + "loss": 1.192, + "mean_token_accuracy": 0.648725152015686, + "num_tokens": 7400067.0, + "step": 289 + }, + { + "epoch": 0.03184713375796178, + "grad_norm": 4.7172746658325195, + "learning_rate": 1.5861690450054885e-07, + "loss": 1.1863, + "mean_token_accuracy": 0.6631999015808105, + "num_tokens": 7429305.0, + "step": 290 + }, + { + "epoch": 0.03195695146057544, + "grad_norm": 5.357011318206787, + "learning_rate": 1.5916575192096596e-07, + "loss": 1.2611, + "mean_token_accuracy": 0.6508187651634216, + "num_tokens": 7453865.0, + "step": 291 + }, + { + "epoch": 0.0320667691631891, + "grad_norm": 4.473448753356934, + "learning_rate": 1.597145993413831e-07, + "loss": 1.3346, + "mean_token_accuracy": 0.6263618469238281, + "num_tokens": 7486367.0, + "step": 292 + }, + { + "epoch": 0.03217658686580277, + "grad_norm": 4.72819709777832, + "learning_rate": 1.602634467618002e-07, + "loss": 1.2732, + "mean_token_accuracy": 0.6446936130523682, + "num_tokens": 7515889.0, + "step": 293 + }, + { + "epoch": 0.03228640456841643, + "grad_norm": 4.4133381843566895, + "learning_rate": 1.6081229418221733e-07, + "loss": 1.2706, + "mean_token_accuracy": 0.6385302543640137, + "num_tokens": 7544467.0, + "step": 294 + }, + { + "epoch": 0.03239622227103009, + "grad_norm": 4.30421781539917, + "learning_rate": 1.6136114160263445e-07, + "loss": 1.1691, + "mean_token_accuracy": 0.6643193960189819, + "num_tokens": 7571748.0, + "step": 295 + }, + { + "epoch": 0.03250603997364375, + "grad_norm": 4.938171863555908, + "learning_rate": 1.619099890230516e-07, + "loss": 1.1658, + "mean_token_accuracy": 0.6612244248390198, + "num_tokens": 7600291.0, + "step": 296 + }, + { + "epoch": 0.03261585767625741, + "grad_norm": 4.805749893188477, + "learning_rate": 1.6245883644346873e-07, + "loss": 1.251, + "mean_token_accuracy": 0.6417202949523926, + "num_tokens": 7626028.0, + "step": 297 + }, + { + "epoch": 0.032725675378871076, + "grad_norm": 4.984499931335449, + "learning_rate": 1.6300768386388582e-07, + "loss": 1.1087, + "mean_token_accuracy": 0.6830609440803528, + "num_tokens": 7649670.0, + "step": 298 + }, + { + "epoch": 0.032835493081484736, + "grad_norm": 4.923973083496094, + "learning_rate": 1.6355653128430296e-07, + "loss": 1.2854, + "mean_token_accuracy": 0.6456958055496216, + "num_tokens": 7671983.0, + "step": 299 + }, + { + "epoch": 0.032945310784098396, + "grad_norm": 4.845790863037109, + "learning_rate": 1.6410537870472008e-07, + "loss": 1.1814, + "mean_token_accuracy": 0.6628363728523254, + "num_tokens": 7702151.0, + "step": 300 + }, + { + "epoch": 0.033055128486712056, + "grad_norm": 5.005503177642822, + "learning_rate": 1.6465422612513722e-07, + "loss": 1.1541, + "mean_token_accuracy": 0.6669400930404663, + "num_tokens": 7724230.0, + "step": 301 + }, + { + "epoch": 0.03316494618932572, + "grad_norm": 6.291779518127441, + "learning_rate": 1.6520307354555433e-07, + "loss": 1.1796, + "mean_token_accuracy": 0.6580267548561096, + "num_tokens": 7744146.0, + "step": 302 + }, + { + "epoch": 0.03327476389193938, + "grad_norm": 4.594903945922852, + "learning_rate": 1.6575192096597145e-07, + "loss": 1.2302, + "mean_token_accuracy": 0.6598083972930908, + "num_tokens": 7767892.0, + "step": 303 + }, + { + "epoch": 0.03338458159455304, + "grad_norm": 5.029612064361572, + "learning_rate": 1.6630076838638856e-07, + "loss": 1.2319, + "mean_token_accuracy": 0.6519712209701538, + "num_tokens": 7800894.0, + "step": 304 + }, + { + "epoch": 0.0334943992971667, + "grad_norm": 5.085905075073242, + "learning_rate": 1.668496158068057e-07, + "loss": 1.2131, + "mean_token_accuracy": 0.6528096795082092, + "num_tokens": 7827149.0, + "step": 305 + }, + { + "epoch": 0.03360421699978036, + "grad_norm": 4.438549995422363, + "learning_rate": 1.6739846322722282e-07, + "loss": 1.1531, + "mean_token_accuracy": 0.6716302633285522, + "num_tokens": 7856886.0, + "step": 306 + }, + { + "epoch": 0.03371403470239403, + "grad_norm": 4.344805717468262, + "learning_rate": 1.6794731064763996e-07, + "loss": 1.2557, + "mean_token_accuracy": 0.6498540639877319, + "num_tokens": 7883960.0, + "step": 307 + }, + { + "epoch": 0.03382385240500769, + "grad_norm": 5.8958539962768555, + "learning_rate": 1.6849615806805705e-07, + "loss": 1.2422, + "mean_token_accuracy": 0.6465470790863037, + "num_tokens": 7907601.0, + "step": 308 + }, + { + "epoch": 0.03393367010762135, + "grad_norm": 5.223424434661865, + "learning_rate": 1.690450054884742e-07, + "loss": 1.1924, + "mean_token_accuracy": 0.6635427474975586, + "num_tokens": 7931322.0, + "step": 309 + }, + { + "epoch": 0.03404348781023501, + "grad_norm": 4.792941093444824, + "learning_rate": 1.6959385290889134e-07, + "loss": 1.2411, + "mean_token_accuracy": 0.6530506610870361, + "num_tokens": 7962336.0, + "step": 310 + }, + { + "epoch": 0.03415330551284867, + "grad_norm": 5.27213716506958, + "learning_rate": 1.7014270032930845e-07, + "loss": 1.2397, + "mean_token_accuracy": 0.6516285538673401, + "num_tokens": 7995196.0, + "step": 311 + }, + { + "epoch": 0.034263123215462335, + "grad_norm": 4.582456111907959, + "learning_rate": 1.7069154774972557e-07, + "loss": 1.2539, + "mean_token_accuracy": 0.6435145139694214, + "num_tokens": 8023004.0, + "step": 312 + }, + { + "epoch": 0.034372940918075995, + "grad_norm": 5.090002059936523, + "learning_rate": 1.7124039517014268e-07, + "loss": 1.1474, + "mean_token_accuracy": 0.6787323951721191, + "num_tokens": 8047112.0, + "step": 313 + }, + { + "epoch": 0.034482758620689655, + "grad_norm": 5.707190036773682, + "learning_rate": 1.7178924259055982e-07, + "loss": 1.2445, + "mean_token_accuracy": 0.6491540670394897, + "num_tokens": 8075601.0, + "step": 314 + }, + { + "epoch": 0.034592576323303315, + "grad_norm": 4.948573589324951, + "learning_rate": 1.7233809001097694e-07, + "loss": 1.1394, + "mean_token_accuracy": 0.6772005558013916, + "num_tokens": 8099696.0, + "step": 315 + }, + { + "epoch": 0.034702394025916974, + "grad_norm": 6.033024311065674, + "learning_rate": 1.7288693743139408e-07, + "loss": 1.1211, + "mean_token_accuracy": 0.6746683120727539, + "num_tokens": 8125145.0, + "step": 316 + }, + { + "epoch": 0.03481221172853064, + "grad_norm": 4.485821723937988, + "learning_rate": 1.7343578485181117e-07, + "loss": 1.1586, + "mean_token_accuracy": 0.6655963063240051, + "num_tokens": 8147462.0, + "step": 317 + }, + { + "epoch": 0.0349220294311443, + "grad_norm": 4.463705539703369, + "learning_rate": 1.739846322722283e-07, + "loss": 1.2009, + "mean_token_accuracy": 0.6625060439109802, + "num_tokens": 8177087.0, + "step": 318 + }, + { + "epoch": 0.03503184713375796, + "grad_norm": 4.856771945953369, + "learning_rate": 1.7453347969264543e-07, + "loss": 1.1137, + "mean_token_accuracy": 0.6748943328857422, + "num_tokens": 8202579.0, + "step": 319 + }, + { + "epoch": 0.03514166483637162, + "grad_norm": 4.413636684417725, + "learning_rate": 1.7508232711306257e-07, + "loss": 1.1744, + "mean_token_accuracy": 0.6678944230079651, + "num_tokens": 8223518.0, + "step": 320 + }, + { + "epoch": 0.03525148253898529, + "grad_norm": 6.218851089477539, + "learning_rate": 1.756311745334797e-07, + "loss": 1.1942, + "mean_token_accuracy": 0.6684645414352417, + "num_tokens": 8242582.0, + "step": 321 + }, + { + "epoch": 0.03536130024159895, + "grad_norm": 3.812760353088379, + "learning_rate": 1.761800219538968e-07, + "loss": 1.2052, + "mean_token_accuracy": 0.6649820804595947, + "num_tokens": 8270475.0, + "step": 322 + }, + { + "epoch": 0.03547111794421261, + "grad_norm": 4.867440223693848, + "learning_rate": 1.7672886937431394e-07, + "loss": 1.1304, + "mean_token_accuracy": 0.6665424108505249, + "num_tokens": 8288061.0, + "step": 323 + }, + { + "epoch": 0.03558093564682627, + "grad_norm": 4.8383259773254395, + "learning_rate": 1.7727771679473105e-07, + "loss": 1.2788, + "mean_token_accuracy": 0.6339805126190186, + "num_tokens": 8315024.0, + "step": 324 + }, + { + "epoch": 0.03569075334943993, + "grad_norm": 4.784774303436279, + "learning_rate": 1.778265642151482e-07, + "loss": 1.2894, + "mean_token_accuracy": 0.6424678564071655, + "num_tokens": 8339637.0, + "step": 325 + }, + { + "epoch": 0.035800571052053594, + "grad_norm": 4.742081165313721, + "learning_rate": 1.7837541163556529e-07, + "loss": 1.118, + "mean_token_accuracy": 0.6756885051727295, + "num_tokens": 8361581.0, + "step": 326 + }, + { + "epoch": 0.035910388754667254, + "grad_norm": 3.813265800476074, + "learning_rate": 1.7892425905598243e-07, + "loss": 1.2661, + "mean_token_accuracy": 0.6443532705307007, + "num_tokens": 8388977.0, + "step": 327 + }, + { + "epoch": 0.036020206457280914, + "grad_norm": 4.5399489402771, + "learning_rate": 1.7947310647639954e-07, + "loss": 1.2115, + "mean_token_accuracy": 0.6679291129112244, + "num_tokens": 8418340.0, + "step": 328 + }, + { + "epoch": 0.03613002415989457, + "grad_norm": 4.1351823806762695, + "learning_rate": 1.8002195389681668e-07, + "loss": 1.2668, + "mean_token_accuracy": 0.6400627493858337, + "num_tokens": 8444765.0, + "step": 329 + }, + { + "epoch": 0.03623984186250823, + "grad_norm": 5.012885093688965, + "learning_rate": 1.805708013172338e-07, + "loss": 1.3143, + "mean_token_accuracy": 0.6406410336494446, + "num_tokens": 8471904.0, + "step": 330 + }, + { + "epoch": 0.0363496595651219, + "grad_norm": 5.158116817474365, + "learning_rate": 1.8111964873765091e-07, + "loss": 1.3073, + "mean_token_accuracy": 0.632429838180542, + "num_tokens": 8496713.0, + "step": 331 + }, + { + "epoch": 0.03645947726773556, + "grad_norm": 5.722342491149902, + "learning_rate": 1.8166849615806803e-07, + "loss": 1.2276, + "mean_token_accuracy": 0.6465221643447876, + "num_tokens": 8521959.0, + "step": 332 + }, + { + "epoch": 0.03656929497034922, + "grad_norm": 4.688882350921631, + "learning_rate": 1.8221734357848517e-07, + "loss": 1.1458, + "mean_token_accuracy": 0.674392819404602, + "num_tokens": 8545395.0, + "step": 333 + }, + { + "epoch": 0.03667911267296288, + "grad_norm": 3.7839643955230713, + "learning_rate": 1.827661909989023e-07, + "loss": 1.214, + "mean_token_accuracy": 0.6586793661117554, + "num_tokens": 8576886.0, + "step": 334 + }, + { + "epoch": 0.03678893037557654, + "grad_norm": 5.205403804779053, + "learning_rate": 1.8331503841931943e-07, + "loss": 1.1287, + "mean_token_accuracy": 0.6786304712295532, + "num_tokens": 8597484.0, + "step": 335 + }, + { + "epoch": 0.036898748078190206, + "grad_norm": 4.3159260749816895, + "learning_rate": 1.8386388583973654e-07, + "loss": 1.1614, + "mean_token_accuracy": 0.6638450622558594, + "num_tokens": 8617614.0, + "step": 336 + }, + { + "epoch": 0.037008565780803866, + "grad_norm": 4.173069000244141, + "learning_rate": 1.8441273326015366e-07, + "loss": 1.158, + "mean_token_accuracy": 0.6714608073234558, + "num_tokens": 8641142.0, + "step": 337 + }, + { + "epoch": 0.037118383483417526, + "grad_norm": 4.422873020172119, + "learning_rate": 1.849615806805708e-07, + "loss": 1.1683, + "mean_token_accuracy": 0.6608454585075378, + "num_tokens": 8666279.0, + "step": 338 + }, + { + "epoch": 0.037228201186031186, + "grad_norm": 4.58045768737793, + "learning_rate": 1.8551042810098792e-07, + "loss": 1.149, + "mean_token_accuracy": 0.6704791784286499, + "num_tokens": 8690795.0, + "step": 339 + }, + { + "epoch": 0.03733801888864485, + "grad_norm": 5.128715515136719, + "learning_rate": 1.8605927552140506e-07, + "loss": 1.1261, + "mean_token_accuracy": 0.6797670125961304, + "num_tokens": 8710978.0, + "step": 340 + }, + { + "epoch": 0.03744783659125851, + "grad_norm": 4.900345802307129, + "learning_rate": 1.8660812294182215e-07, + "loss": 1.241, + "mean_token_accuracy": 0.6545315384864807, + "num_tokens": 8736185.0, + "step": 341 + }, + { + "epoch": 0.03755765429387217, + "grad_norm": 3.9048378467559814, + "learning_rate": 1.871569703622393e-07, + "loss": 1.2564, + "mean_token_accuracy": 0.6413177847862244, + "num_tokens": 8766469.0, + "step": 342 + }, + { + "epoch": 0.03766747199648583, + "grad_norm": 5.212855339050293, + "learning_rate": 1.877058177826564e-07, + "loss": 1.1793, + "mean_token_accuracy": 0.6645893454551697, + "num_tokens": 8790117.0, + "step": 343 + }, + { + "epoch": 0.03777728969909949, + "grad_norm": 5.721282005310059, + "learning_rate": 1.8825466520307354e-07, + "loss": 1.2448, + "mean_token_accuracy": 0.6466518640518188, + "num_tokens": 8814427.0, + "step": 344 + }, + { + "epoch": 0.03788710740171316, + "grad_norm": 5.715538501739502, + "learning_rate": 1.8880351262349069e-07, + "loss": 1.1155, + "mean_token_accuracy": 0.6762257218360901, + "num_tokens": 8833353.0, + "step": 345 + }, + { + "epoch": 0.03799692510432682, + "grad_norm": 4.4080681800842285, + "learning_rate": 1.8935236004390778e-07, + "loss": 1.2648, + "mean_token_accuracy": 0.6424624919891357, + "num_tokens": 8862485.0, + "step": 346 + }, + { + "epoch": 0.03810674280694048, + "grad_norm": 4.225715637207031, + "learning_rate": 1.8990120746432492e-07, + "loss": 1.2631, + "mean_token_accuracy": 0.6386491060256958, + "num_tokens": 8892074.0, + "step": 347 + }, + { + "epoch": 0.03821656050955414, + "grad_norm": 4.2640767097473145, + "learning_rate": 1.9045005488474203e-07, + "loss": 1.2321, + "mean_token_accuracy": 0.6544698476791382, + "num_tokens": 8924125.0, + "step": 348 + }, + { + "epoch": 0.0383263782121678, + "grad_norm": 4.540669918060303, + "learning_rate": 1.9099890230515917e-07, + "loss": 1.1719, + "mean_token_accuracy": 0.660098671913147, + "num_tokens": 8946744.0, + "step": 349 + }, + { + "epoch": 0.038436195914781465, + "grad_norm": 5.106960296630859, + "learning_rate": 1.9154774972557626e-07, + "loss": 1.0916, + "mean_token_accuracy": 0.6771001815795898, + "num_tokens": 8965834.0, + "step": 350 + }, + { + "epoch": 0.038546013617395125, + "grad_norm": 4.817321300506592, + "learning_rate": 1.920965971459934e-07, + "loss": 1.2145, + "mean_token_accuracy": 0.6686131954193115, + "num_tokens": 8990522.0, + "step": 351 + }, + { + "epoch": 0.038655831320008785, + "grad_norm": 4.790185451507568, + "learning_rate": 1.9264544456641052e-07, + "loss": 1.1162, + "mean_token_accuracy": 0.6758798360824585, + "num_tokens": 9013768.0, + "step": 352 + }, + { + "epoch": 0.038765649022622445, + "grad_norm": 4.167245388031006, + "learning_rate": 1.9319429198682766e-07, + "loss": 1.2977, + "mean_token_accuracy": 0.6381700038909912, + "num_tokens": 9043160.0, + "step": 353 + }, + { + "epoch": 0.038875466725236105, + "grad_norm": 5.048829555511475, + "learning_rate": 1.9374313940724478e-07, + "loss": 1.1738, + "mean_token_accuracy": 0.6642544269561768, + "num_tokens": 9065524.0, + "step": 354 + }, + { + "epoch": 0.03898528442784977, + "grad_norm": 4.458340644836426, + "learning_rate": 1.942919868276619e-07, + "loss": 1.2325, + "mean_token_accuracy": 0.6509512662887573, + "num_tokens": 9087831.0, + "step": 355 + }, + { + "epoch": 0.03909510213046343, + "grad_norm": 4.748081207275391, + "learning_rate": 1.9484083424807903e-07, + "loss": 1.1656, + "mean_token_accuracy": 0.6694294214248657, + "num_tokens": 9115033.0, + "step": 356 + }, + { + "epoch": 0.03920491983307709, + "grad_norm": 4.702285289764404, + "learning_rate": 1.9538968166849615e-07, + "loss": 1.2637, + "mean_token_accuracy": 0.6513940691947937, + "num_tokens": 9141063.0, + "step": 357 + }, + { + "epoch": 0.03931473753569075, + "grad_norm": 4.442079067230225, + "learning_rate": 1.959385290889133e-07, + "loss": 1.1655, + "mean_token_accuracy": 0.668415904045105, + "num_tokens": 9165211.0, + "step": 358 + }, + { + "epoch": 0.03942455523830442, + "grad_norm": 4.431364059448242, + "learning_rate": 1.964873765093304e-07, + "loss": 1.1954, + "mean_token_accuracy": 0.6540940403938293, + "num_tokens": 9192273.0, + "step": 359 + }, + { + "epoch": 0.03953437294091808, + "grad_norm": 4.209264278411865, + "learning_rate": 1.9703622392974752e-07, + "loss": 1.1797, + "mean_token_accuracy": 0.6714420318603516, + "num_tokens": 9214775.0, + "step": 360 + }, + { + "epoch": 0.03964419064353174, + "grad_norm": 4.075558662414551, + "learning_rate": 1.9758507135016464e-07, + "loss": 1.1893, + "mean_token_accuracy": 0.6551254391670227, + "num_tokens": 9239602.0, + "step": 361 + }, + { + "epoch": 0.0397540083461454, + "grad_norm": 3.9245712757110596, + "learning_rate": 1.9813391877058178e-07, + "loss": 1.2024, + "mean_token_accuracy": 0.6534669995307922, + "num_tokens": 9267845.0, + "step": 362 + }, + { + "epoch": 0.03986382604875906, + "grad_norm": 3.7516539096832275, + "learning_rate": 1.986827661909989e-07, + "loss": 1.2108, + "mean_token_accuracy": 0.6546672582626343, + "num_tokens": 9297930.0, + "step": 363 + }, + { + "epoch": 0.039973643751372724, + "grad_norm": 4.294806003570557, + "learning_rate": 1.9923161361141603e-07, + "loss": 1.2215, + "mean_token_accuracy": 0.6655240058898926, + "num_tokens": 9323092.0, + "step": 364 + }, + { + "epoch": 0.040083461453986384, + "grad_norm": 3.342374086380005, + "learning_rate": 1.9978046103183312e-07, + "loss": 1.2616, + "mean_token_accuracy": 0.6423945426940918, + "num_tokens": 9352852.0, + "step": 365 + }, + { + "epoch": 0.040193279156600044, + "grad_norm": 4.526165008544922, + "learning_rate": 2.0032930845225027e-07, + "loss": 1.2259, + "mean_token_accuracy": 0.6445549726486206, + "num_tokens": 9373759.0, + "step": 366 + }, + { + "epoch": 0.040303096859213704, + "grad_norm": 3.6048614978790283, + "learning_rate": 2.0087815587266738e-07, + "loss": 1.1765, + "mean_token_accuracy": 0.6739569902420044, + "num_tokens": 9403051.0, + "step": 367 + }, + { + "epoch": 0.040412914561827364, + "grad_norm": 3.767897605895996, + "learning_rate": 2.0142700329308452e-07, + "loss": 1.1731, + "mean_token_accuracy": 0.6602878570556641, + "num_tokens": 9428895.0, + "step": 368 + }, + { + "epoch": 0.04052273226444103, + "grad_norm": 3.9848031997680664, + "learning_rate": 2.0197585071350164e-07, + "loss": 1.1933, + "mean_token_accuracy": 0.6823248267173767, + "num_tokens": 9453309.0, + "step": 369 + }, + { + "epoch": 0.04063254996705469, + "grad_norm": 3.6782124042510986, + "learning_rate": 2.0252469813391875e-07, + "loss": 1.0533, + "mean_token_accuracy": 0.6982205510139465, + "num_tokens": 9477981.0, + "step": 370 + }, + { + "epoch": 0.04074236766966835, + "grad_norm": 4.573125839233398, + "learning_rate": 2.030735455543359e-07, + "loss": 1.1238, + "mean_token_accuracy": 0.692509114742279, + "num_tokens": 9501771.0, + "step": 371 + }, + { + "epoch": 0.04085218537228201, + "grad_norm": 3.338397264480591, + "learning_rate": 2.03622392974753e-07, + "loss": 1.1522, + "mean_token_accuracy": 0.6661827564239502, + "num_tokens": 9533625.0, + "step": 372 + }, + { + "epoch": 0.04096200307489567, + "grad_norm": 4.401088237762451, + "learning_rate": 2.0417124039517015e-07, + "loss": 1.1866, + "mean_token_accuracy": 0.6687107086181641, + "num_tokens": 9557069.0, + "step": 373 + }, + { + "epoch": 0.04107182077750934, + "grad_norm": 3.0849111080169678, + "learning_rate": 2.0472008781558724e-07, + "loss": 1.1749, + "mean_token_accuracy": 0.6652166247367859, + "num_tokens": 9593371.0, + "step": 374 + }, + { + "epoch": 0.041181638480122996, + "grad_norm": 3.62843918800354, + "learning_rate": 2.0526893523600438e-07, + "loss": 1.253, + "mean_token_accuracy": 0.645747184753418, + "num_tokens": 9623207.0, + "step": 375 + }, + { + "epoch": 0.041291456182736656, + "grad_norm": 3.633140802383423, + "learning_rate": 2.058177826564215e-07, + "loss": 1.2107, + "mean_token_accuracy": 0.6589434146881104, + "num_tokens": 9652939.0, + "step": 376 + }, + { + "epoch": 0.041401273885350316, + "grad_norm": 5.123833179473877, + "learning_rate": 2.0636663007683864e-07, + "loss": 1.2039, + "mean_token_accuracy": 0.6542313694953918, + "num_tokens": 9673970.0, + "step": 377 + }, + { + "epoch": 0.04151109158796398, + "grad_norm": 4.49476432800293, + "learning_rate": 2.0691547749725575e-07, + "loss": 1.2318, + "mean_token_accuracy": 0.6471540331840515, + "num_tokens": 9699461.0, + "step": 378 + }, + { + "epoch": 0.04162090929057764, + "grad_norm": 3.147996664047241, + "learning_rate": 2.0746432491767287e-07, + "loss": 1.2464, + "mean_token_accuracy": 0.6395561099052429, + "num_tokens": 9729151.0, + "step": 379 + }, + { + "epoch": 0.0417307269931913, + "grad_norm": 4.064396381378174, + "learning_rate": 2.0801317233809e-07, + "loss": 1.2294, + "mean_token_accuracy": 0.6532146334648132, + "num_tokens": 9756438.0, + "step": 380 + }, + { + "epoch": 0.04184054469580496, + "grad_norm": 3.679680585861206, + "learning_rate": 2.0856201975850713e-07, + "loss": 1.1942, + "mean_token_accuracy": 0.6640514135360718, + "num_tokens": 9788075.0, + "step": 381 + }, + { + "epoch": 0.04195036239841862, + "grad_norm": 3.4784293174743652, + "learning_rate": 2.0911086717892427e-07, + "loss": 1.1491, + "mean_token_accuracy": 0.665923535823822, + "num_tokens": 9813396.0, + "step": 382 + }, + { + "epoch": 0.04206018010103229, + "grad_norm": 3.8020691871643066, + "learning_rate": 2.0965971459934136e-07, + "loss": 1.2436, + "mean_token_accuracy": 0.6515878438949585, + "num_tokens": 9841369.0, + "step": 383 + }, + { + "epoch": 0.04216999780364595, + "grad_norm": 4.042391300201416, + "learning_rate": 2.102085620197585e-07, + "loss": 1.1566, + "mean_token_accuracy": 0.6677910685539246, + "num_tokens": 9867620.0, + "step": 384 + }, + { + "epoch": 0.04227981550625961, + "grad_norm": 4.059485912322998, + "learning_rate": 2.1075740944017561e-07, + "loss": 1.1834, + "mean_token_accuracy": 0.660812497138977, + "num_tokens": 9893936.0, + "step": 385 + }, + { + "epoch": 0.04238963320887327, + "grad_norm": 3.774405002593994, + "learning_rate": 2.1130625686059276e-07, + "loss": 1.1929, + "mean_token_accuracy": 0.6566463112831116, + "num_tokens": 9918796.0, + "step": 386 + }, + { + "epoch": 0.04249945091148693, + "grad_norm": 3.138674020767212, + "learning_rate": 2.1185510428100987e-07, + "loss": 1.1983, + "mean_token_accuracy": 0.658737301826477, + "num_tokens": 9946529.0, + "step": 387 + }, + { + "epoch": 0.042609268614100596, + "grad_norm": 3.8448877334594727, + "learning_rate": 2.1240395170142699e-07, + "loss": 1.1586, + "mean_token_accuracy": 0.6672120094299316, + "num_tokens": 9974023.0, + "step": 388 + }, + { + "epoch": 0.042719086316714255, + "grad_norm": 4.3187665939331055, + "learning_rate": 2.129527991218441e-07, + "loss": 1.1253, + "mean_token_accuracy": 0.6776610612869263, + "num_tokens": 9996684.0, + "step": 389 + }, + { + "epoch": 0.042828904019327915, + "grad_norm": 4.696625709533691, + "learning_rate": 2.1350164654226124e-07, + "loss": 1.1502, + "mean_token_accuracy": 0.6642169952392578, + "num_tokens": 10015749.0, + "step": 390 + }, + { + "epoch": 0.042938721721941575, + "grad_norm": 4.0754547119140625, + "learning_rate": 2.1405049396267836e-07, + "loss": 1.1487, + "mean_token_accuracy": 0.6687294244766235, + "num_tokens": 10043080.0, + "step": 391 + }, + { + "epoch": 0.043048539424555235, + "grad_norm": 3.9738986492156982, + "learning_rate": 2.145993413830955e-07, + "loss": 1.1126, + "mean_token_accuracy": 0.6771374940872192, + "num_tokens": 10068914.0, + "step": 392 + }, + { + "epoch": 0.0431583571271689, + "grad_norm": 4.123415946960449, + "learning_rate": 2.1514818880351262e-07, + "loss": 1.1193, + "mean_token_accuracy": 0.6786293983459473, + "num_tokens": 10091189.0, + "step": 393 + }, + { + "epoch": 0.04326817482978256, + "grad_norm": 3.4091949462890625, + "learning_rate": 2.1569703622392973e-07, + "loss": 1.235, + "mean_token_accuracy": 0.6451846957206726, + "num_tokens": 10116202.0, + "step": 394 + }, + { + "epoch": 0.04337799253239622, + "grad_norm": 3.8823699951171875, + "learning_rate": 2.1624588364434687e-07, + "loss": 1.2121, + "mean_token_accuracy": 0.6685320734977722, + "num_tokens": 10144107.0, + "step": 395 + }, + { + "epoch": 0.04348781023500988, + "grad_norm": 3.482656240463257, + "learning_rate": 2.16794731064764e-07, + "loss": 1.2797, + "mean_token_accuracy": 0.6488708853721619, + "num_tokens": 10170246.0, + "step": 396 + }, + { + "epoch": 0.04359762793762355, + "grad_norm": 4.895567893981934, + "learning_rate": 2.1734357848518113e-07, + "loss": 1.1329, + "mean_token_accuracy": 0.6739936470985413, + "num_tokens": 10187807.0, + "step": 397 + }, + { + "epoch": 0.04370744564023721, + "grad_norm": 3.4219818115234375, + "learning_rate": 2.1789242590559822e-07, + "loss": 1.252, + "mean_token_accuracy": 0.6420021057128906, + "num_tokens": 10215789.0, + "step": 398 + }, + { + "epoch": 0.04381726334285087, + "grad_norm": 3.0420260429382324, + "learning_rate": 2.1844127332601536e-07, + "loss": 1.2186, + "mean_token_accuracy": 0.6500681042671204, + "num_tokens": 10248606.0, + "step": 399 + }, + { + "epoch": 0.04392708104546453, + "grad_norm": 3.2574219703674316, + "learning_rate": 2.1899012074643247e-07, + "loss": 1.1312, + "mean_token_accuracy": 0.6725400686264038, + "num_tokens": 10277992.0, + "step": 400 + }, + { + "epoch": 0.04403689874807819, + "grad_norm": 4.031012058258057, + "learning_rate": 2.1953896816684962e-07, + "loss": 1.0762, + "mean_token_accuracy": 0.6878728866577148, + "num_tokens": 10300979.0, + "step": 401 + }, + { + "epoch": 0.044146716450691854, + "grad_norm": 3.2007157802581787, + "learning_rate": 2.200878155872667e-07, + "loss": 1.1977, + "mean_token_accuracy": 0.6607226133346558, + "num_tokens": 10332379.0, + "step": 402 + }, + { + "epoch": 0.044256534153305514, + "grad_norm": 3.456029176712036, + "learning_rate": 2.2063666300768385e-07, + "loss": 1.2341, + "mean_token_accuracy": 0.6439775228500366, + "num_tokens": 10358190.0, + "step": 403 + }, + { + "epoch": 0.044366351855919174, + "grad_norm": 3.3312692642211914, + "learning_rate": 2.21185510428101e-07, + "loss": 1.3193, + "mean_token_accuracy": 0.6382639408111572, + "num_tokens": 10387246.0, + "step": 404 + }, + { + "epoch": 0.044476169558532834, + "grad_norm": 2.9523539543151855, + "learning_rate": 2.217343578485181e-07, + "loss": 1.2222, + "mean_token_accuracy": 0.6531848907470703, + "num_tokens": 10417256.0, + "step": 405 + }, + { + "epoch": 0.044585987261146494, + "grad_norm": 3.740719795227051, + "learning_rate": 2.2228320526893525e-07, + "loss": 1.2452, + "mean_token_accuracy": 0.6471304893493652, + "num_tokens": 10444061.0, + "step": 406 + }, + { + "epoch": 0.04469580496376016, + "grad_norm": 2.9100823402404785, + "learning_rate": 2.2283205268935233e-07, + "loss": 1.1728, + "mean_token_accuracy": 0.6714359521865845, + "num_tokens": 10472064.0, + "step": 407 + }, + { + "epoch": 0.04480562266637382, + "grad_norm": 3.4896013736724854, + "learning_rate": 2.2338090010976948e-07, + "loss": 1.0843, + "mean_token_accuracy": 0.6888709664344788, + "num_tokens": 10496295.0, + "step": 408 + }, + { + "epoch": 0.04491544036898748, + "grad_norm": 3.171466588973999, + "learning_rate": 2.239297475301866e-07, + "loss": 1.1279, + "mean_token_accuracy": 0.672443151473999, + "num_tokens": 10521466.0, + "step": 409 + }, + { + "epoch": 0.04502525807160114, + "grad_norm": 3.5170490741729736, + "learning_rate": 2.2447859495060373e-07, + "loss": 1.1912, + "mean_token_accuracy": 0.6675073504447937, + "num_tokens": 10546479.0, + "step": 410 + }, + { + "epoch": 0.0451350757742148, + "grad_norm": 3.570960521697998, + "learning_rate": 2.2502744237102085e-07, + "loss": 1.1865, + "mean_token_accuracy": 0.6669046878814697, + "num_tokens": 10574612.0, + "step": 411 + }, + { + "epoch": 0.04524489347682847, + "grad_norm": 3.492722749710083, + "learning_rate": 2.2557628979143796e-07, + "loss": 1.1603, + "mean_token_accuracy": 0.6766692996025085, + "num_tokens": 10596913.0, + "step": 412 + }, + { + "epoch": 0.04535471117944213, + "grad_norm": 3.335620880126953, + "learning_rate": 2.2612513721185508e-07, + "loss": 1.1566, + "mean_token_accuracy": 0.6630131006240845, + "num_tokens": 10621885.0, + "step": 413 + }, + { + "epoch": 0.04546452888205579, + "grad_norm": 3.86489200592041, + "learning_rate": 2.2667398463227222e-07, + "loss": 1.1371, + "mean_token_accuracy": 0.6754186153411865, + "num_tokens": 10641344.0, + "step": 414 + }, + { + "epoch": 0.045574346584669446, + "grad_norm": 4.061831951141357, + "learning_rate": 2.2722283205268934e-07, + "loss": 1.1884, + "mean_token_accuracy": 0.656424880027771, + "num_tokens": 10667502.0, + "step": 415 + }, + { + "epoch": 0.04568416428728311, + "grad_norm": 3.357311725616455, + "learning_rate": 2.2777167947310648e-07, + "loss": 1.194, + "mean_token_accuracy": 0.6600022315979004, + "num_tokens": 10697756.0, + "step": 416 + }, + { + "epoch": 0.04579398198989677, + "grad_norm": 4.168095111846924, + "learning_rate": 2.283205268935236e-07, + "loss": 1.1738, + "mean_token_accuracy": 0.6639667749404907, + "num_tokens": 10719007.0, + "step": 417 + }, + { + "epoch": 0.04590379969251043, + "grad_norm": 3.296605348587036, + "learning_rate": 2.288693743139407e-07, + "loss": 1.1809, + "mean_token_accuracy": 0.6660019159317017, + "num_tokens": 10749842.0, + "step": 418 + }, + { + "epoch": 0.04601361739512409, + "grad_norm": 3.4950826168060303, + "learning_rate": 2.2941822173435785e-07, + "loss": 1.1561, + "mean_token_accuracy": 0.6677848100662231, + "num_tokens": 10777618.0, + "step": 419 + }, + { + "epoch": 0.04612343509773775, + "grad_norm": 3.370145559310913, + "learning_rate": 2.2996706915477496e-07, + "loss": 1.1447, + "mean_token_accuracy": 0.671439528465271, + "num_tokens": 10800047.0, + "step": 420 + }, + { + "epoch": 0.04623325280035142, + "grad_norm": 3.5526089668273926, + "learning_rate": 2.3051591657519208e-07, + "loss": 1.102, + "mean_token_accuracy": 0.6764316558837891, + "num_tokens": 10824578.0, + "step": 421 + }, + { + "epoch": 0.04634307050296508, + "grad_norm": 4.008546829223633, + "learning_rate": 2.310647639956092e-07, + "loss": 1.2889, + "mean_token_accuracy": 0.6340821981430054, + "num_tokens": 10848651.0, + "step": 422 + }, + { + "epoch": 0.04645288820557874, + "grad_norm": 3.6832923889160156, + "learning_rate": 2.3161361141602634e-07, + "loss": 1.193, + "mean_token_accuracy": 0.6643158197402954, + "num_tokens": 10872522.0, + "step": 423 + }, + { + "epoch": 0.0465627059081924, + "grad_norm": 3.7269413471221924, + "learning_rate": 2.3216245883644345e-07, + "loss": 1.1153, + "mean_token_accuracy": 0.6773744821548462, + "num_tokens": 10892663.0, + "step": 424 + }, + { + "epoch": 0.04667252361080606, + "grad_norm": 3.5343477725982666, + "learning_rate": 2.327113062568606e-07, + "loss": 1.1879, + "mean_token_accuracy": 0.657645583152771, + "num_tokens": 10914562.0, + "step": 425 + }, + { + "epoch": 0.046782341313419726, + "grad_norm": 3.124443531036377, + "learning_rate": 2.3326015367727768e-07, + "loss": 1.0894, + "mean_token_accuracy": 0.6855737566947937, + "num_tokens": 10939123.0, + "step": 426 + }, + { + "epoch": 0.046892159016033386, + "grad_norm": 3.323241710662842, + "learning_rate": 2.3380900109769482e-07, + "loss": 1.2009, + "mean_token_accuracy": 0.6742897033691406, + "num_tokens": 10964837.0, + "step": 427 + }, + { + "epoch": 0.047001976718647046, + "grad_norm": 3.3786442279815674, + "learning_rate": 2.3435784851811197e-07, + "loss": 1.1268, + "mean_token_accuracy": 0.6696939468383789, + "num_tokens": 10991462.0, + "step": 428 + }, + { + "epoch": 0.047111794421260705, + "grad_norm": 3.032527208328247, + "learning_rate": 2.3490669593852908e-07, + "loss": 1.1431, + "mean_token_accuracy": 0.6687542200088501, + "num_tokens": 11019672.0, + "step": 429 + }, + { + "epoch": 0.047221612123874365, + "grad_norm": 2.931227445602417, + "learning_rate": 2.3545554335894622e-07, + "loss": 1.2596, + "mean_token_accuracy": 0.6393845081329346, + "num_tokens": 11051408.0, + "step": 430 + }, + { + "epoch": 0.04733142982648803, + "grad_norm": 3.2953851222991943, + "learning_rate": 2.360043907793633e-07, + "loss": 1.1417, + "mean_token_accuracy": 0.6741083860397339, + "num_tokens": 11075714.0, + "step": 431 + }, + { + "epoch": 0.04744124752910169, + "grad_norm": 2.8578946590423584, + "learning_rate": 2.3655323819978045e-07, + "loss": 1.1966, + "mean_token_accuracy": 0.6535441875457764, + "num_tokens": 11103537.0, + "step": 432 + }, + { + "epoch": 0.04755106523171535, + "grad_norm": 2.986063241958618, + "learning_rate": 2.3710208562019757e-07, + "loss": 1.1374, + "mean_token_accuracy": 0.674083411693573, + "num_tokens": 11128936.0, + "step": 433 + }, + { + "epoch": 0.04766088293432901, + "grad_norm": 2.7277274131774902, + "learning_rate": 2.376509330406147e-07, + "loss": 1.2225, + "mean_token_accuracy": 0.6480405926704407, + "num_tokens": 11160249.0, + "step": 434 + }, + { + "epoch": 0.04777070063694268, + "grad_norm": 3.4796829223632812, + "learning_rate": 2.3819978046103183e-07, + "loss": 1.1627, + "mean_token_accuracy": 0.6622583270072937, + "num_tokens": 11180504.0, + "step": 435 + }, + { + "epoch": 0.04788051833955634, + "grad_norm": 3.6681220531463623, + "learning_rate": 2.3874862788144894e-07, + "loss": 1.1186, + "mean_token_accuracy": 0.6778759360313416, + "num_tokens": 11200815.0, + "step": 436 + }, + { + "epoch": 0.04799033604217, + "grad_norm": 2.964482307434082, + "learning_rate": 2.3929747530186606e-07, + "loss": 1.0936, + "mean_token_accuracy": 0.6852169632911682, + "num_tokens": 11229536.0, + "step": 437 + }, + { + "epoch": 0.04810015374478366, + "grad_norm": 3.969973564147949, + "learning_rate": 2.3984632272228317e-07, + "loss": 1.2086, + "mean_token_accuracy": 0.6535100936889648, + "num_tokens": 11249966.0, + "step": 438 + }, + { + "epoch": 0.04820997144739732, + "grad_norm": 3.1626155376434326, + "learning_rate": 2.4039517014270034e-07, + "loss": 1.1685, + "mean_token_accuracy": 0.6574693918228149, + "num_tokens": 11278735.0, + "step": 439 + }, + { + "epoch": 0.048319789150010985, + "grad_norm": 4.140015602111816, + "learning_rate": 2.4094401756311745e-07, + "loss": 1.2117, + "mean_token_accuracy": 0.6500813961029053, + "num_tokens": 11298221.0, + "step": 440 + }, + { + "epoch": 0.048429606852624645, + "grad_norm": 3.293653726577759, + "learning_rate": 2.4149286498353457e-07, + "loss": 1.1823, + "mean_token_accuracy": 0.656653881072998, + "num_tokens": 11327764.0, + "step": 441 + }, + { + "epoch": 0.048539424555238304, + "grad_norm": 3.5810067653656006, + "learning_rate": 2.420417124039517e-07, + "loss": 1.1202, + "mean_token_accuracy": 0.6739243268966675, + "num_tokens": 11349547.0, + "step": 442 + }, + { + "epoch": 0.048649242257851964, + "grad_norm": 3.0180885791778564, + "learning_rate": 2.425905598243688e-07, + "loss": 1.2255, + "mean_token_accuracy": 0.6489447951316833, + "num_tokens": 11380154.0, + "step": 443 + }, + { + "epoch": 0.048759059960465624, + "grad_norm": 2.7908709049224854, + "learning_rate": 2.431394072447859e-07, + "loss": 1.1435, + "mean_token_accuracy": 0.6750683784484863, + "num_tokens": 11407980.0, + "step": 444 + }, + { + "epoch": 0.04886887766307929, + "grad_norm": 3.237319231033325, + "learning_rate": 2.436882546652031e-07, + "loss": 1.1482, + "mean_token_accuracy": 0.671581506729126, + "num_tokens": 11431417.0, + "step": 445 + }, + { + "epoch": 0.04897869536569295, + "grad_norm": 3.4036712646484375, + "learning_rate": 2.442371020856202e-07, + "loss": 1.0679, + "mean_token_accuracy": 0.6871690154075623, + "num_tokens": 11454490.0, + "step": 446 + }, + { + "epoch": 0.04908851306830661, + "grad_norm": 4.133213043212891, + "learning_rate": 2.447859495060373e-07, + "loss": 1.1675, + "mean_token_accuracy": 0.6613529920578003, + "num_tokens": 11474238.0, + "step": 447 + }, + { + "epoch": 0.04919833077092027, + "grad_norm": 3.630682945251465, + "learning_rate": 2.4533479692645443e-07, + "loss": 1.1936, + "mean_token_accuracy": 0.651604413986206, + "num_tokens": 11494924.0, + "step": 448 + }, + { + "epoch": 0.04930814847353393, + "grad_norm": 3.1463706493377686, + "learning_rate": 2.4588364434687154e-07, + "loss": 1.1435, + "mean_token_accuracy": 0.6703542470932007, + "num_tokens": 11519667.0, + "step": 449 + }, + { + "epoch": 0.0494179661761476, + "grad_norm": 3.282409191131592, + "learning_rate": 2.4643249176728866e-07, + "loss": 1.1254, + "mean_token_accuracy": 0.6807973384857178, + "num_tokens": 11546769.0, + "step": 450 + }, + { + "epoch": 0.04952778387876126, + "grad_norm": 3.6076998710632324, + "learning_rate": 2.4698133918770583e-07, + "loss": 1.1983, + "mean_token_accuracy": 0.6551201343536377, + "num_tokens": 11569240.0, + "step": 451 + }, + { + "epoch": 0.04963760158137492, + "grad_norm": 3.290274143218994, + "learning_rate": 2.4753018660812294e-07, + "loss": 1.1334, + "mean_token_accuracy": 0.6707939505577087, + "num_tokens": 11592862.0, + "step": 452 + }, + { + "epoch": 0.04974741928398858, + "grad_norm": 2.9652655124664307, + "learning_rate": 2.4807903402854006e-07, + "loss": 1.1867, + "mean_token_accuracy": 0.6548534631729126, + "num_tokens": 11617692.0, + "step": 453 + }, + { + "epoch": 0.049857236986602244, + "grad_norm": 3.1510283946990967, + "learning_rate": 2.486278814489572e-07, + "loss": 1.1867, + "mean_token_accuracy": 0.6599615812301636, + "num_tokens": 11639816.0, + "step": 454 + }, + { + "epoch": 0.0499670546892159, + "grad_norm": 3.6012353897094727, + "learning_rate": 2.491767288693743e-07, + "loss": 1.1122, + "mean_token_accuracy": 0.6677559614181519, + "num_tokens": 11661671.0, + "step": 455 + }, + { + "epoch": 0.05007687239182956, + "grad_norm": 3.0392727851867676, + "learning_rate": 2.4972557628979146e-07, + "loss": 1.2495, + "mean_token_accuracy": 0.6418200135231018, + "num_tokens": 11691676.0, + "step": 456 + }, + { + "epoch": 0.05018669009444322, + "grad_norm": 2.714388608932495, + "learning_rate": 2.5027442371020857e-07, + "loss": 1.142, + "mean_token_accuracy": 0.6866999864578247, + "num_tokens": 11718964.0, + "step": 457 + }, + { + "epoch": 0.05029650779705688, + "grad_norm": 3.1240084171295166, + "learning_rate": 2.508232711306257e-07, + "loss": 1.1847, + "mean_token_accuracy": 0.6602873802185059, + "num_tokens": 11744111.0, + "step": 458 + }, + { + "epoch": 0.05040632549967055, + "grad_norm": 2.8125317096710205, + "learning_rate": 2.513721185510428e-07, + "loss": 1.1683, + "mean_token_accuracy": 0.6706335544586182, + "num_tokens": 11771685.0, + "step": 459 + }, + { + "epoch": 0.05051614320228421, + "grad_norm": 2.5294265747070312, + "learning_rate": 2.519209659714599e-07, + "loss": 1.215, + "mean_token_accuracy": 0.6509800553321838, + "num_tokens": 11806324.0, + "step": 460 + }, + { + "epoch": 0.05062596090489787, + "grad_norm": 2.8408241271972656, + "learning_rate": 2.524698133918771e-07, + "loss": 1.1029, + "mean_token_accuracy": 0.6777981519699097, + "num_tokens": 11835649.0, + "step": 461 + }, + { + "epoch": 0.05073577860751153, + "grad_norm": 3.914680004119873, + "learning_rate": 2.5301866081229415e-07, + "loss": 1.1826, + "mean_token_accuracy": 0.660986602306366, + "num_tokens": 11862291.0, + "step": 462 + }, + { + "epoch": 0.05084559631012519, + "grad_norm": 2.9291460514068604, + "learning_rate": 2.5356750823271126e-07, + "loss": 1.1435, + "mean_token_accuracy": 0.67010897397995, + "num_tokens": 11890532.0, + "step": 463 + }, + { + "epoch": 0.050955414012738856, + "grad_norm": 3.0258138179779053, + "learning_rate": 2.5411635565312843e-07, + "loss": 1.1491, + "mean_token_accuracy": 0.666933000087738, + "num_tokens": 11915387.0, + "step": 464 + }, + { + "epoch": 0.051065231715352516, + "grad_norm": 2.8931541442871094, + "learning_rate": 2.5466520307354555e-07, + "loss": 1.1179, + "mean_token_accuracy": 0.6810283660888672, + "num_tokens": 11946245.0, + "step": 465 + }, + { + "epoch": 0.051175049417966176, + "grad_norm": 3.4376108646392822, + "learning_rate": 2.5521405049396266e-07, + "loss": 1.104, + "mean_token_accuracy": 0.6771970391273499, + "num_tokens": 11967486.0, + "step": 466 + }, + { + "epoch": 0.051284867120579836, + "grad_norm": 3.360867977142334, + "learning_rate": 2.557628979143798e-07, + "loss": 1.1245, + "mean_token_accuracy": 0.6736201047897339, + "num_tokens": 11990241.0, + "step": 467 + }, + { + "epoch": 0.051394684823193496, + "grad_norm": 2.9831905364990234, + "learning_rate": 2.5631174533479695e-07, + "loss": 1.0258, + "mean_token_accuracy": 0.6966726779937744, + "num_tokens": 12013907.0, + "step": 468 + }, + { + "epoch": 0.05150450252580716, + "grad_norm": 2.9584596157073975, + "learning_rate": 2.5686059275521406e-07, + "loss": 1.1665, + "mean_token_accuracy": 0.667405366897583, + "num_tokens": 12038557.0, + "step": 469 + }, + { + "epoch": 0.05161432022842082, + "grad_norm": 2.8480851650238037, + "learning_rate": 2.574094401756312e-07, + "loss": 1.1694, + "mean_token_accuracy": 0.6671779155731201, + "num_tokens": 12064483.0, + "step": 470 + }, + { + "epoch": 0.05172413793103448, + "grad_norm": 2.9270970821380615, + "learning_rate": 2.5795828759604824e-07, + "loss": 1.3122, + "mean_token_accuracy": 0.6285170912742615, + "num_tokens": 12094965.0, + "step": 471 + }, + { + "epoch": 0.05183395563364814, + "grad_norm": 3.1700737476348877, + "learning_rate": 2.585071350164654e-07, + "loss": 1.1453, + "mean_token_accuracy": 0.6642838716506958, + "num_tokens": 12117540.0, + "step": 472 + }, + { + "epoch": 0.05194377333626181, + "grad_norm": 2.806879997253418, + "learning_rate": 2.590559824368825e-07, + "loss": 1.0943, + "mean_token_accuracy": 0.6816945672035217, + "num_tokens": 12143214.0, + "step": 473 + }, + { + "epoch": 0.05205359103887547, + "grad_norm": 3.1003410816192627, + "learning_rate": 2.5960482985729964e-07, + "loss": 1.1221, + "mean_token_accuracy": 0.6755359172821045, + "num_tokens": 12165004.0, + "step": 474 + }, + { + "epoch": 0.05216340874148913, + "grad_norm": 2.698079824447632, + "learning_rate": 2.601536772777168e-07, + "loss": 1.1513, + "mean_token_accuracy": 0.6742467284202576, + "num_tokens": 12193627.0, + "step": 475 + }, + { + "epoch": 0.05227322644410279, + "grad_norm": 2.887342691421509, + "learning_rate": 2.607025246981339e-07, + "loss": 1.1622, + "mean_token_accuracy": 0.6667620539665222, + "num_tokens": 12220553.0, + "step": 476 + }, + { + "epoch": 0.05238304414671645, + "grad_norm": 3.092759847640991, + "learning_rate": 2.6125137211855104e-07, + "loss": 1.2226, + "mean_token_accuracy": 0.6603109240531921, + "num_tokens": 12246083.0, + "step": 477 + }, + { + "epoch": 0.052492861849330115, + "grad_norm": 3.1315975189208984, + "learning_rate": 2.6180021953896815e-07, + "loss": 1.0998, + "mean_token_accuracy": 0.6825853586196899, + "num_tokens": 12267660.0, + "step": 478 + }, + { + "epoch": 0.052602679551943775, + "grad_norm": 3.0190930366516113, + "learning_rate": 2.623490669593853e-07, + "loss": 1.2346, + "mean_token_accuracy": 0.646297812461853, + "num_tokens": 12292223.0, + "step": 479 + }, + { + "epoch": 0.052712497254557435, + "grad_norm": 2.8660390377044678, + "learning_rate": 2.6289791437980244e-07, + "loss": 1.0658, + "mean_token_accuracy": 0.6882008910179138, + "num_tokens": 12316698.0, + "step": 480 + }, + { + "epoch": 0.052822314957171095, + "grad_norm": 2.8453469276428223, + "learning_rate": 2.634467618002195e-07, + "loss": 1.1192, + "mean_token_accuracy": 0.6773805618286133, + "num_tokens": 12343689.0, + "step": 481 + }, + { + "epoch": 0.052932132659784754, + "grad_norm": 2.7977566719055176, + "learning_rate": 2.639956092206366e-07, + "loss": 1.0354, + "mean_token_accuracy": 0.6940751075744629, + "num_tokens": 12369871.0, + "step": 482 + }, + { + "epoch": 0.05304195036239842, + "grad_norm": 2.4073808193206787, + "learning_rate": 2.645444566410538e-07, + "loss": 1.0687, + "mean_token_accuracy": 0.6821310520172119, + "num_tokens": 12400693.0, + "step": 483 + }, + { + "epoch": 0.05315176806501208, + "grad_norm": 2.658350944519043, + "learning_rate": 2.650933040614709e-07, + "loss": 1.2707, + "mean_token_accuracy": 0.6451667547225952, + "num_tokens": 12430564.0, + "step": 484 + }, + { + "epoch": 0.05326158576762574, + "grad_norm": 3.4002156257629395, + "learning_rate": 2.65642151481888e-07, + "loss": 1.1651, + "mean_token_accuracy": 0.6635078191757202, + "num_tokens": 12450785.0, + "step": 485 + }, + { + "epoch": 0.0533714034702394, + "grad_norm": 3.3879051208496094, + "learning_rate": 2.6619099890230513e-07, + "loss": 1.2362, + "mean_token_accuracy": 0.6495607495307922, + "num_tokens": 12475626.0, + "step": 486 + }, + { + "epoch": 0.05348122117285306, + "grad_norm": 3.059056043624878, + "learning_rate": 2.667398463227223e-07, + "loss": 1.0711, + "mean_token_accuracy": 0.6813543438911438, + "num_tokens": 12498208.0, + "step": 487 + }, + { + "epoch": 0.05359103887546673, + "grad_norm": 2.6529603004455566, + "learning_rate": 2.672886937431394e-07, + "loss": 1.1203, + "mean_token_accuracy": 0.6784849762916565, + "num_tokens": 12527307.0, + "step": 488 + }, + { + "epoch": 0.05370085657808039, + "grad_norm": 2.9653172492980957, + "learning_rate": 2.678375411635565e-07, + "loss": 1.0905, + "mean_token_accuracy": 0.6773719787597656, + "num_tokens": 12551265.0, + "step": 489 + }, + { + "epoch": 0.05381067428069405, + "grad_norm": 2.883727550506592, + "learning_rate": 2.6838638858397364e-07, + "loss": 1.0984, + "mean_token_accuracy": 0.6764349937438965, + "num_tokens": 12573666.0, + "step": 490 + }, + { + "epoch": 0.05392049198330771, + "grad_norm": 3.073984146118164, + "learning_rate": 2.6893523600439076e-07, + "loss": 1.145, + "mean_token_accuracy": 0.6698759198188782, + "num_tokens": 12598291.0, + "step": 491 + }, + { + "epoch": 0.054030309685921374, + "grad_norm": 3.2819459438323975, + "learning_rate": 2.6948408342480787e-07, + "loss": 1.115, + "mean_token_accuracy": 0.6723285913467407, + "num_tokens": 12621480.0, + "step": 492 + }, + { + "epoch": 0.054140127388535034, + "grad_norm": 2.7337517738342285, + "learning_rate": 2.70032930845225e-07, + "loss": 1.0839, + "mean_token_accuracy": 0.6811389923095703, + "num_tokens": 12648364.0, + "step": 493 + }, + { + "epoch": 0.054249945091148694, + "grad_norm": 3.0832650661468506, + "learning_rate": 2.7058177826564215e-07, + "loss": 1.1861, + "mean_token_accuracy": 0.6642299890518188, + "num_tokens": 12675332.0, + "step": 494 + }, + { + "epoch": 0.05435976279376235, + "grad_norm": 3.281343936920166, + "learning_rate": 2.7113062568605927e-07, + "loss": 1.1148, + "mean_token_accuracy": 0.6767305135726929, + "num_tokens": 12698179.0, + "step": 495 + }, + { + "epoch": 0.05446958049637601, + "grad_norm": 2.710768938064575, + "learning_rate": 2.716794731064764e-07, + "loss": 1.1548, + "mean_token_accuracy": 0.6675187349319458, + "num_tokens": 12725546.0, + "step": 496 + }, + { + "epoch": 0.05457939819898968, + "grad_norm": 3.240056037902832, + "learning_rate": 2.722283205268935e-07, + "loss": 1.0748, + "mean_token_accuracy": 0.6876198053359985, + "num_tokens": 12746157.0, + "step": 497 + }, + { + "epoch": 0.05468921590160334, + "grad_norm": 2.6083977222442627, + "learning_rate": 2.7277716794731067e-07, + "loss": 1.2379, + "mean_token_accuracy": 0.6457153558731079, + "num_tokens": 12774899.0, + "step": 498 + }, + { + "epoch": 0.054799033604217, + "grad_norm": 2.7578530311584473, + "learning_rate": 2.733260153677278e-07, + "loss": 1.182, + "mean_token_accuracy": 0.6580971479415894, + "num_tokens": 12803890.0, + "step": 499 + }, + { + "epoch": 0.05490885130683066, + "grad_norm": 2.90543270111084, + "learning_rate": 2.7387486278814485e-07, + "loss": 1.0983, + "mean_token_accuracy": 0.6749898791313171, + "num_tokens": 12827113.0, + "step": 500 + }, + { + "epoch": 0.05501866900944432, + "grad_norm": 2.632321357727051, + "learning_rate": 2.74423710208562e-07, + "loss": 1.1159, + "mean_token_accuracy": 0.6745585203170776, + "num_tokens": 12851692.0, + "step": 501 + }, + { + "epoch": 0.055128486712057986, + "grad_norm": 2.4654643535614014, + "learning_rate": 2.7497255762897913e-07, + "loss": 1.2182, + "mean_token_accuracy": 0.650181770324707, + "num_tokens": 12882328.0, + "step": 502 + }, + { + "epoch": 0.055238304414671646, + "grad_norm": 3.1645944118499756, + "learning_rate": 2.7552140504939624e-07, + "loss": 1.1668, + "mean_token_accuracy": 0.6542862057685852, + "num_tokens": 12903041.0, + "step": 503 + }, + { + "epoch": 0.055348122117285306, + "grad_norm": 2.5699563026428223, + "learning_rate": 2.7607025246981336e-07, + "loss": 1.1111, + "mean_token_accuracy": 0.6777710914611816, + "num_tokens": 12931102.0, + "step": 504 + }, + { + "epoch": 0.055457939819898966, + "grad_norm": 3.2762081623077393, + "learning_rate": 2.7661909989023053e-07, + "loss": 1.0281, + "mean_token_accuracy": 0.6959967613220215, + "num_tokens": 12952312.0, + "step": 505 + }, + { + "epoch": 0.055567757522512626, + "grad_norm": 2.769303560256958, + "learning_rate": 2.7716794731064764e-07, + "loss": 1.1828, + "mean_token_accuracy": 0.6641866564750671, + "num_tokens": 12979880.0, + "step": 506 + }, + { + "epoch": 0.05567757522512629, + "grad_norm": 2.9624063968658447, + "learning_rate": 2.7771679473106476e-07, + "loss": 1.0883, + "mean_token_accuracy": 0.6833668947219849, + "num_tokens": 13003013.0, + "step": 507 + }, + { + "epoch": 0.05578739292773995, + "grad_norm": 3.1297948360443115, + "learning_rate": 2.7826564215148187e-07, + "loss": 1.1527, + "mean_token_accuracy": 0.6673184633255005, + "num_tokens": 13024395.0, + "step": 508 + }, + { + "epoch": 0.05589721063035361, + "grad_norm": 2.8415236473083496, + "learning_rate": 2.78814489571899e-07, + "loss": 1.0209, + "mean_token_accuracy": 0.6917712688446045, + "num_tokens": 13045856.0, + "step": 509 + }, + { + "epoch": 0.05600702833296727, + "grad_norm": 3.0517003536224365, + "learning_rate": 2.793633369923161e-07, + "loss": 1.0096, + "mean_token_accuracy": 0.7054553627967834, + "num_tokens": 13066297.0, + "step": 510 + }, + { + "epoch": 0.05611684603558094, + "grad_norm": 2.732541561126709, + "learning_rate": 2.799121844127332e-07, + "loss": 1.1716, + "mean_token_accuracy": 0.6625338196754456, + "num_tokens": 13091745.0, + "step": 511 + }, + { + "epoch": 0.0562266637381946, + "grad_norm": 2.9642605781555176, + "learning_rate": 2.804610318331504e-07, + "loss": 1.1254, + "mean_token_accuracy": 0.6761672496795654, + "num_tokens": 13113941.0, + "step": 512 + }, + { + "epoch": 0.05633648144080826, + "grad_norm": 3.1821115016937256, + "learning_rate": 2.810098792535675e-07, + "loss": 1.0496, + "mean_token_accuracy": 0.6921664476394653, + "num_tokens": 13135148.0, + "step": 513 + }, + { + "epoch": 0.05644629914342192, + "grad_norm": 3.603625535964966, + "learning_rate": 2.815587266739846e-07, + "loss": 1.0963, + "mean_token_accuracy": 0.6789394617080688, + "num_tokens": 13152735.0, + "step": 514 + }, + { + "epoch": 0.05655611684603558, + "grad_norm": 3.2870125770568848, + "learning_rate": 2.8210757409440173e-07, + "loss": 1.0913, + "mean_token_accuracy": 0.683628499507904, + "num_tokens": 13175691.0, + "step": 515 + }, + { + "epoch": 0.056665934548649245, + "grad_norm": 2.689068555831909, + "learning_rate": 2.826564215148189e-07, + "loss": 1.1881, + "mean_token_accuracy": 0.6525189876556396, + "num_tokens": 13203987.0, + "step": 516 + }, + { + "epoch": 0.056775752251262905, + "grad_norm": 2.9231715202331543, + "learning_rate": 2.83205268935236e-07, + "loss": 1.0721, + "mean_token_accuracy": 0.6950573325157166, + "num_tokens": 13226761.0, + "step": 517 + }, + { + "epoch": 0.056885569953876565, + "grad_norm": 2.438208818435669, + "learning_rate": 2.8375411635565313e-07, + "loss": 1.1197, + "mean_token_accuracy": 0.6760292649269104, + "num_tokens": 13258157.0, + "step": 518 + }, + { + "epoch": 0.056995387656490225, + "grad_norm": 2.6710262298583984, + "learning_rate": 2.843029637760702e-07, + "loss": 1.1914, + "mean_token_accuracy": 0.6589288711547852, + "num_tokens": 13285619.0, + "step": 519 + }, + { + "epoch": 0.057105205359103885, + "grad_norm": 2.828749895095825, + "learning_rate": 2.8485181119648736e-07, + "loss": 1.021, + "mean_token_accuracy": 0.6973201036453247, + "num_tokens": 13312008.0, + "step": 520 + }, + { + "epoch": 0.05721502306171755, + "grad_norm": 2.8221583366394043, + "learning_rate": 2.854006586169045e-07, + "loss": 1.1175, + "mean_token_accuracy": 0.6767851114273071, + "num_tokens": 13337915.0, + "step": 521 + }, + { + "epoch": 0.05732484076433121, + "grad_norm": 2.4107959270477295, + "learning_rate": 2.859495060373216e-07, + "loss": 1.2076, + "mean_token_accuracy": 0.6588525772094727, + "num_tokens": 13367407.0, + "step": 522 + }, + { + "epoch": 0.05743465846694487, + "grad_norm": 2.711371421813965, + "learning_rate": 2.8649835345773876e-07, + "loss": 1.2049, + "mean_token_accuracy": 0.6634594202041626, + "num_tokens": 13391886.0, + "step": 523 + }, + { + "epoch": 0.05754447616955853, + "grad_norm": 2.868455410003662, + "learning_rate": 2.870472008781559e-07, + "loss": 1.1341, + "mean_token_accuracy": 0.6703625321388245, + "num_tokens": 13414307.0, + "step": 524 + }, + { + "epoch": 0.05765429387217219, + "grad_norm": 2.728666067123413, + "learning_rate": 2.87596048298573e-07, + "loss": 1.1802, + "mean_token_accuracy": 0.6546539068222046, + "num_tokens": 13439390.0, + "step": 525 + }, + { + "epoch": 0.05776411157478586, + "grad_norm": 2.9521799087524414, + "learning_rate": 2.881448957189901e-07, + "loss": 1.1078, + "mean_token_accuracy": 0.6655943393707275, + "num_tokens": 13461243.0, + "step": 526 + }, + { + "epoch": 0.05787392927739952, + "grad_norm": 2.8348939418792725, + "learning_rate": 2.886937431394073e-07, + "loss": 1.1342, + "mean_token_accuracy": 0.666171669960022, + "num_tokens": 13484439.0, + "step": 527 + }, + { + "epoch": 0.05798374698001318, + "grad_norm": 2.6914567947387695, + "learning_rate": 2.8924259055982434e-07, + "loss": 1.1511, + "mean_token_accuracy": 0.6788211464881897, + "num_tokens": 13509211.0, + "step": 528 + }, + { + "epoch": 0.05809356468262684, + "grad_norm": 2.983921766281128, + "learning_rate": 2.8979143798024145e-07, + "loss": 1.1279, + "mean_token_accuracy": 0.6693599820137024, + "num_tokens": 13529467.0, + "step": 529 + }, + { + "epoch": 0.058203382385240504, + "grad_norm": 2.799428939819336, + "learning_rate": 2.9034028540065857e-07, + "loss": 1.114, + "mean_token_accuracy": 0.6703866720199585, + "num_tokens": 13553998.0, + "step": 530 + }, + { + "epoch": 0.058313200087854164, + "grad_norm": 2.8770217895507812, + "learning_rate": 2.9088913282107574e-07, + "loss": 1.0897, + "mean_token_accuracy": 0.6825820207595825, + "num_tokens": 13578734.0, + "step": 531 + }, + { + "epoch": 0.058423017790467824, + "grad_norm": 2.7728030681610107, + "learning_rate": 2.9143798024149285e-07, + "loss": 1.1463, + "mean_token_accuracy": 0.6744999885559082, + "num_tokens": 13604646.0, + "step": 532 + }, + { + "epoch": 0.058532835493081484, + "grad_norm": 2.638007402420044, + "learning_rate": 2.9198682766190997e-07, + "loss": 1.0537, + "mean_token_accuracy": 0.686143696308136, + "num_tokens": 13630798.0, + "step": 533 + }, + { + "epoch": 0.058642653195695144, + "grad_norm": 3.013847827911377, + "learning_rate": 2.9253567508232713e-07, + "loss": 1.0033, + "mean_token_accuracy": 0.7039613127708435, + "num_tokens": 13650425.0, + "step": 534 + }, + { + "epoch": 0.05875247089830881, + "grad_norm": 3.3560562133789062, + "learning_rate": 2.9308452250274425e-07, + "loss": 1.1228, + "mean_token_accuracy": 0.6700910329818726, + "num_tokens": 13668806.0, + "step": 535 + }, + { + "epoch": 0.05886228860092247, + "grad_norm": 2.8054747581481934, + "learning_rate": 2.9363336992316136e-07, + "loss": 1.1089, + "mean_token_accuracy": 0.6787279844284058, + "num_tokens": 13693058.0, + "step": 536 + }, + { + "epoch": 0.05897210630353613, + "grad_norm": 2.934924364089966, + "learning_rate": 2.941822173435785e-07, + "loss": 1.0607, + "mean_token_accuracy": 0.6852309703826904, + "num_tokens": 13714602.0, + "step": 537 + }, + { + "epoch": 0.05908192400614979, + "grad_norm": 2.3669047355651855, + "learning_rate": 2.947310647639956e-07, + "loss": 1.1112, + "mean_token_accuracy": 0.6814795732498169, + "num_tokens": 13746608.0, + "step": 538 + }, + { + "epoch": 0.05919174170876345, + "grad_norm": 2.3134806156158447, + "learning_rate": 2.952799121844127e-07, + "loss": 1.1773, + "mean_token_accuracy": 0.6554323434829712, + "num_tokens": 13777608.0, + "step": 539 + }, + { + "epoch": 0.05930155941137712, + "grad_norm": 2.5348520278930664, + "learning_rate": 2.958287596048298e-07, + "loss": 1.2056, + "mean_token_accuracy": 0.6531015634536743, + "num_tokens": 13806648.0, + "step": 540 + }, + { + "epoch": 0.059411377113990776, + "grad_norm": 2.801880359649658, + "learning_rate": 2.9637760702524694e-07, + "loss": 1.1836, + "mean_token_accuracy": 0.6611485481262207, + "num_tokens": 13830348.0, + "step": 541 + }, + { + "epoch": 0.059521194816604436, + "grad_norm": 2.433107852935791, + "learning_rate": 2.969264544456641e-07, + "loss": 1.0963, + "mean_token_accuracy": 0.6832491159439087, + "num_tokens": 13857470.0, + "step": 542 + }, + { + "epoch": 0.059631012519218096, + "grad_norm": 2.5923335552215576, + "learning_rate": 2.974753018660812e-07, + "loss": 1.1259, + "mean_token_accuracy": 0.6813538074493408, + "num_tokens": 13883205.0, + "step": 543 + }, + { + "epoch": 0.059740830221831756, + "grad_norm": 2.4768753051757812, + "learning_rate": 2.9802414928649834e-07, + "loss": 1.1788, + "mean_token_accuracy": 0.6607742309570312, + "num_tokens": 13912934.0, + "step": 544 + }, + { + "epoch": 0.05985064792444542, + "grad_norm": 2.693138599395752, + "learning_rate": 2.9857299670691546e-07, + "loss": 1.186, + "mean_token_accuracy": 0.6578078866004944, + "num_tokens": 13938533.0, + "step": 545 + }, + { + "epoch": 0.05996046562705908, + "grad_norm": 2.5871453285217285, + "learning_rate": 2.991218441273326e-07, + "loss": 1.1016, + "mean_token_accuracy": 0.676104724407196, + "num_tokens": 13967802.0, + "step": 546 + }, + { + "epoch": 0.06007028332967274, + "grad_norm": 2.4650206565856934, + "learning_rate": 2.996706915477497e-07, + "loss": 1.117, + "mean_token_accuracy": 0.6739547252655029, + "num_tokens": 13996835.0, + "step": 547 + }, + { + "epoch": 0.0601801010322864, + "grad_norm": 2.2236521244049072, + "learning_rate": 3.002195389681668e-07, + "loss": 1.2451, + "mean_token_accuracy": 0.6393991708755493, + "num_tokens": 14033902.0, + "step": 548 + }, + { + "epoch": 0.06028991873490007, + "grad_norm": 2.463886022567749, + "learning_rate": 3.0076838638858397e-07, + "loss": 1.1056, + "mean_token_accuracy": 0.6753267049789429, + "num_tokens": 14062254.0, + "step": 549 + }, + { + "epoch": 0.06039973643751373, + "grad_norm": 2.6000351905822754, + "learning_rate": 3.013172338090011e-07, + "loss": 1.145, + "mean_token_accuracy": 0.6673499345779419, + "num_tokens": 14090504.0, + "step": 550 + }, + { + "epoch": 0.06050955414012739, + "grad_norm": 2.468200445175171, + "learning_rate": 3.018660812294182e-07, + "loss": 1.0781, + "mean_token_accuracy": 0.6820721626281738, + "num_tokens": 14119129.0, + "step": 551 + }, + { + "epoch": 0.06061937184274105, + "grad_norm": 2.64125394821167, + "learning_rate": 3.024149286498353e-07, + "loss": 1.1813, + "mean_token_accuracy": 0.6577390432357788, + "num_tokens": 14144531.0, + "step": 552 + }, + { + "epoch": 0.06072918954535471, + "grad_norm": 2.4491803646087646, + "learning_rate": 3.029637760702525e-07, + "loss": 1.1034, + "mean_token_accuracy": 0.6821731328964233, + "num_tokens": 14175635.0, + "step": 553 + }, + { + "epoch": 0.060839007247968376, + "grad_norm": 2.3585867881774902, + "learning_rate": 3.035126234906696e-07, + "loss": 1.214, + "mean_token_accuracy": 0.6556034088134766, + "num_tokens": 14209116.0, + "step": 554 + }, + { + "epoch": 0.060948824950582035, + "grad_norm": 2.6563117504119873, + "learning_rate": 3.040614709110867e-07, + "loss": 1.1603, + "mean_token_accuracy": 0.6637877225875854, + "num_tokens": 14234732.0, + "step": 555 + }, + { + "epoch": 0.061058642653195695, + "grad_norm": 2.9935991764068604, + "learning_rate": 3.046103183315038e-07, + "loss": 1.1546, + "mean_token_accuracy": 0.6707183122634888, + "num_tokens": 14254717.0, + "step": 556 + }, + { + "epoch": 0.061168460355809355, + "grad_norm": 2.7665767669677734, + "learning_rate": 3.0515916575192094e-07, + "loss": 1.1202, + "mean_token_accuracy": 0.6778849363327026, + "num_tokens": 14277674.0, + "step": 557 + }, + { + "epoch": 0.061278278058423015, + "grad_norm": 2.507866621017456, + "learning_rate": 3.0570801317233806e-07, + "loss": 1.1375, + "mean_token_accuracy": 0.6725822687149048, + "num_tokens": 14306031.0, + "step": 558 + }, + { + "epoch": 0.06138809576103668, + "grad_norm": 2.4755313396453857, + "learning_rate": 3.062568605927552e-07, + "loss": 1.0554, + "mean_token_accuracy": 0.6900932788848877, + "num_tokens": 14333768.0, + "step": 559 + }, + { + "epoch": 0.06149791346365034, + "grad_norm": 3.0875961780548096, + "learning_rate": 3.0680570801317234e-07, + "loss": 1.1248, + "mean_token_accuracy": 0.6716578006744385, + "num_tokens": 14358193.0, + "step": 560 + }, + { + "epoch": 0.061607731166264, + "grad_norm": 2.6479523181915283, + "learning_rate": 3.0735455543358946e-07, + "loss": 1.2009, + "mean_token_accuracy": 0.6664425730705261, + "num_tokens": 14383088.0, + "step": 561 + }, + { + "epoch": 0.06171754886887766, + "grad_norm": 2.1953694820404053, + "learning_rate": 3.0790340285400657e-07, + "loss": 1.0963, + "mean_token_accuracy": 0.6813512444496155, + "num_tokens": 14419916.0, + "step": 562 + }, + { + "epoch": 0.06182736657149132, + "grad_norm": 2.7643883228302, + "learning_rate": 3.084522502744237e-07, + "loss": 1.0922, + "mean_token_accuracy": 0.6784512400627136, + "num_tokens": 14443147.0, + "step": 563 + }, + { + "epoch": 0.06193718427410499, + "grad_norm": 2.1949679851531982, + "learning_rate": 3.0900109769484086e-07, + "loss": 1.2435, + "mean_token_accuracy": 0.6430532932281494, + "num_tokens": 14478348.0, + "step": 564 + }, + { + "epoch": 0.06204700197671865, + "grad_norm": 2.6816627979278564, + "learning_rate": 3.0954994511525797e-07, + "loss": 1.084, + "mean_token_accuracy": 0.6738042831420898, + "num_tokens": 14503502.0, + "step": 565 + }, + { + "epoch": 0.06215681967933231, + "grad_norm": 2.893843173980713, + "learning_rate": 3.1009879253567503e-07, + "loss": 1.1455, + "mean_token_accuracy": 0.6710913777351379, + "num_tokens": 14524558.0, + "step": 566 + }, + { + "epoch": 0.06226663738194597, + "grad_norm": 2.17971134185791, + "learning_rate": 3.1064763995609215e-07, + "loss": 1.1618, + "mean_token_accuracy": 0.6686532497406006, + "num_tokens": 14560425.0, + "step": 567 + }, + { + "epoch": 0.062376455084559634, + "grad_norm": 2.2719364166259766, + "learning_rate": 3.111964873765093e-07, + "loss": 1.1542, + "mean_token_accuracy": 0.6646029949188232, + "num_tokens": 14590756.0, + "step": 568 + }, + { + "epoch": 0.062486272787173294, + "grad_norm": 2.7703609466552734, + "learning_rate": 3.1174533479692643e-07, + "loss": 1.1005, + "mean_token_accuracy": 0.6805640459060669, + "num_tokens": 14613591.0, + "step": 569 + }, + { + "epoch": 0.06259609048978695, + "grad_norm": 2.19685697555542, + "learning_rate": 3.1229418221734355e-07, + "loss": 1.219, + "mean_token_accuracy": 0.6516499519348145, + "num_tokens": 14647993.0, + "step": 570 + }, + { + "epoch": 0.06270590819240061, + "grad_norm": 3.1371371746063232, + "learning_rate": 3.128430296377607e-07, + "loss": 1.0319, + "mean_token_accuracy": 0.6959694027900696, + "num_tokens": 14667902.0, + "step": 571 + }, + { + "epoch": 0.06281572589501427, + "grad_norm": 2.696753978729248, + "learning_rate": 3.1339187705817783e-07, + "loss": 1.1549, + "mean_token_accuracy": 0.6533870697021484, + "num_tokens": 14693619.0, + "step": 572 + }, + { + "epoch": 0.06292554359762793, + "grad_norm": 2.9814260005950928, + "learning_rate": 3.1394072447859495e-07, + "loss": 1.0729, + "mean_token_accuracy": 0.6830410361289978, + "num_tokens": 14715866.0, + "step": 573 + }, + { + "epoch": 0.0630353613002416, + "grad_norm": 2.6987199783325195, + "learning_rate": 3.1448957189901206e-07, + "loss": 1.1324, + "mean_token_accuracy": 0.6728073358535767, + "num_tokens": 14743363.0, + "step": 574 + }, + { + "epoch": 0.06314517900285525, + "grad_norm": 2.577051877975464, + "learning_rate": 3.1503841931942923e-07, + "loss": 1.1338, + "mean_token_accuracy": 0.6698927879333496, + "num_tokens": 14772333.0, + "step": 575 + }, + { + "epoch": 0.06325499670546893, + "grad_norm": 3.0731990337371826, + "learning_rate": 3.155872667398463e-07, + "loss": 1.0937, + "mean_token_accuracy": 0.6783382892608643, + "num_tokens": 14793677.0, + "step": 576 + }, + { + "epoch": 0.06336481440808259, + "grad_norm": 2.150485038757324, + "learning_rate": 3.161361141602634e-07, + "loss": 1.2999, + "mean_token_accuracy": 0.628605842590332, + "num_tokens": 14830194.0, + "step": 577 + }, + { + "epoch": 0.06347463211069625, + "grad_norm": 2.394794225692749, + "learning_rate": 3.166849615806805e-07, + "loss": 1.1115, + "mean_token_accuracy": 0.6758230924606323, + "num_tokens": 14859278.0, + "step": 578 + }, + { + "epoch": 0.0635844498133099, + "grad_norm": 2.642153024673462, + "learning_rate": 3.172338090010977e-07, + "loss": 1.1835, + "mean_token_accuracy": 0.6624569296836853, + "num_tokens": 14886851.0, + "step": 579 + }, + { + "epoch": 0.06369426751592357, + "grad_norm": 2.6435186862945557, + "learning_rate": 3.177826564215148e-07, + "loss": 1.0882, + "mean_token_accuracy": 0.6832517385482788, + "num_tokens": 14911595.0, + "step": 580 + }, + { + "epoch": 0.06380408521853723, + "grad_norm": 2.8614954948425293, + "learning_rate": 3.183315038419319e-07, + "loss": 1.0966, + "mean_token_accuracy": 0.67786705493927, + "num_tokens": 14933951.0, + "step": 581 + }, + { + "epoch": 0.06391390292115089, + "grad_norm": 2.5580031871795654, + "learning_rate": 3.188803512623491e-07, + "loss": 1.227, + "mean_token_accuracy": 0.6430533528327942, + "num_tokens": 14960598.0, + "step": 582 + }, + { + "epoch": 0.06402372062376455, + "grad_norm": 2.751417636871338, + "learning_rate": 3.194291986827662e-07, + "loss": 1.1173, + "mean_token_accuracy": 0.6746628284454346, + "num_tokens": 14983414.0, + "step": 583 + }, + { + "epoch": 0.0641335383263782, + "grad_norm": 3.115168333053589, + "learning_rate": 3.199780461031833e-07, + "loss": 1.1616, + "mean_token_accuracy": 0.6809801459312439, + "num_tokens": 15004723.0, + "step": 584 + }, + { + "epoch": 0.06424335602899188, + "grad_norm": 2.6080808639526367, + "learning_rate": 3.205268935236004e-07, + "loss": 1.2144, + "mean_token_accuracy": 0.6438786387443542, + "num_tokens": 15031630.0, + "step": 585 + }, + { + "epoch": 0.06435317373160554, + "grad_norm": 3.0342447757720947, + "learning_rate": 3.2107574094401755e-07, + "loss": 1.1661, + "mean_token_accuracy": 0.6674572229385376, + "num_tokens": 15051551.0, + "step": 586 + }, + { + "epoch": 0.0644629914342192, + "grad_norm": 2.458705425262451, + "learning_rate": 3.2162458836443467e-07, + "loss": 1.0298, + "mean_token_accuracy": 0.7027183175086975, + "num_tokens": 15078407.0, + "step": 587 + }, + { + "epoch": 0.06457280913683286, + "grad_norm": 2.591729164123535, + "learning_rate": 3.221734357848518e-07, + "loss": 1.1909, + "mean_token_accuracy": 0.6644593477249146, + "num_tokens": 15106353.0, + "step": 588 + }, + { + "epoch": 0.06468262683944652, + "grad_norm": 2.6461715698242188, + "learning_rate": 3.227222832052689e-07, + "loss": 1.1265, + "mean_token_accuracy": 0.6753824949264526, + "num_tokens": 15132550.0, + "step": 589 + }, + { + "epoch": 0.06479244454206018, + "grad_norm": 2.2604427337646484, + "learning_rate": 3.2327113062568606e-07, + "loss": 1.0866, + "mean_token_accuracy": 0.6915555000305176, + "num_tokens": 15163161.0, + "step": 590 + }, + { + "epoch": 0.06490226224467384, + "grad_norm": 2.9815688133239746, + "learning_rate": 3.238199780461032e-07, + "loss": 1.0899, + "mean_token_accuracy": 0.6867023706436157, + "num_tokens": 15183991.0, + "step": 591 + }, + { + "epoch": 0.0650120799472875, + "grad_norm": 2.511213541030884, + "learning_rate": 3.243688254665203e-07, + "loss": 1.0853, + "mean_token_accuracy": 0.6879722476005554, + "num_tokens": 15212113.0, + "step": 592 + }, + { + "epoch": 0.06512189764990116, + "grad_norm": 3.12070369720459, + "learning_rate": 3.2491767288693746e-07, + "loss": 0.9124, + "mean_token_accuracy": 0.7227336764335632, + "num_tokens": 15230116.0, + "step": 593 + }, + { + "epoch": 0.06523171535251482, + "grad_norm": 2.5953805446624756, + "learning_rate": 3.254665203073546e-07, + "loss": 0.9906, + "mean_token_accuracy": 0.7046810388565063, + "num_tokens": 15255979.0, + "step": 594 + }, + { + "epoch": 0.06534153305512849, + "grad_norm": 2.7410287857055664, + "learning_rate": 3.2601536772777164e-07, + "loss": 1.1101, + "mean_token_accuracy": 0.6806069016456604, + "num_tokens": 15279314.0, + "step": 595 + }, + { + "epoch": 0.06545135075774215, + "grad_norm": 2.3294785022735596, + "learning_rate": 3.2656421514818876e-07, + "loss": 1.0686, + "mean_token_accuracy": 0.6911052465438843, + "num_tokens": 15311408.0, + "step": 596 + }, + { + "epoch": 0.06556116846035581, + "grad_norm": 2.3473148345947266, + "learning_rate": 3.271130625686059e-07, + "loss": 1.1358, + "mean_token_accuracy": 0.672370970249176, + "num_tokens": 15338715.0, + "step": 597 + }, + { + "epoch": 0.06567098616296947, + "grad_norm": 2.3466391563415527, + "learning_rate": 3.2766190998902304e-07, + "loss": 1.112, + "mean_token_accuracy": 0.677478551864624, + "num_tokens": 15369478.0, + "step": 598 + }, + { + "epoch": 0.06578080386558313, + "grad_norm": 2.747398614883423, + "learning_rate": 3.2821075740944015e-07, + "loss": 1.1374, + "mean_token_accuracy": 0.6748858690261841, + "num_tokens": 15391622.0, + "step": 599 + }, + { + "epoch": 0.06589062156819679, + "grad_norm": 2.767810344696045, + "learning_rate": 3.2875960482985727e-07, + "loss": 1.0955, + "mean_token_accuracy": 0.6800887584686279, + "num_tokens": 15413578.0, + "step": 600 + }, + { + "epoch": 0.06600043927081045, + "grad_norm": 2.5086464881896973, + "learning_rate": 3.2930845225027444e-07, + "loss": 1.2056, + "mean_token_accuracy": 0.6576444506645203, + "num_tokens": 15442264.0, + "step": 601 + }, + { + "epoch": 0.06611025697342411, + "grad_norm": 2.5486438274383545, + "learning_rate": 3.2985729967069155e-07, + "loss": 1.0365, + "mean_token_accuracy": 0.6971932649612427, + "num_tokens": 15467175.0, + "step": 602 + }, + { + "epoch": 0.06622007467603777, + "grad_norm": 2.6656017303466797, + "learning_rate": 3.3040614709110867e-07, + "loss": 1.1311, + "mean_token_accuracy": 0.6733261942863464, + "num_tokens": 15491173.0, + "step": 603 + }, + { + "epoch": 0.06632989237865144, + "grad_norm": 2.198904514312744, + "learning_rate": 3.3095499451152573e-07, + "loss": 1.1165, + "mean_token_accuracy": 0.6746196746826172, + "num_tokens": 15524033.0, + "step": 604 + }, + { + "epoch": 0.0664397100812651, + "grad_norm": 2.9076759815216064, + "learning_rate": 3.315038419319429e-07, + "loss": 1.0775, + "mean_token_accuracy": 0.6787432432174683, + "num_tokens": 15543227.0, + "step": 605 + }, + { + "epoch": 0.06654952778387876, + "grad_norm": 2.520766258239746, + "learning_rate": 3.3205268935236e-07, + "loss": 1.0488, + "mean_token_accuracy": 0.6970123052597046, + "num_tokens": 15567296.0, + "step": 606 + }, + { + "epoch": 0.06665934548649242, + "grad_norm": 2.3668792247772217, + "learning_rate": 3.3260153677277713e-07, + "loss": 1.1715, + "mean_token_accuracy": 0.6548634767532349, + "num_tokens": 15597413.0, + "step": 607 + }, + { + "epoch": 0.06676916318910608, + "grad_norm": 2.700356960296631, + "learning_rate": 3.331503841931943e-07, + "loss": 1.1789, + "mean_token_accuracy": 0.6599701642990112, + "num_tokens": 15619596.0, + "step": 608 + }, + { + "epoch": 0.06687898089171974, + "grad_norm": 2.5308539867401123, + "learning_rate": 3.336992316136114e-07, + "loss": 1.0541, + "mean_token_accuracy": 0.6875091791152954, + "num_tokens": 15643922.0, + "step": 609 + }, + { + "epoch": 0.0669887985943334, + "grad_norm": 2.254575490951538, + "learning_rate": 3.3424807903402853e-07, + "loss": 1.1337, + "mean_token_accuracy": 0.6653784513473511, + "num_tokens": 15675120.0, + "step": 610 + }, + { + "epoch": 0.06709861629694706, + "grad_norm": 2.5762534141540527, + "learning_rate": 3.3479692645444564e-07, + "loss": 1.062, + "mean_token_accuracy": 0.6944459676742554, + "num_tokens": 15701109.0, + "step": 611 + }, + { + "epoch": 0.06720843399956072, + "grad_norm": 2.6771764755249023, + "learning_rate": 3.353457738748628e-07, + "loss": 1.0475, + "mean_token_accuracy": 0.690322995185852, + "num_tokens": 15723911.0, + "step": 612 + }, + { + "epoch": 0.06731825170217438, + "grad_norm": 2.5904245376586914, + "learning_rate": 3.3589462129527993e-07, + "loss": 1.099, + "mean_token_accuracy": 0.676381528377533, + "num_tokens": 15748769.0, + "step": 613 + }, + { + "epoch": 0.06742806940478806, + "grad_norm": 2.316159963607788, + "learning_rate": 3.36443468715697e-07, + "loss": 1.0858, + "mean_token_accuracy": 0.6929275989532471, + "num_tokens": 15779008.0, + "step": 614 + }, + { + "epoch": 0.06753788710740172, + "grad_norm": 2.9597222805023193, + "learning_rate": 3.369923161361141e-07, + "loss": 1.123, + "mean_token_accuracy": 0.6830363869667053, + "num_tokens": 15798879.0, + "step": 615 + }, + { + "epoch": 0.06764770481001538, + "grad_norm": 2.5438473224639893, + "learning_rate": 3.3754116355653127e-07, + "loss": 1.0682, + "mean_token_accuracy": 0.6923069953918457, + "num_tokens": 15822787.0, + "step": 616 + }, + { + "epoch": 0.06775752251262904, + "grad_norm": 2.773244619369507, + "learning_rate": 3.380900109769484e-07, + "loss": 1.1471, + "mean_token_accuracy": 0.6672619581222534, + "num_tokens": 15845389.0, + "step": 617 + }, + { + "epoch": 0.0678673402152427, + "grad_norm": 2.580054998397827, + "learning_rate": 3.386388583973655e-07, + "loss": 1.0915, + "mean_token_accuracy": 0.6735794544219971, + "num_tokens": 15869365.0, + "step": 618 + }, + { + "epoch": 0.06797715791785636, + "grad_norm": 2.2239601612091064, + "learning_rate": 3.3918770581778267e-07, + "loss": 1.1222, + "mean_token_accuracy": 0.6665246486663818, + "num_tokens": 15901989.0, + "step": 619 + }, + { + "epoch": 0.06808697562047002, + "grad_norm": 2.60553240776062, + "learning_rate": 3.397365532381998e-07, + "loss": 1.0955, + "mean_token_accuracy": 0.6800912618637085, + "num_tokens": 15928756.0, + "step": 620 + }, + { + "epoch": 0.06819679332308368, + "grad_norm": 2.4210305213928223, + "learning_rate": 3.402854006586169e-07, + "loss": 1.1795, + "mean_token_accuracy": 0.6519486904144287, + "num_tokens": 15957351.0, + "step": 621 + }, + { + "epoch": 0.06830661102569734, + "grad_norm": 2.6291873455047607, + "learning_rate": 3.40834248079034e-07, + "loss": 1.1506, + "mean_token_accuracy": 0.6639854907989502, + "num_tokens": 15981452.0, + "step": 622 + }, + { + "epoch": 0.06841642872831101, + "grad_norm": 2.6699538230895996, + "learning_rate": 3.4138309549945113e-07, + "loss": 1.0789, + "mean_token_accuracy": 0.6833028197288513, + "num_tokens": 16004363.0, + "step": 623 + }, + { + "epoch": 0.06852624643092467, + "grad_norm": 2.7250149250030518, + "learning_rate": 3.4193194291986825e-07, + "loss": 1.111, + "mean_token_accuracy": 0.6853190064430237, + "num_tokens": 16026402.0, + "step": 624 + }, + { + "epoch": 0.06863606413353833, + "grad_norm": 2.7360901832580566, + "learning_rate": 3.4248079034028536e-07, + "loss": 1.0253, + "mean_token_accuracy": 0.6894401907920837, + "num_tokens": 16048414.0, + "step": 625 + }, + { + "epoch": 0.06874588183615199, + "grad_norm": 2.592611789703369, + "learning_rate": 3.430296377607025e-07, + "loss": 1.1109, + "mean_token_accuracy": 0.6791844964027405, + "num_tokens": 16073120.0, + "step": 626 + }, + { + "epoch": 0.06885569953876565, + "grad_norm": 2.5370290279388428, + "learning_rate": 3.4357848518111965e-07, + "loss": 1.0663, + "mean_token_accuracy": 0.6879292130470276, + "num_tokens": 16097935.0, + "step": 627 + }, + { + "epoch": 0.06896551724137931, + "grad_norm": 2.743555784225464, + "learning_rate": 3.4412733260153676e-07, + "loss": 1.1103, + "mean_token_accuracy": 0.6807119846343994, + "num_tokens": 16119646.0, + "step": 628 + }, + { + "epoch": 0.06907533494399297, + "grad_norm": 2.12931489944458, + "learning_rate": 3.446761800219539e-07, + "loss": 1.1008, + "mean_token_accuracy": 0.6883732676506042, + "num_tokens": 16150853.0, + "step": 629 + }, + { + "epoch": 0.06918515264660663, + "grad_norm": 3.0084280967712402, + "learning_rate": 3.4522502744237104e-07, + "loss": 1.0457, + "mean_token_accuracy": 0.6999592781066895, + "num_tokens": 16171600.0, + "step": 630 + }, + { + "epoch": 0.06929497034922029, + "grad_norm": 2.1144802570343018, + "learning_rate": 3.4577387486278816e-07, + "loss": 1.1181, + "mean_token_accuracy": 0.6747449040412903, + "num_tokens": 16206140.0, + "step": 631 + }, + { + "epoch": 0.06940478805183395, + "grad_norm": 2.35823392868042, + "learning_rate": 3.463227222832053e-07, + "loss": 1.0966, + "mean_token_accuracy": 0.6763581037521362, + "num_tokens": 16234217.0, + "step": 632 + }, + { + "epoch": 0.06951460575444762, + "grad_norm": 2.4686923027038574, + "learning_rate": 3.4687156970362234e-07, + "loss": 1.0391, + "mean_token_accuracy": 0.6924425959587097, + "num_tokens": 16259760.0, + "step": 633 + }, + { + "epoch": 0.06962442345706128, + "grad_norm": 2.938199281692505, + "learning_rate": 3.474204171240395e-07, + "loss": 1.2437, + "mean_token_accuracy": 0.6609053015708923, + "num_tokens": 16280540.0, + "step": 634 + }, + { + "epoch": 0.06973424115967494, + "grad_norm": 2.738225221633911, + "learning_rate": 3.479692645444566e-07, + "loss": 1.0548, + "mean_token_accuracy": 0.6867402791976929, + "num_tokens": 16304664.0, + "step": 635 + }, + { + "epoch": 0.0698440588622886, + "grad_norm": 2.521322011947632, + "learning_rate": 3.4851811196487374e-07, + "loss": 1.1851, + "mean_token_accuracy": 0.6598394513130188, + "num_tokens": 16330068.0, + "step": 636 + }, + { + "epoch": 0.06995387656490226, + "grad_norm": 2.7781729698181152, + "learning_rate": 3.4906695938529085e-07, + "loss": 1.175, + "mean_token_accuracy": 0.6595427393913269, + "num_tokens": 16354012.0, + "step": 637 + }, + { + "epoch": 0.07006369426751592, + "grad_norm": 2.2689108848571777, + "learning_rate": 3.49615806805708e-07, + "loss": 1.0589, + "mean_token_accuracy": 0.6918047666549683, + "num_tokens": 16381633.0, + "step": 638 + }, + { + "epoch": 0.07017351197012958, + "grad_norm": 2.841705083847046, + "learning_rate": 3.5016465422612513e-07, + "loss": 1.0497, + "mean_token_accuracy": 0.6925162672996521, + "num_tokens": 16402544.0, + "step": 639 + }, + { + "epoch": 0.07028332967274324, + "grad_norm": 3.1136622428894043, + "learning_rate": 3.5071350164654225e-07, + "loss": 1.0079, + "mean_token_accuracy": 0.7010324001312256, + "num_tokens": 16422607.0, + "step": 640 + }, + { + "epoch": 0.0703931473753569, + "grad_norm": 2.229156494140625, + "learning_rate": 3.512623490669594e-07, + "loss": 1.1876, + "mean_token_accuracy": 0.656669020652771, + "num_tokens": 16452032.0, + "step": 641 + }, + { + "epoch": 0.07050296507797058, + "grad_norm": 2.7056922912597656, + "learning_rate": 3.518111964873765e-07, + "loss": 1.055, + "mean_token_accuracy": 0.6837881803512573, + "num_tokens": 16472805.0, + "step": 642 + }, + { + "epoch": 0.07061278278058424, + "grad_norm": 2.9788930416107178, + "learning_rate": 3.523600439077936e-07, + "loss": 1.0887, + "mean_token_accuracy": 0.6822879314422607, + "num_tokens": 16492181.0, + "step": 643 + }, + { + "epoch": 0.0707226004831979, + "grad_norm": 2.4914419651031494, + "learning_rate": 3.529088913282107e-07, + "loss": 1.1295, + "mean_token_accuracy": 0.6697427034378052, + "num_tokens": 16517769.0, + "step": 644 + }, + { + "epoch": 0.07083241818581155, + "grad_norm": 2.534257173538208, + "learning_rate": 3.534577387486279e-07, + "loss": 1.1201, + "mean_token_accuracy": 0.6748893857002258, + "num_tokens": 16541630.0, + "step": 645 + }, + { + "epoch": 0.07094223588842521, + "grad_norm": 2.3248815536499023, + "learning_rate": 3.54006586169045e-07, + "loss": 1.1796, + "mean_token_accuracy": 0.6557697653770447, + "num_tokens": 16571294.0, + "step": 646 + }, + { + "epoch": 0.07105205359103887, + "grad_norm": 2.4070427417755127, + "learning_rate": 3.545554335894621e-07, + "loss": 1.2043, + "mean_token_accuracy": 0.6587406396865845, + "num_tokens": 16600965.0, + "step": 647 + }, + { + "epoch": 0.07116187129365253, + "grad_norm": 2.8779642581939697, + "learning_rate": 3.551042810098792e-07, + "loss": 1.1688, + "mean_token_accuracy": 0.6697878837585449, + "num_tokens": 16623873.0, + "step": 648 + }, + { + "epoch": 0.0712716889962662, + "grad_norm": 2.521301031112671, + "learning_rate": 3.556531284302964e-07, + "loss": 1.0196, + "mean_token_accuracy": 0.7015433311462402, + "num_tokens": 16647603.0, + "step": 649 + }, + { + "epoch": 0.07138150669887985, + "grad_norm": 2.480233907699585, + "learning_rate": 3.562019758507135e-07, + "loss": 1.0751, + "mean_token_accuracy": 0.6845015287399292, + "num_tokens": 16672803.0, + "step": 650 + }, + { + "epoch": 0.07149132440149351, + "grad_norm": 2.9731550216674805, + "learning_rate": 3.5675082327113057e-07, + "loss": 1.1227, + "mean_token_accuracy": 0.6771201491355896, + "num_tokens": 16693732.0, + "step": 651 + }, + { + "epoch": 0.07160114210410719, + "grad_norm": 2.396324396133423, + "learning_rate": 3.5729967069154774e-07, + "loss": 1.0578, + "mean_token_accuracy": 0.6866417527198792, + "num_tokens": 16720849.0, + "step": 652 + }, + { + "epoch": 0.07171095980672085, + "grad_norm": 2.411879062652588, + "learning_rate": 3.5784851811196485e-07, + "loss": 1.0958, + "mean_token_accuracy": 0.6791224479675293, + "num_tokens": 16749324.0, + "step": 653 + }, + { + "epoch": 0.07182077750933451, + "grad_norm": 2.6282739639282227, + "learning_rate": 3.5839736553238197e-07, + "loss": 1.1036, + "mean_token_accuracy": 0.6803787350654602, + "num_tokens": 16773589.0, + "step": 654 + }, + { + "epoch": 0.07193059521194817, + "grad_norm": 2.790273427963257, + "learning_rate": 3.589462129527991e-07, + "loss": 1.0275, + "mean_token_accuracy": 0.6931105852127075, + "num_tokens": 16795332.0, + "step": 655 + }, + { + "epoch": 0.07204041291456183, + "grad_norm": 2.49639630317688, + "learning_rate": 3.5949506037321625e-07, + "loss": 1.0901, + "mean_token_accuracy": 0.6828168630599976, + "num_tokens": 16821021.0, + "step": 656 + }, + { + "epoch": 0.07215023061717549, + "grad_norm": 2.633225202560425, + "learning_rate": 3.6004390779363337e-07, + "loss": 1.1867, + "mean_token_accuracy": 0.6545661687850952, + "num_tokens": 16844661.0, + "step": 657 + }, + { + "epoch": 0.07226004831978915, + "grad_norm": 2.2806155681610107, + "learning_rate": 3.605927552140505e-07, + "loss": 1.1689, + "mean_token_accuracy": 0.6639583110809326, + "num_tokens": 16874521.0, + "step": 658 + }, + { + "epoch": 0.0723698660224028, + "grad_norm": 2.695030689239502, + "learning_rate": 3.611416026344676e-07, + "loss": 1.1203, + "mean_token_accuracy": 0.6710394024848938, + "num_tokens": 16896950.0, + "step": 659 + }, + { + "epoch": 0.07247968372501647, + "grad_norm": 2.4714279174804688, + "learning_rate": 3.6169045005488477e-07, + "loss": 1.1109, + "mean_token_accuracy": 0.6797852516174316, + "num_tokens": 16921768.0, + "step": 660 + }, + { + "epoch": 0.07258950142763014, + "grad_norm": 2.308823823928833, + "learning_rate": 3.6223929747530183e-07, + "loss": 1.0976, + "mean_token_accuracy": 0.6856310367584229, + "num_tokens": 16951856.0, + "step": 661 + }, + { + "epoch": 0.0726993191302438, + "grad_norm": 2.364854335784912, + "learning_rate": 3.6278814489571894e-07, + "loss": 1.1186, + "mean_token_accuracy": 0.6730933785438538, + "num_tokens": 16979012.0, + "step": 662 + }, + { + "epoch": 0.07280913683285746, + "grad_norm": 2.4095916748046875, + "learning_rate": 3.6333699231613606e-07, + "loss": 1.1797, + "mean_token_accuracy": 0.6556129455566406, + "num_tokens": 17007746.0, + "step": 663 + }, + { + "epoch": 0.07291895453547112, + "grad_norm": 2.4271693229675293, + "learning_rate": 3.6388583973655323e-07, + "loss": 1.1424, + "mean_token_accuracy": 0.665442168712616, + "num_tokens": 17035418.0, + "step": 664 + }, + { + "epoch": 0.07302877223808478, + "grad_norm": 2.459888219833374, + "learning_rate": 3.6443468715697034e-07, + "loss": 1.0929, + "mean_token_accuracy": 0.6806082725524902, + "num_tokens": 17062318.0, + "step": 665 + }, + { + "epoch": 0.07313858994069844, + "grad_norm": 2.3677895069122314, + "learning_rate": 3.6498353457738746e-07, + "loss": 1.0924, + "mean_token_accuracy": 0.6790418028831482, + "num_tokens": 17089068.0, + "step": 666 + }, + { + "epoch": 0.0732484076433121, + "grad_norm": 2.5713417530059814, + "learning_rate": 3.655323819978046e-07, + "loss": 1.0635, + "mean_token_accuracy": 0.6855461597442627, + "num_tokens": 17113190.0, + "step": 667 + }, + { + "epoch": 0.07335822534592576, + "grad_norm": 2.6068947315216064, + "learning_rate": 3.6608122941822174e-07, + "loss": 1.1665, + "mean_token_accuracy": 0.6723207235336304, + "num_tokens": 17137271.0, + "step": 668 + }, + { + "epoch": 0.07346804304853942, + "grad_norm": 2.6865427494049072, + "learning_rate": 3.6663007683863886e-07, + "loss": 0.9953, + "mean_token_accuracy": 0.7036597728729248, + "num_tokens": 17157400.0, + "step": 669 + }, + { + "epoch": 0.07357786075115308, + "grad_norm": 2.363703966140747, + "learning_rate": 3.671789242590559e-07, + "loss": 1.1056, + "mean_token_accuracy": 0.6790207624435425, + "num_tokens": 17183189.0, + "step": 670 + }, + { + "epoch": 0.07368767845376675, + "grad_norm": 2.2481791973114014, + "learning_rate": 3.677277716794731e-07, + "loss": 1.1071, + "mean_token_accuracy": 0.6811281442642212, + "num_tokens": 17211690.0, + "step": 671 + }, + { + "epoch": 0.07379749615638041, + "grad_norm": 2.4844911098480225, + "learning_rate": 3.682766190998902e-07, + "loss": 1.1311, + "mean_token_accuracy": 0.6732844710350037, + "num_tokens": 17237813.0, + "step": 672 + }, + { + "epoch": 0.07390731385899407, + "grad_norm": 2.543410301208496, + "learning_rate": 3.688254665203073e-07, + "loss": 1.008, + "mean_token_accuracy": 0.7036850452423096, + "num_tokens": 17261551.0, + "step": 673 + }, + { + "epoch": 0.07401713156160773, + "grad_norm": 2.347018003463745, + "learning_rate": 3.6937431394072443e-07, + "loss": 1.1534, + "mean_token_accuracy": 0.6607087850570679, + "num_tokens": 17288485.0, + "step": 674 + }, + { + "epoch": 0.07412694926422139, + "grad_norm": 2.3106493949890137, + "learning_rate": 3.699231613611416e-07, + "loss": 1.1321, + "mean_token_accuracy": 0.6693618297576904, + "num_tokens": 17320177.0, + "step": 675 + }, + { + "epoch": 0.07423676696683505, + "grad_norm": 2.7093868255615234, + "learning_rate": 3.704720087815587e-07, + "loss": 1.0097, + "mean_token_accuracy": 0.7030143737792969, + "num_tokens": 17340881.0, + "step": 676 + }, + { + "epoch": 0.07434658466944871, + "grad_norm": 2.8233025074005127, + "learning_rate": 3.7102085620197583e-07, + "loss": 1.1473, + "mean_token_accuracy": 0.6597262620925903, + "num_tokens": 17362201.0, + "step": 677 + }, + { + "epoch": 0.07445640237206237, + "grad_norm": 3.157916307449341, + "learning_rate": 3.71569703622393e-07, + "loss": 1.0953, + "mean_token_accuracy": 0.6775996685028076, + "num_tokens": 17382355.0, + "step": 678 + }, + { + "epoch": 0.07456622007467603, + "grad_norm": 2.460973024368286, + "learning_rate": 3.721185510428101e-07, + "loss": 1.0778, + "mean_token_accuracy": 0.6771106719970703, + "num_tokens": 17407555.0, + "step": 679 + }, + { + "epoch": 0.0746760377772897, + "grad_norm": 2.7393903732299805, + "learning_rate": 3.726673984632272e-07, + "loss": 1.0392, + "mean_token_accuracy": 0.7039675712585449, + "num_tokens": 17430348.0, + "step": 680 + }, + { + "epoch": 0.07478585547990337, + "grad_norm": 2.5004971027374268, + "learning_rate": 3.732162458836443e-07, + "loss": 1.1176, + "mean_token_accuracy": 0.670315146446228, + "num_tokens": 17455543.0, + "step": 681 + }, + { + "epoch": 0.07489567318251703, + "grad_norm": 2.450714349746704, + "learning_rate": 3.7376509330406146e-07, + "loss": 1.1904, + "mean_token_accuracy": 0.652765154838562, + "num_tokens": 17483285.0, + "step": 682 + }, + { + "epoch": 0.07500549088513069, + "grad_norm": 2.5356736183166504, + "learning_rate": 3.743139407244786e-07, + "loss": 1.0579, + "mean_token_accuracy": 0.6877691745758057, + "num_tokens": 17508786.0, + "step": 683 + }, + { + "epoch": 0.07511530858774434, + "grad_norm": 2.442704439163208, + "learning_rate": 3.748627881448957e-07, + "loss": 1.106, + "mean_token_accuracy": 0.671086847782135, + "num_tokens": 17535425.0, + "step": 684 + }, + { + "epoch": 0.075225126290358, + "grad_norm": 2.6198694705963135, + "learning_rate": 3.754116355653128e-07, + "loss": 1.0635, + "mean_token_accuracy": 0.6842988133430481, + "num_tokens": 17559363.0, + "step": 685 + }, + { + "epoch": 0.07533494399297166, + "grad_norm": 2.4905192852020264, + "learning_rate": 3.7596048298573e-07, + "loss": 1.2334, + "mean_token_accuracy": 0.6463784575462341, + "num_tokens": 17586472.0, + "step": 686 + }, + { + "epoch": 0.07544476169558532, + "grad_norm": 2.6304030418395996, + "learning_rate": 3.765093304061471e-07, + "loss": 1.1415, + "mean_token_accuracy": 0.6730507612228394, + "num_tokens": 17610333.0, + "step": 687 + }, + { + "epoch": 0.07555457939819898, + "grad_norm": 2.3066818714141846, + "learning_rate": 3.770581778265642e-07, + "loss": 1.0271, + "mean_token_accuracy": 0.6979837417602539, + "num_tokens": 17636766.0, + "step": 688 + }, + { + "epoch": 0.07566439710081264, + "grad_norm": 2.495253324508667, + "learning_rate": 3.7760702524698137e-07, + "loss": 1.0338, + "mean_token_accuracy": 0.6921461820602417, + "num_tokens": 17660932.0, + "step": 689 + }, + { + "epoch": 0.07577421480342632, + "grad_norm": 2.204279899597168, + "learning_rate": 3.7815587266739844e-07, + "loss": 1.1034, + "mean_token_accuracy": 0.6831374168395996, + "num_tokens": 17689635.0, + "step": 690 + }, + { + "epoch": 0.07588403250603998, + "grad_norm": 2.4499096870422363, + "learning_rate": 3.7870472008781555e-07, + "loss": 1.1274, + "mean_token_accuracy": 0.6707121729850769, + "num_tokens": 17714292.0, + "step": 691 + }, + { + "epoch": 0.07599385020865364, + "grad_norm": 2.590224027633667, + "learning_rate": 3.7925356750823267e-07, + "loss": 1.0714, + "mean_token_accuracy": 0.6801975965499878, + "num_tokens": 17736652.0, + "step": 692 + }, + { + "epoch": 0.0761036679112673, + "grad_norm": 2.1003262996673584, + "learning_rate": 3.7980241492864983e-07, + "loss": 1.1965, + "mean_token_accuracy": 0.6534062027931213, + "num_tokens": 17773162.0, + "step": 693 + }, + { + "epoch": 0.07621348561388096, + "grad_norm": 2.3571877479553223, + "learning_rate": 3.8035126234906695e-07, + "loss": 1.0832, + "mean_token_accuracy": 0.6850154399871826, + "num_tokens": 17803131.0, + "step": 694 + }, + { + "epoch": 0.07632330331649462, + "grad_norm": 3.078366279602051, + "learning_rate": 3.8090010976948406e-07, + "loss": 0.984, + "mean_token_accuracy": 0.6991131901741028, + "num_tokens": 17820798.0, + "step": 695 + }, + { + "epoch": 0.07643312101910828, + "grad_norm": 2.4148013591766357, + "learning_rate": 3.814489571899012e-07, + "loss": 1.0961, + "mean_token_accuracy": 0.6755293011665344, + "num_tokens": 17847722.0, + "step": 696 + }, + { + "epoch": 0.07654293872172194, + "grad_norm": 2.568009614944458, + "learning_rate": 3.8199780461031835e-07, + "loss": 1.1194, + "mean_token_accuracy": 0.676038384437561, + "num_tokens": 17870979.0, + "step": 697 + }, + { + "epoch": 0.0766527564243356, + "grad_norm": 2.5824780464172363, + "learning_rate": 3.8254665203073546e-07, + "loss": 1.0999, + "mean_token_accuracy": 0.6751207113265991, + "num_tokens": 17894861.0, + "step": 698 + }, + { + "epoch": 0.07676257412694927, + "grad_norm": 2.300475597381592, + "learning_rate": 3.830954994511525e-07, + "loss": 1.143, + "mean_token_accuracy": 0.6720792651176453, + "num_tokens": 17925211.0, + "step": 699 + }, + { + "epoch": 0.07687239182956293, + "grad_norm": 2.773813486099243, + "learning_rate": 3.836443468715697e-07, + "loss": 1.1147, + "mean_token_accuracy": 0.6766594052314758, + "num_tokens": 17946300.0, + "step": 700 + }, + { + "epoch": 0.07698220953217659, + "grad_norm": 2.8061718940734863, + "learning_rate": 3.841931942919868e-07, + "loss": 0.9827, + "mean_token_accuracy": 0.7067165374755859, + "num_tokens": 17965341.0, + "step": 701 + }, + { + "epoch": 0.07709202723479025, + "grad_norm": 3.0227534770965576, + "learning_rate": 3.847420417124039e-07, + "loss": 1.0033, + "mean_token_accuracy": 0.6995372176170349, + "num_tokens": 17982282.0, + "step": 702 + }, + { + "epoch": 0.07720184493740391, + "grad_norm": 2.5946502685546875, + "learning_rate": 3.8529088913282104e-07, + "loss": 0.969, + "mean_token_accuracy": 0.7085658311843872, + "num_tokens": 18003689.0, + "step": 703 + }, + { + "epoch": 0.07731166264001757, + "grad_norm": 3.006908655166626, + "learning_rate": 3.858397365532382e-07, + "loss": 0.9547, + "mean_token_accuracy": 0.7040994763374329, + "num_tokens": 18021214.0, + "step": 704 + }, + { + "epoch": 0.07742148034263123, + "grad_norm": 2.4002676010131836, + "learning_rate": 3.863885839736553e-07, + "loss": 1.1567, + "mean_token_accuracy": 0.6568931341171265, + "num_tokens": 18048604.0, + "step": 705 + }, + { + "epoch": 0.07753129804524489, + "grad_norm": 2.5290112495422363, + "learning_rate": 3.8693743139407244e-07, + "loss": 1.0746, + "mean_token_accuracy": 0.6873378753662109, + "num_tokens": 18073163.0, + "step": 706 + }, + { + "epoch": 0.07764111574785855, + "grad_norm": 2.3045096397399902, + "learning_rate": 3.8748627881448955e-07, + "loss": 1.1333, + "mean_token_accuracy": 0.6682860851287842, + "num_tokens": 18103067.0, + "step": 707 + }, + { + "epoch": 0.07775093345047221, + "grad_norm": 2.5175583362579346, + "learning_rate": 3.880351262349067e-07, + "loss": 1.0252, + "mean_token_accuracy": 0.6955397725105286, + "num_tokens": 18126646.0, + "step": 708 + }, + { + "epoch": 0.07786075115308588, + "grad_norm": 2.5710015296936035, + "learning_rate": 3.885839736553238e-07, + "loss": 1.0216, + "mean_token_accuracy": 0.6924290657043457, + "num_tokens": 18148406.0, + "step": 709 + }, + { + "epoch": 0.07797056885569954, + "grad_norm": 2.511143922805786, + "learning_rate": 3.891328210757409e-07, + "loss": 1.0529, + "mean_token_accuracy": 0.6852750778198242, + "num_tokens": 18172707.0, + "step": 710 + }, + { + "epoch": 0.0780803865583132, + "grad_norm": 2.335981845855713, + "learning_rate": 3.8968166849615807e-07, + "loss": 1.1209, + "mean_token_accuracy": 0.674749493598938, + "num_tokens": 18201735.0, + "step": 711 + }, + { + "epoch": 0.07819020426092686, + "grad_norm": 2.188856363296509, + "learning_rate": 3.902305159165752e-07, + "loss": 1.1351, + "mean_token_accuracy": 0.6753536462783813, + "num_tokens": 18233001.0, + "step": 712 + }, + { + "epoch": 0.07830002196354052, + "grad_norm": 2.6300692558288574, + "learning_rate": 3.907793633369923e-07, + "loss": 1.0668, + "mean_token_accuracy": 0.6887871026992798, + "num_tokens": 18255447.0, + "step": 713 + }, + { + "epoch": 0.07840983966615418, + "grad_norm": 2.5239205360412598, + "learning_rate": 3.913282107574094e-07, + "loss": 1.0603, + "mean_token_accuracy": 0.6827736496925354, + "num_tokens": 18276472.0, + "step": 714 + }, + { + "epoch": 0.07851965736876784, + "grad_norm": 2.4609615802764893, + "learning_rate": 3.918770581778266e-07, + "loss": 1.0981, + "mean_token_accuracy": 0.6770231127738953, + "num_tokens": 18300773.0, + "step": 715 + }, + { + "epoch": 0.0786294750713815, + "grad_norm": 2.3882644176483154, + "learning_rate": 3.924259055982437e-07, + "loss": 1.0391, + "mean_token_accuracy": 0.7151299118995667, + "num_tokens": 18327741.0, + "step": 716 + }, + { + "epoch": 0.07873929277399516, + "grad_norm": 2.6140201091766357, + "learning_rate": 3.929747530186608e-07, + "loss": 1.0408, + "mean_token_accuracy": 0.690610408782959, + "num_tokens": 18350642.0, + "step": 717 + }, + { + "epoch": 0.07884911047660884, + "grad_norm": 2.414454460144043, + "learning_rate": 3.935236004390779e-07, + "loss": 1.0448, + "mean_token_accuracy": 0.6903553605079651, + "num_tokens": 18379140.0, + "step": 718 + }, + { + "epoch": 0.0789589281792225, + "grad_norm": 2.220470666885376, + "learning_rate": 3.9407244785949504e-07, + "loss": 1.1716, + "mean_token_accuracy": 0.6799123287200928, + "num_tokens": 18409923.0, + "step": 719 + }, + { + "epoch": 0.07906874588183616, + "grad_norm": 2.3281354904174805, + "learning_rate": 3.9462129527991216e-07, + "loss": 1.0503, + "mean_token_accuracy": 0.6854423880577087, + "num_tokens": 18438024.0, + "step": 720 + }, + { + "epoch": 0.07917856358444982, + "grad_norm": 2.3180594444274902, + "learning_rate": 3.9517014270032927e-07, + "loss": 1.072, + "mean_token_accuracy": 0.6826770901679993, + "num_tokens": 18465508.0, + "step": 721 + }, + { + "epoch": 0.07928838128706348, + "grad_norm": 2.6750969886779785, + "learning_rate": 3.957189901207464e-07, + "loss": 1.1277, + "mean_token_accuracy": 0.6677232384681702, + "num_tokens": 18487863.0, + "step": 722 + }, + { + "epoch": 0.07939819898967714, + "grad_norm": 2.1013779640197754, + "learning_rate": 3.9626783754116356e-07, + "loss": 1.0889, + "mean_token_accuracy": 0.6766408681869507, + "num_tokens": 18518851.0, + "step": 723 + }, + { + "epoch": 0.0795080166922908, + "grad_norm": 2.3961191177368164, + "learning_rate": 3.9681668496158067e-07, + "loss": 1.0644, + "mean_token_accuracy": 0.6829008460044861, + "num_tokens": 18543975.0, + "step": 724 + }, + { + "epoch": 0.07961783439490445, + "grad_norm": 2.280468225479126, + "learning_rate": 3.973655323819978e-07, + "loss": 1.057, + "mean_token_accuracy": 0.6997352838516235, + "num_tokens": 18572814.0, + "step": 725 + }, + { + "epoch": 0.07972765209751811, + "grad_norm": 2.404911994934082, + "learning_rate": 3.9791437980241495e-07, + "loss": 1.0038, + "mean_token_accuracy": 0.6979445219039917, + "num_tokens": 18596995.0, + "step": 726 + }, + { + "epoch": 0.07983746980013177, + "grad_norm": 2.150167465209961, + "learning_rate": 3.9846322722283207e-07, + "loss": 1.0522, + "mean_token_accuracy": 0.6902391910552979, + "num_tokens": 18629513.0, + "step": 727 + }, + { + "epoch": 0.07994728750274545, + "grad_norm": 2.3294522762298584, + "learning_rate": 3.9901207464324913e-07, + "loss": 1.03, + "mean_token_accuracy": 0.6990851163864136, + "num_tokens": 18657101.0, + "step": 728 + }, + { + "epoch": 0.08005710520535911, + "grad_norm": 2.623384475708008, + "learning_rate": 3.9956092206366625e-07, + "loss": 1.1191, + "mean_token_accuracy": 0.6724417805671692, + "num_tokens": 18682677.0, + "step": 729 + }, + { + "epoch": 0.08016692290797277, + "grad_norm": 3.0882627964019775, + "learning_rate": 4.001097694840834e-07, + "loss": 1.148, + "mean_token_accuracy": 0.6849895715713501, + "num_tokens": 18700959.0, + "step": 730 + }, + { + "epoch": 0.08027674061058643, + "grad_norm": 2.444460868835449, + "learning_rate": 4.0065861690450053e-07, + "loss": 1.0822, + "mean_token_accuracy": 0.6817034482955933, + "num_tokens": 18727623.0, + "step": 731 + }, + { + "epoch": 0.08038655831320009, + "grad_norm": 2.30405330657959, + "learning_rate": 4.0120746432491765e-07, + "loss": 1.2036, + "mean_token_accuracy": 0.6513879299163818, + "num_tokens": 18756130.0, + "step": 732 + }, + { + "epoch": 0.08049637601581375, + "grad_norm": 2.361161708831787, + "learning_rate": 4.0175631174533476e-07, + "loss": 1.056, + "mean_token_accuracy": 0.68129962682724, + "num_tokens": 18785463.0, + "step": 733 + }, + { + "epoch": 0.08060619371842741, + "grad_norm": 2.2664716243743896, + "learning_rate": 4.0230515916575193e-07, + "loss": 1.1416, + "mean_token_accuracy": 0.6724100112915039, + "num_tokens": 18811888.0, + "step": 734 + }, + { + "epoch": 0.08071601142104107, + "grad_norm": 2.1231443881988525, + "learning_rate": 4.0285400658616904e-07, + "loss": 1.1005, + "mean_token_accuracy": 0.6898476481437683, + "num_tokens": 18844743.0, + "step": 735 + }, + { + "epoch": 0.08082582912365473, + "grad_norm": 2.693868637084961, + "learning_rate": 4.0340285400658616e-07, + "loss": 1.0788, + "mean_token_accuracy": 0.6812940835952759, + "num_tokens": 18866184.0, + "step": 736 + }, + { + "epoch": 0.0809356468262684, + "grad_norm": 2.6185832023620605, + "learning_rate": 4.039517014270033e-07, + "loss": 0.9703, + "mean_token_accuracy": 0.7131417989730835, + "num_tokens": 18888175.0, + "step": 737 + }, + { + "epoch": 0.08104546452888206, + "grad_norm": 2.7321019172668457, + "learning_rate": 4.045005488474204e-07, + "loss": 1.1262, + "mean_token_accuracy": 0.6736863851547241, + "num_tokens": 18908724.0, + "step": 738 + }, + { + "epoch": 0.08115528223149572, + "grad_norm": 2.383855104446411, + "learning_rate": 4.050493962678375e-07, + "loss": 1.0767, + "mean_token_accuracy": 0.6786609292030334, + "num_tokens": 18933920.0, + "step": 739 + }, + { + "epoch": 0.08126509993410938, + "grad_norm": 2.361989736557007, + "learning_rate": 4.055982436882546e-07, + "loss": 0.9708, + "mean_token_accuracy": 0.7128248810768127, + "num_tokens": 18958089.0, + "step": 740 + }, + { + "epoch": 0.08137491763672304, + "grad_norm": 2.2279837131500244, + "learning_rate": 4.061470911086718e-07, + "loss": 1.0115, + "mean_token_accuracy": 0.6980305910110474, + "num_tokens": 18985237.0, + "step": 741 + }, + { + "epoch": 0.0814847353393367, + "grad_norm": 2.4572060108184814, + "learning_rate": 4.066959385290889e-07, + "loss": 1.1195, + "mean_token_accuracy": 0.6712609529495239, + "num_tokens": 19009884.0, + "step": 742 + }, + { + "epoch": 0.08159455304195036, + "grad_norm": 2.2394344806671143, + "learning_rate": 4.07244785949506e-07, + "loss": 1.1241, + "mean_token_accuracy": 0.6753557324409485, + "num_tokens": 19038885.0, + "step": 743 + }, + { + "epoch": 0.08170437074456402, + "grad_norm": 2.253659725189209, + "learning_rate": 4.0779363336992313e-07, + "loss": 1.0757, + "mean_token_accuracy": 0.6809805631637573, + "num_tokens": 19066392.0, + "step": 744 + }, + { + "epoch": 0.08181418844717768, + "grad_norm": 2.3039488792419434, + "learning_rate": 4.083424807903403e-07, + "loss": 1.1117, + "mean_token_accuracy": 0.6728465557098389, + "num_tokens": 19094917.0, + "step": 745 + }, + { + "epoch": 0.08192400614979134, + "grad_norm": 2.2565414905548096, + "learning_rate": 4.0889132821075737e-07, + "loss": 0.9822, + "mean_token_accuracy": 0.710456371307373, + "num_tokens": 19124560.0, + "step": 746 + }, + { + "epoch": 0.08203382385240501, + "grad_norm": 2.093130350112915, + "learning_rate": 4.094401756311745e-07, + "loss": 1.0124, + "mean_token_accuracy": 0.6972812414169312, + "num_tokens": 19154969.0, + "step": 747 + }, + { + "epoch": 0.08214364155501867, + "grad_norm": 2.247746706008911, + "learning_rate": 4.0998902305159165e-07, + "loss": 1.0754, + "mean_token_accuracy": 0.6832829117774963, + "num_tokens": 19183408.0, + "step": 748 + }, + { + "epoch": 0.08225345925763233, + "grad_norm": 2.3130719661712646, + "learning_rate": 4.1053787047200876e-07, + "loss": 1.2002, + "mean_token_accuracy": 0.6527788043022156, + "num_tokens": 19211989.0, + "step": 749 + }, + { + "epoch": 0.08236327696024599, + "grad_norm": 2.6903340816497803, + "learning_rate": 4.110867178924259e-07, + "loss": 1.1103, + "mean_token_accuracy": 0.6774872541427612, + "num_tokens": 19233535.0, + "step": 750 + }, + { + "epoch": 0.08247309466285965, + "grad_norm": 2.1399950981140137, + "learning_rate": 4.11635565312843e-07, + "loss": 1.0527, + "mean_token_accuracy": 0.6899723410606384, + "num_tokens": 19264886.0, + "step": 751 + }, + { + "epoch": 0.08258291236547331, + "grad_norm": 2.0817952156066895, + "learning_rate": 4.1218441273326016e-07, + "loss": 1.1472, + "mean_token_accuracy": 0.6673771142959595, + "num_tokens": 19295968.0, + "step": 752 + }, + { + "epoch": 0.08269273006808697, + "grad_norm": 2.386847972869873, + "learning_rate": 4.127332601536773e-07, + "loss": 0.99, + "mean_token_accuracy": 0.7059304714202881, + "num_tokens": 19320203.0, + "step": 753 + }, + { + "epoch": 0.08280254777070063, + "grad_norm": 2.4439172744750977, + "learning_rate": 4.132821075740944e-07, + "loss": 1.1267, + "mean_token_accuracy": 0.6782941818237305, + "num_tokens": 19347614.0, + "step": 754 + }, + { + "epoch": 0.08291236547331429, + "grad_norm": 2.577627420425415, + "learning_rate": 4.138309549945115e-07, + "loss": 1.0849, + "mean_token_accuracy": 0.6837658882141113, + "num_tokens": 19371516.0, + "step": 755 + }, + { + "epoch": 0.08302218317592797, + "grad_norm": 2.186253547668457, + "learning_rate": 4.143798024149286e-07, + "loss": 1.1251, + "mean_token_accuracy": 0.6748446822166443, + "num_tokens": 19399963.0, + "step": 756 + }, + { + "epoch": 0.08313200087854163, + "grad_norm": 2.635699510574341, + "learning_rate": 4.1492864983534574e-07, + "loss": 1.1084, + "mean_token_accuracy": 0.6720764636993408, + "num_tokens": 19421044.0, + "step": 757 + }, + { + "epoch": 0.08324181858115529, + "grad_norm": 2.6160526275634766, + "learning_rate": 4.1547749725576285e-07, + "loss": 1.0282, + "mean_token_accuracy": 0.6950212717056274, + "num_tokens": 19441620.0, + "step": 758 + }, + { + "epoch": 0.08335163628376895, + "grad_norm": 2.2539446353912354, + "learning_rate": 4.1602634467618e-07, + "loss": 1.0301, + "mean_token_accuracy": 0.6876161098480225, + "num_tokens": 19467889.0, + "step": 759 + }, + { + "epoch": 0.0834614539863826, + "grad_norm": 2.4545555114746094, + "learning_rate": 4.1657519209659714e-07, + "loss": 0.9694, + "mean_token_accuracy": 0.707760751247406, + "num_tokens": 19491134.0, + "step": 760 + }, + { + "epoch": 0.08357127168899627, + "grad_norm": 2.4721732139587402, + "learning_rate": 4.1712403951701425e-07, + "loss": 1.0083, + "mean_token_accuracy": 0.7005251049995422, + "num_tokens": 19514752.0, + "step": 761 + }, + { + "epoch": 0.08368108939160993, + "grad_norm": 2.191091775894165, + "learning_rate": 4.1767288693743137e-07, + "loss": 1.0712, + "mean_token_accuracy": 0.6903305053710938, + "num_tokens": 19545986.0, + "step": 762 + }, + { + "epoch": 0.08379090709422359, + "grad_norm": 2.5211338996887207, + "learning_rate": 4.1822173435784854e-07, + "loss": 1.0913, + "mean_token_accuracy": 0.6867043972015381, + "num_tokens": 19571866.0, + "step": 763 + }, + { + "epoch": 0.08390072479683724, + "grad_norm": 1.911689043045044, + "learning_rate": 4.1877058177826565e-07, + "loss": 1.0824, + "mean_token_accuracy": 0.677861213684082, + "num_tokens": 19607475.0, + "step": 764 + }, + { + "epoch": 0.0840105424994509, + "grad_norm": 2.081098794937134, + "learning_rate": 4.193194291986827e-07, + "loss": 1.1063, + "mean_token_accuracy": 0.6765960454940796, + "num_tokens": 19640298.0, + "step": 765 + }, + { + "epoch": 0.08412036020206458, + "grad_norm": 2.5141682624816895, + "learning_rate": 4.1986827661909983e-07, + "loss": 1.0512, + "mean_token_accuracy": 0.6851490139961243, + "num_tokens": 19665538.0, + "step": 766 + }, + { + "epoch": 0.08423017790467824, + "grad_norm": 2.292631149291992, + "learning_rate": 4.20417124039517e-07, + "loss": 1.1322, + "mean_token_accuracy": 0.6678103804588318, + "num_tokens": 19693023.0, + "step": 767 + }, + { + "epoch": 0.0843399956072919, + "grad_norm": 2.209502935409546, + "learning_rate": 4.209659714599341e-07, + "loss": 1.1292, + "mean_token_accuracy": 0.6682542562484741, + "num_tokens": 19722692.0, + "step": 768 + }, + { + "epoch": 0.08444981330990556, + "grad_norm": 2.2340147495269775, + "learning_rate": 4.2151481888035123e-07, + "loss": 1.0925, + "mean_token_accuracy": 0.6800203919410706, + "num_tokens": 19751027.0, + "step": 769 + }, + { + "epoch": 0.08455963101251922, + "grad_norm": 2.49928617477417, + "learning_rate": 4.2206366630076834e-07, + "loss": 1.204, + "mean_token_accuracy": 0.6547297239303589, + "num_tokens": 19777198.0, + "step": 770 + }, + { + "epoch": 0.08466944871513288, + "grad_norm": 2.2668728828430176, + "learning_rate": 4.226125137211855e-07, + "loss": 1.1642, + "mean_token_accuracy": 0.6700109243392944, + "num_tokens": 19806149.0, + "step": 771 + }, + { + "epoch": 0.08477926641774654, + "grad_norm": 3.035637855529785, + "learning_rate": 4.231613611416026e-07, + "loss": 1.0196, + "mean_token_accuracy": 0.6937167644500732, + "num_tokens": 19824356.0, + "step": 772 + }, + { + "epoch": 0.0848890841203602, + "grad_norm": 2.125206708908081, + "learning_rate": 4.2371020856201974e-07, + "loss": 1.0649, + "mean_token_accuracy": 0.6862284541130066, + "num_tokens": 19857209.0, + "step": 773 + }, + { + "epoch": 0.08499890182297386, + "grad_norm": 2.506124258041382, + "learning_rate": 4.242590559824369e-07, + "loss": 1.1972, + "mean_token_accuracy": 0.6565247774124146, + "num_tokens": 19884020.0, + "step": 774 + }, + { + "epoch": 0.08510871952558753, + "grad_norm": 2.2737226486206055, + "learning_rate": 4.2480790340285397e-07, + "loss": 1.1957, + "mean_token_accuracy": 0.6588541269302368, + "num_tokens": 19912147.0, + "step": 775 + }, + { + "epoch": 0.08521853722820119, + "grad_norm": 2.480297565460205, + "learning_rate": 4.253567508232711e-07, + "loss": 1.1249, + "mean_token_accuracy": 0.673790454864502, + "num_tokens": 19935856.0, + "step": 776 + }, + { + "epoch": 0.08532835493081485, + "grad_norm": 2.1523990631103516, + "learning_rate": 4.259055982436882e-07, + "loss": 1.1717, + "mean_token_accuracy": 0.6632651090621948, + "num_tokens": 19965850.0, + "step": 777 + }, + { + "epoch": 0.08543817263342851, + "grad_norm": 2.3573405742645264, + "learning_rate": 4.2645444566410537e-07, + "loss": 1.1094, + "mean_token_accuracy": 0.6818739175796509, + "num_tokens": 19991082.0, + "step": 778 + }, + { + "epoch": 0.08554799033604217, + "grad_norm": 2.185602903366089, + "learning_rate": 4.270032930845225e-07, + "loss": 1.116, + "mean_token_accuracy": 0.6742486357688904, + "num_tokens": 20021049.0, + "step": 779 + }, + { + "epoch": 0.08565780803865583, + "grad_norm": 2.25234317779541, + "learning_rate": 4.275521405049396e-07, + "loss": 1.1364, + "mean_token_accuracy": 0.6684938073158264, + "num_tokens": 20049437.0, + "step": 780 + }, + { + "epoch": 0.08576762574126949, + "grad_norm": 2.538602352142334, + "learning_rate": 4.281009879253567e-07, + "loss": 1.0141, + "mean_token_accuracy": 0.6979867815971375, + "num_tokens": 20073176.0, + "step": 781 + }, + { + "epoch": 0.08587744344388315, + "grad_norm": 2.3528060913085938, + "learning_rate": 4.286498353457739e-07, + "loss": 1.1092, + "mean_token_accuracy": 0.6734472513198853, + "num_tokens": 20099076.0, + "step": 782 + }, + { + "epoch": 0.08598726114649681, + "grad_norm": 2.3408329486846924, + "learning_rate": 4.29198682766191e-07, + "loss": 1.0627, + "mean_token_accuracy": 0.6842948198318481, + "num_tokens": 20125901.0, + "step": 783 + }, + { + "epoch": 0.08609707884911047, + "grad_norm": 2.224896192550659, + "learning_rate": 4.2974753018660806e-07, + "loss": 1.1452, + "mean_token_accuracy": 0.6751638650894165, + "num_tokens": 20154008.0, + "step": 784 + }, + { + "epoch": 0.08620689655172414, + "grad_norm": 2.7532548904418945, + "learning_rate": 4.3029637760702523e-07, + "loss": 1.1503, + "mean_token_accuracy": 0.6674484014511108, + "num_tokens": 20176932.0, + "step": 785 + }, + { + "epoch": 0.0863167142543378, + "grad_norm": 2.7000975608825684, + "learning_rate": 4.3084522502744235e-07, + "loss": 1.0388, + "mean_token_accuracy": 0.6917681694030762, + "num_tokens": 20198954.0, + "step": 786 + }, + { + "epoch": 0.08642653195695146, + "grad_norm": 2.5763604640960693, + "learning_rate": 4.3139407244785946e-07, + "loss": 1.0851, + "mean_token_accuracy": 0.6896263360977173, + "num_tokens": 20221828.0, + "step": 787 + }, + { + "epoch": 0.08653634965956512, + "grad_norm": 2.4961109161376953, + "learning_rate": 4.319429198682766e-07, + "loss": 1.1291, + "mean_token_accuracy": 0.6638756394386292, + "num_tokens": 20246948.0, + "step": 788 + }, + { + "epoch": 0.08664616736217878, + "grad_norm": 2.579271078109741, + "learning_rate": 4.3249176728869374e-07, + "loss": 1.1216, + "mean_token_accuracy": 0.6723210215568542, + "num_tokens": 20272143.0, + "step": 789 + }, + { + "epoch": 0.08675598506479244, + "grad_norm": 2.3372111320495605, + "learning_rate": 4.3304061470911086e-07, + "loss": 1.0542, + "mean_token_accuracy": 0.6881837844848633, + "num_tokens": 20298197.0, + "step": 790 + }, + { + "epoch": 0.0868658027674061, + "grad_norm": 2.3553643226623535, + "learning_rate": 4.33589462129528e-07, + "loss": 1.0675, + "mean_token_accuracy": 0.6812276244163513, + "num_tokens": 20324013.0, + "step": 791 + }, + { + "epoch": 0.08697562047001976, + "grad_norm": 2.38334059715271, + "learning_rate": 4.341383095499451e-07, + "loss": 1.0837, + "mean_token_accuracy": 0.6803816556930542, + "num_tokens": 20349091.0, + "step": 792 + }, + { + "epoch": 0.08708543817263342, + "grad_norm": 2.5894100666046143, + "learning_rate": 4.3468715697036226e-07, + "loss": 1.0532, + "mean_token_accuracy": 0.6921955943107605, + "num_tokens": 20372053.0, + "step": 793 + }, + { + "epoch": 0.0871952558752471, + "grad_norm": 2.186720848083496, + "learning_rate": 4.352360043907793e-07, + "loss": 1.038, + "mean_token_accuracy": 0.6900081634521484, + "num_tokens": 20401552.0, + "step": 794 + }, + { + "epoch": 0.08730507357786076, + "grad_norm": 2.090082883834839, + "learning_rate": 4.3578485181119644e-07, + "loss": 1.1053, + "mean_token_accuracy": 0.6680967807769775, + "num_tokens": 20434081.0, + "step": 795 + }, + { + "epoch": 0.08741489128047442, + "grad_norm": 2.4356772899627686, + "learning_rate": 4.363336992316136e-07, + "loss": 1.0622, + "mean_token_accuracy": 0.6848263144493103, + "num_tokens": 20457274.0, + "step": 796 + }, + { + "epoch": 0.08752470898308808, + "grad_norm": 2.1144983768463135, + "learning_rate": 4.368825466520307e-07, + "loss": 1.1226, + "mean_token_accuracy": 0.6715147495269775, + "num_tokens": 20488879.0, + "step": 797 + }, + { + "epoch": 0.08763452668570174, + "grad_norm": 2.78913950920105, + "learning_rate": 4.3743139407244783e-07, + "loss": 1.0828, + "mean_token_accuracy": 0.6802106499671936, + "num_tokens": 20511214.0, + "step": 798 + }, + { + "epoch": 0.0877443443883154, + "grad_norm": 2.460181951522827, + "learning_rate": 4.3798024149286495e-07, + "loss": 1.1419, + "mean_token_accuracy": 0.6662626266479492, + "num_tokens": 20535878.0, + "step": 799 + }, + { + "epoch": 0.08785416209092906, + "grad_norm": 2.6616175174713135, + "learning_rate": 4.385290889132821e-07, + "loss": 1.0648, + "mean_token_accuracy": 0.6862415075302124, + "num_tokens": 20558305.0, + "step": 800 + }, + { + "epoch": 0.08796397979354272, + "grad_norm": 2.0486419200897217, + "learning_rate": 4.3907793633369923e-07, + "loss": 1.0103, + "mean_token_accuracy": 0.6949938535690308, + "num_tokens": 20589039.0, + "step": 801 + }, + { + "epoch": 0.08807379749615638, + "grad_norm": 2.4138269424438477, + "learning_rate": 4.3962678375411635e-07, + "loss": 1.0258, + "mean_token_accuracy": 0.6927492618560791, + "num_tokens": 20614233.0, + "step": 802 + }, + { + "epoch": 0.08818361519877004, + "grad_norm": 2.4055469036102295, + "learning_rate": 4.401756311745334e-07, + "loss": 1.0936, + "mean_token_accuracy": 0.6699685454368591, + "num_tokens": 20639168.0, + "step": 803 + }, + { + "epoch": 0.08829343290138371, + "grad_norm": 2.395672559738159, + "learning_rate": 4.407244785949506e-07, + "loss": 1.0185, + "mean_token_accuracy": 0.6914088726043701, + "num_tokens": 20664464.0, + "step": 804 + }, + { + "epoch": 0.08840325060399737, + "grad_norm": 2.7142388820648193, + "learning_rate": 4.412733260153677e-07, + "loss": 1.1337, + "mean_token_accuracy": 0.6704899072647095, + "num_tokens": 20688195.0, + "step": 805 + }, + { + "epoch": 0.08851306830661103, + "grad_norm": 2.674867868423462, + "learning_rate": 4.418221734357848e-07, + "loss": 1.1308, + "mean_token_accuracy": 0.6652841567993164, + "num_tokens": 20710648.0, + "step": 806 + }, + { + "epoch": 0.08862288600922469, + "grad_norm": 2.431236505508423, + "learning_rate": 4.42371020856202e-07, + "loss": 1.039, + "mean_token_accuracy": 0.6802240610122681, + "num_tokens": 20735110.0, + "step": 807 + }, + { + "epoch": 0.08873270371183835, + "grad_norm": 2.302696943283081, + "learning_rate": 4.429198682766191e-07, + "loss": 1.0852, + "mean_token_accuracy": 0.6739278435707092, + "num_tokens": 20762412.0, + "step": 808 + }, + { + "epoch": 0.08884252141445201, + "grad_norm": 2.4406259059906006, + "learning_rate": 4.434687156970362e-07, + "loss": 0.9405, + "mean_token_accuracy": 0.718915581703186, + "num_tokens": 20788793.0, + "step": 809 + }, + { + "epoch": 0.08895233911706567, + "grad_norm": 2.3508241176605225, + "learning_rate": 4.440175631174533e-07, + "loss": 1.1316, + "mean_token_accuracy": 0.6689288020133972, + "num_tokens": 20815556.0, + "step": 810 + }, + { + "epoch": 0.08906215681967933, + "grad_norm": 2.4759912490844727, + "learning_rate": 4.445664105378705e-07, + "loss": 1.1161, + "mean_token_accuracy": 0.6680377125740051, + "num_tokens": 20841742.0, + "step": 811 + }, + { + "epoch": 0.08917197452229299, + "grad_norm": 2.4200587272644043, + "learning_rate": 4.451152579582876e-07, + "loss": 1.0225, + "mean_token_accuracy": 0.7044981718063354, + "num_tokens": 20868364.0, + "step": 812 + }, + { + "epoch": 0.08928179222490666, + "grad_norm": 2.4644806385040283, + "learning_rate": 4.4566410537870467e-07, + "loss": 1.007, + "mean_token_accuracy": 0.7019332647323608, + "num_tokens": 20891443.0, + "step": 813 + }, + { + "epoch": 0.08939160992752032, + "grad_norm": 2.6152846813201904, + "learning_rate": 4.462129527991218e-07, + "loss": 1.145, + "mean_token_accuracy": 0.6728000044822693, + "num_tokens": 20913114.0, + "step": 814 + }, + { + "epoch": 0.08950142763013398, + "grad_norm": 2.4557173252105713, + "learning_rate": 4.4676180021953895e-07, + "loss": 1.1464, + "mean_token_accuracy": 0.6634953618049622, + "num_tokens": 20939343.0, + "step": 815 + }, + { + "epoch": 0.08961124533274764, + "grad_norm": 2.6156365871429443, + "learning_rate": 4.4731064763995607e-07, + "loss": 0.9611, + "mean_token_accuracy": 0.7074037194252014, + "num_tokens": 20960760.0, + "step": 816 + }, + { + "epoch": 0.0897210630353613, + "grad_norm": 2.6649982929229736, + "learning_rate": 4.478594950603732e-07, + "loss": 1.005, + "mean_token_accuracy": 0.7005557417869568, + "num_tokens": 20980833.0, + "step": 817 + }, + { + "epoch": 0.08983088073797496, + "grad_norm": 2.4575982093811035, + "learning_rate": 4.4840834248079035e-07, + "loss": 1.0389, + "mean_token_accuracy": 0.6961930990219116, + "num_tokens": 21005826.0, + "step": 818 + }, + { + "epoch": 0.08994069844058862, + "grad_norm": 2.616219997406006, + "learning_rate": 4.4895718990120747e-07, + "loss": 1.0759, + "mean_token_accuracy": 0.685961127281189, + "num_tokens": 21027666.0, + "step": 819 + }, + { + "epoch": 0.09005051614320228, + "grad_norm": 2.1327788829803467, + "learning_rate": 4.495060373216246e-07, + "loss": 1.1086, + "mean_token_accuracy": 0.6739267110824585, + "num_tokens": 21057121.0, + "step": 820 + }, + { + "epoch": 0.09016033384581594, + "grad_norm": 2.462629795074463, + "learning_rate": 4.500548847420417e-07, + "loss": 1.1515, + "mean_token_accuracy": 0.6670717000961304, + "num_tokens": 21082774.0, + "step": 821 + }, + { + "epoch": 0.0902701515484296, + "grad_norm": 2.4696030616760254, + "learning_rate": 4.5060373216245886e-07, + "loss": 1.0054, + "mean_token_accuracy": 0.6997690200805664, + "num_tokens": 21104606.0, + "step": 822 + }, + { + "epoch": 0.09037996925104327, + "grad_norm": 2.21535587310791, + "learning_rate": 4.5115257958287593e-07, + "loss": 1.1278, + "mean_token_accuracy": 0.6726930141448975, + "num_tokens": 21133670.0, + "step": 823 + }, + { + "epoch": 0.09048978695365693, + "grad_norm": 2.4708166122436523, + "learning_rate": 4.5170142700329304e-07, + "loss": 1.0805, + "mean_token_accuracy": 0.681106448173523, + "num_tokens": 21158898.0, + "step": 824 + }, + { + "epoch": 0.0905996046562706, + "grad_norm": 2.3445470333099365, + "learning_rate": 4.5225027442371016e-07, + "loss": 1.1603, + "mean_token_accuracy": 0.6644158363342285, + "num_tokens": 21185944.0, + "step": 825 + }, + { + "epoch": 0.09070942235888425, + "grad_norm": 2.2488858699798584, + "learning_rate": 4.527991218441273e-07, + "loss": 1.0882, + "mean_token_accuracy": 0.6779791116714478, + "num_tokens": 21213957.0, + "step": 826 + }, + { + "epoch": 0.09081924006149791, + "grad_norm": 2.3038270473480225, + "learning_rate": 4.5334796926454444e-07, + "loss": 1.1075, + "mean_token_accuracy": 0.6733283400535583, + "num_tokens": 21239944.0, + "step": 827 + }, + { + "epoch": 0.09092905776411157, + "grad_norm": 2.1938140392303467, + "learning_rate": 4.5389681668496156e-07, + "loss": 1.1888, + "mean_token_accuracy": 0.6621208190917969, + "num_tokens": 21269702.0, + "step": 828 + }, + { + "epoch": 0.09103887546672523, + "grad_norm": 2.587217092514038, + "learning_rate": 4.5444566410537867e-07, + "loss": 1.0823, + "mean_token_accuracy": 0.6787174940109253, + "num_tokens": 21292329.0, + "step": 829 + }, + { + "epoch": 0.09114869316933889, + "grad_norm": 2.4544029235839844, + "learning_rate": 4.5499451152579584e-07, + "loss": 1.0841, + "mean_token_accuracy": 0.6867804527282715, + "num_tokens": 21315085.0, + "step": 830 + }, + { + "epoch": 0.09125851087195255, + "grad_norm": 2.3399910926818848, + "learning_rate": 4.5554335894621295e-07, + "loss": 1.0735, + "mean_token_accuracy": 0.6774392127990723, + "num_tokens": 21341153.0, + "step": 831 + }, + { + "epoch": 0.09136832857456623, + "grad_norm": 2.16534161567688, + "learning_rate": 4.5609220636663e-07, + "loss": 1.1378, + "mean_token_accuracy": 0.667436957359314, + "num_tokens": 21371548.0, + "step": 832 + }, + { + "epoch": 0.09147814627717989, + "grad_norm": 2.7076659202575684, + "learning_rate": 4.566410537870472e-07, + "loss": 1.0544, + "mean_token_accuracy": 0.6851515173912048, + "num_tokens": 21392144.0, + "step": 833 + }, + { + "epoch": 0.09158796397979355, + "grad_norm": 2.5023720264434814, + "learning_rate": 4.571899012074643e-07, + "loss": 1.0199, + "mean_token_accuracy": 0.6939181089401245, + "num_tokens": 21413151.0, + "step": 834 + }, + { + "epoch": 0.0916977816824072, + "grad_norm": 2.322545051574707, + "learning_rate": 4.577387486278814e-07, + "loss": 1.05, + "mean_token_accuracy": 0.6875224113464355, + "num_tokens": 21439843.0, + "step": 835 + }, + { + "epoch": 0.09180759938502087, + "grad_norm": 2.3126509189605713, + "learning_rate": 4.5828759604829853e-07, + "loss": 1.0119, + "mean_token_accuracy": 0.7028838992118835, + "num_tokens": 21464010.0, + "step": 836 + }, + { + "epoch": 0.09191741708763453, + "grad_norm": 2.6650424003601074, + "learning_rate": 4.588364434687157e-07, + "loss": 1.0412, + "mean_token_accuracy": 0.6913748979568481, + "num_tokens": 21484591.0, + "step": 837 + }, + { + "epoch": 0.09202723479024819, + "grad_norm": 2.628610849380493, + "learning_rate": 4.593852908891328e-07, + "loss": 1.1256, + "mean_token_accuracy": 0.6669789552688599, + "num_tokens": 21505856.0, + "step": 838 + }, + { + "epoch": 0.09213705249286185, + "grad_norm": 2.4864413738250732, + "learning_rate": 4.5993413830954993e-07, + "loss": 1.0336, + "mean_token_accuracy": 0.698610246181488, + "num_tokens": 21529421.0, + "step": 839 + }, + { + "epoch": 0.0922468701954755, + "grad_norm": 2.2125346660614014, + "learning_rate": 4.6048298572996704e-07, + "loss": 1.1187, + "mean_token_accuracy": 0.6749681234359741, + "num_tokens": 21559874.0, + "step": 840 + }, + { + "epoch": 0.09235668789808917, + "grad_norm": 2.838808059692383, + "learning_rate": 4.6103183315038416e-07, + "loss": 0.9819, + "mean_token_accuracy": 0.7125028371810913, + "num_tokens": 21578740.0, + "step": 841 + }, + { + "epoch": 0.09246650560070284, + "grad_norm": 2.3047022819519043, + "learning_rate": 4.615806805708013e-07, + "loss": 1.1028, + "mean_token_accuracy": 0.6823028326034546, + "num_tokens": 21605115.0, + "step": 842 + }, + { + "epoch": 0.0925763233033165, + "grad_norm": 2.0676968097686768, + "learning_rate": 4.621295279912184e-07, + "loss": 1.1719, + "mean_token_accuracy": 0.6617246866226196, + "num_tokens": 21636921.0, + "step": 843 + }, + { + "epoch": 0.09268614100593016, + "grad_norm": 2.3887763023376465, + "learning_rate": 4.6267837541163556e-07, + "loss": 1.0357, + "mean_token_accuracy": 0.6952193379402161, + "num_tokens": 21662944.0, + "step": 844 + }, + { + "epoch": 0.09279595870854382, + "grad_norm": 2.615318775177002, + "learning_rate": 4.632272228320527e-07, + "loss": 1.0266, + "mean_token_accuracy": 0.697870135307312, + "num_tokens": 21684700.0, + "step": 845 + }, + { + "epoch": 0.09290577641115748, + "grad_norm": 2.2009801864624023, + "learning_rate": 4.637760702524698e-07, + "loss": 1.0369, + "mean_token_accuracy": 0.6979143619537354, + "num_tokens": 21713402.0, + "step": 846 + }, + { + "epoch": 0.09301559411377114, + "grad_norm": 2.6169424057006836, + "learning_rate": 4.643249176728869e-07, + "loss": 1.0849, + "mean_token_accuracy": 0.6809564828872681, + "num_tokens": 21734886.0, + "step": 847 + }, + { + "epoch": 0.0931254118163848, + "grad_norm": 2.521016836166382, + "learning_rate": 4.6487376509330407e-07, + "loss": 1.1534, + "mean_token_accuracy": 0.6597023606300354, + "num_tokens": 21759855.0, + "step": 848 + }, + { + "epoch": 0.09323522951899846, + "grad_norm": 2.5385096073150635, + "learning_rate": 4.654226125137212e-07, + "loss": 1.0487, + "mean_token_accuracy": 0.6892527937889099, + "num_tokens": 21781168.0, + "step": 849 + }, + { + "epoch": 0.09334504722161212, + "grad_norm": 2.837566614151001, + "learning_rate": 4.659714599341383e-07, + "loss": 1.0176, + "mean_token_accuracy": 0.688496470451355, + "num_tokens": 21799088.0, + "step": 850 + }, + { + "epoch": 0.09345486492422579, + "grad_norm": 2.2699739933013916, + "learning_rate": 4.6652030735455537e-07, + "loss": 1.068, + "mean_token_accuracy": 0.685784101486206, + "num_tokens": 21826593.0, + "step": 851 + }, + { + "epoch": 0.09356468262683945, + "grad_norm": 2.2192845344543457, + "learning_rate": 4.6706915477497253e-07, + "loss": 1.0207, + "mean_token_accuracy": 0.6948926448822021, + "num_tokens": 21857593.0, + "step": 852 + }, + { + "epoch": 0.09367450032945311, + "grad_norm": 2.3447446823120117, + "learning_rate": 4.6761800219538965e-07, + "loss": 1.0962, + "mean_token_accuracy": 0.6742383241653442, + "num_tokens": 21883537.0, + "step": 853 + }, + { + "epoch": 0.09378431803206677, + "grad_norm": 2.4922542572021484, + "learning_rate": 4.6816684961580676e-07, + "loss": 1.078, + "mean_token_accuracy": 0.6852073669433594, + "num_tokens": 21906076.0, + "step": 854 + }, + { + "epoch": 0.09389413573468043, + "grad_norm": 2.460796356201172, + "learning_rate": 4.6871569703622393e-07, + "loss": 1.0266, + "mean_token_accuracy": 0.6935828924179077, + "num_tokens": 21928890.0, + "step": 855 + }, + { + "epoch": 0.09400395343729409, + "grad_norm": 2.305561065673828, + "learning_rate": 4.6926454445664105e-07, + "loss": 1.0824, + "mean_token_accuracy": 0.6790280342102051, + "num_tokens": 21953991.0, + "step": 856 + }, + { + "epoch": 0.09411377113990775, + "grad_norm": 2.3964717388153076, + "learning_rate": 4.6981339187705816e-07, + "loss": 1.0539, + "mean_token_accuracy": 0.6907626390457153, + "num_tokens": 21980225.0, + "step": 857 + }, + { + "epoch": 0.09422358884252141, + "grad_norm": 2.2843081951141357, + "learning_rate": 4.703622392974753e-07, + "loss": 1.0183, + "mean_token_accuracy": 0.695124626159668, + "num_tokens": 22006279.0, + "step": 858 + }, + { + "epoch": 0.09433340654513507, + "grad_norm": 2.318634033203125, + "learning_rate": 4.7091108671789245e-07, + "loss": 1.0475, + "mean_token_accuracy": 0.688563346862793, + "num_tokens": 22030633.0, + "step": 859 + }, + { + "epoch": 0.09444322424774873, + "grad_norm": 2.20401668548584, + "learning_rate": 4.714599341383095e-07, + "loss": 1.0999, + "mean_token_accuracy": 0.6756629943847656, + "num_tokens": 22058085.0, + "step": 860 + }, + { + "epoch": 0.0945530419503624, + "grad_norm": 2.3952014446258545, + "learning_rate": 4.720087815587266e-07, + "loss": 1.0913, + "mean_token_accuracy": 0.6746030449867249, + "num_tokens": 22081441.0, + "step": 861 + }, + { + "epoch": 0.09466285965297606, + "grad_norm": 2.2520360946655273, + "learning_rate": 4.7255762897914374e-07, + "loss": 1.0601, + "mean_token_accuracy": 0.6896054148674011, + "num_tokens": 22108402.0, + "step": 862 + }, + { + "epoch": 0.09477267735558972, + "grad_norm": 2.477674722671509, + "learning_rate": 4.731064763995609e-07, + "loss": 1.0935, + "mean_token_accuracy": 0.6830275058746338, + "num_tokens": 22132153.0, + "step": 863 + }, + { + "epoch": 0.09488249505820338, + "grad_norm": 2.1790106296539307, + "learning_rate": 4.73655323819978e-07, + "loss": 1.0267, + "mean_token_accuracy": 0.6953940391540527, + "num_tokens": 22160328.0, + "step": 864 + }, + { + "epoch": 0.09499231276081704, + "grad_norm": 2.0931918621063232, + "learning_rate": 4.7420417124039514e-07, + "loss": 1.122, + "mean_token_accuracy": 0.6659953594207764, + "num_tokens": 22197532.0, + "step": 865 + }, + { + "epoch": 0.0951021304634307, + "grad_norm": 2.574585199356079, + "learning_rate": 4.747530186608123e-07, + "loss": 0.99, + "mean_token_accuracy": 0.7016693353652954, + "num_tokens": 22219202.0, + "step": 866 + }, + { + "epoch": 0.09521194816604436, + "grad_norm": 2.3949825763702393, + "learning_rate": 4.753018660812294e-07, + "loss": 0.9867, + "mean_token_accuracy": 0.6997283697128296, + "num_tokens": 22241303.0, + "step": 867 + }, + { + "epoch": 0.09532176586865802, + "grad_norm": 2.4899866580963135, + "learning_rate": 4.7585071350164654e-07, + "loss": 0.9763, + "mean_token_accuracy": 0.7035152912139893, + "num_tokens": 22266678.0, + "step": 868 + }, + { + "epoch": 0.09543158357127168, + "grad_norm": 2.1296567916870117, + "learning_rate": 4.7639956092206365e-07, + "loss": 1.0177, + "mean_token_accuracy": 0.7041008472442627, + "num_tokens": 22295627.0, + "step": 869 + }, + { + "epoch": 0.09554140127388536, + "grad_norm": 2.580632209777832, + "learning_rate": 4.769484083424808e-07, + "loss": 1.1158, + "mean_token_accuracy": 0.6740199327468872, + "num_tokens": 22318538.0, + "step": 870 + }, + { + "epoch": 0.09565121897649902, + "grad_norm": 2.8554673194885254, + "learning_rate": 4.774972557628979e-07, + "loss": 1.0693, + "mean_token_accuracy": 0.6787522435188293, + "num_tokens": 22336118.0, + "step": 871 + }, + { + "epoch": 0.09576103667911268, + "grad_norm": 2.2348716259002686, + "learning_rate": 4.78046103183315e-07, + "loss": 1.1519, + "mean_token_accuracy": 0.6669076681137085, + "num_tokens": 22364302.0, + "step": 872 + }, + { + "epoch": 0.09587085438172634, + "grad_norm": 2.7598021030426025, + "learning_rate": 4.785949506037321e-07, + "loss": 1.0424, + "mean_token_accuracy": 0.6902213096618652, + "num_tokens": 22384460.0, + "step": 873 + }, + { + "epoch": 0.09598067208434, + "grad_norm": 2.3438470363616943, + "learning_rate": 4.791437980241493e-07, + "loss": 1.0658, + "mean_token_accuracy": 0.6786554455757141, + "num_tokens": 22409412.0, + "step": 874 + }, + { + "epoch": 0.09609048978695366, + "grad_norm": 2.6587846279144287, + "learning_rate": 4.796926454445663e-07, + "loss": 1.0345, + "mean_token_accuracy": 0.6891480684280396, + "num_tokens": 22430327.0, + "step": 875 + }, + { + "epoch": 0.09620030748956732, + "grad_norm": 2.502068042755127, + "learning_rate": 4.802414928649835e-07, + "loss": 1.0564, + "mean_token_accuracy": 0.6912598013877869, + "num_tokens": 22452557.0, + "step": 876 + }, + { + "epoch": 0.09631012519218098, + "grad_norm": 2.2379469871520996, + "learning_rate": 4.807903402854007e-07, + "loss": 1.0186, + "mean_token_accuracy": 0.6912474632263184, + "num_tokens": 22479404.0, + "step": 877 + }, + { + "epoch": 0.09641994289479464, + "grad_norm": 2.4523046016693115, + "learning_rate": 4.813391877058177e-07, + "loss": 1.0135, + "mean_token_accuracy": 0.7015022039413452, + "num_tokens": 22502514.0, + "step": 878 + }, + { + "epoch": 0.0965297605974083, + "grad_norm": 2.428154468536377, + "learning_rate": 4.818880351262349e-07, + "loss": 1.0552, + "mean_token_accuracy": 0.680496335029602, + "num_tokens": 22527514.0, + "step": 879 + }, + { + "epoch": 0.09663957830002197, + "grad_norm": 2.512559652328491, + "learning_rate": 4.82436882546652e-07, + "loss": 1.0861, + "mean_token_accuracy": 0.6808322668075562, + "num_tokens": 22550580.0, + "step": 880 + }, + { + "epoch": 0.09674939600263563, + "grad_norm": 2.6182491779327393, + "learning_rate": 4.829857299670691e-07, + "loss": 1.0715, + "mean_token_accuracy": 0.6807610988616943, + "num_tokens": 22570904.0, + "step": 881 + }, + { + "epoch": 0.09685921370524929, + "grad_norm": 2.285933494567871, + "learning_rate": 4.835345773874863e-07, + "loss": 1.0376, + "mean_token_accuracy": 0.691886305809021, + "num_tokens": 22599342.0, + "step": 882 + }, + { + "epoch": 0.09696903140786295, + "grad_norm": 2.2140612602233887, + "learning_rate": 4.840834248079034e-07, + "loss": 1.0453, + "mean_token_accuracy": 0.6909102201461792, + "num_tokens": 22628881.0, + "step": 883 + }, + { + "epoch": 0.09707884911047661, + "grad_norm": 2.580864906311035, + "learning_rate": 4.846322722283204e-07, + "loss": 1.0045, + "mean_token_accuracy": 0.7043010592460632, + "num_tokens": 22651554.0, + "step": 884 + }, + { + "epoch": 0.09718866681309027, + "grad_norm": 2.3575408458709717, + "learning_rate": 4.851811196487376e-07, + "loss": 0.9964, + "mean_token_accuracy": 0.7007981538772583, + "num_tokens": 22677728.0, + "step": 885 + }, + { + "epoch": 0.09729848451570393, + "grad_norm": 2.031500816345215, + "learning_rate": 4.857299670691548e-07, + "loss": 1.0337, + "mean_token_accuracy": 0.6969214081764221, + "num_tokens": 22710122.0, + "step": 886 + }, + { + "epoch": 0.09740830221831759, + "grad_norm": 2.829509735107422, + "learning_rate": 4.862788144895718e-07, + "loss": 1.0934, + "mean_token_accuracy": 0.6807571649551392, + "num_tokens": 22730685.0, + "step": 887 + }, + { + "epoch": 0.09751811992093125, + "grad_norm": 2.2463135719299316, + "learning_rate": 4.86827661909989e-07, + "loss": 1.0312, + "mean_token_accuracy": 0.6915590763092041, + "num_tokens": 22758710.0, + "step": 888 + }, + { + "epoch": 0.09762793762354492, + "grad_norm": 2.3807482719421387, + "learning_rate": 4.873765093304062e-07, + "loss": 0.9654, + "mean_token_accuracy": 0.715919554233551, + "num_tokens": 22784865.0, + "step": 889 + }, + { + "epoch": 0.09773775532615858, + "grad_norm": 2.4913415908813477, + "learning_rate": 4.879253567508232e-07, + "loss": 0.9765, + "mean_token_accuracy": 0.701215386390686, + "num_tokens": 22808380.0, + "step": 890 + }, + { + "epoch": 0.09784757302877224, + "grad_norm": 2.324230194091797, + "learning_rate": 4.884742041712404e-07, + "loss": 1.0873, + "mean_token_accuracy": 0.6886245012283325, + "num_tokens": 22834543.0, + "step": 891 + }, + { + "epoch": 0.0979573907313859, + "grad_norm": 2.245887279510498, + "learning_rate": 4.890230515916576e-07, + "loss": 1.0929, + "mean_token_accuracy": 0.6781946420669556, + "num_tokens": 22861373.0, + "step": 892 + }, + { + "epoch": 0.09806720843399956, + "grad_norm": 2.233581781387329, + "learning_rate": 4.895718990120746e-07, + "loss": 1.2262, + "mean_token_accuracy": 0.6465514302253723, + "num_tokens": 22892227.0, + "step": 893 + }, + { + "epoch": 0.09817702613661322, + "grad_norm": 2.3874504566192627, + "learning_rate": 4.901207464324917e-07, + "loss": 1.0225, + "mean_token_accuracy": 0.6965537071228027, + "num_tokens": 22916758.0, + "step": 894 + }, + { + "epoch": 0.09828684383922688, + "grad_norm": 2.769242763519287, + "learning_rate": 4.906695938529089e-07, + "loss": 1.0782, + "mean_token_accuracy": 0.6803735494613647, + "num_tokens": 22937628.0, + "step": 895 + }, + { + "epoch": 0.09839666154184054, + "grad_norm": 2.458735942840576, + "learning_rate": 4.91218441273326e-07, + "loss": 0.9266, + "mean_token_accuracy": 0.722328245639801, + "num_tokens": 22960642.0, + "step": 896 + }, + { + "epoch": 0.0985064792444542, + "grad_norm": 2.2739009857177734, + "learning_rate": 4.917672886937431e-07, + "loss": 1.0467, + "mean_token_accuracy": 0.6874101161956787, + "num_tokens": 22987184.0, + "step": 897 + }, + { + "epoch": 0.09861629694706786, + "grad_norm": 2.3825035095214844, + "learning_rate": 4.923161361141603e-07, + "loss": 1.086, + "mean_token_accuracy": 0.6743584871292114, + "num_tokens": 23011571.0, + "step": 898 + }, + { + "epoch": 0.09872611464968153, + "grad_norm": 2.3865115642547607, + "learning_rate": 4.928649835345773e-07, + "loss": 1.1502, + "mean_token_accuracy": 0.6916840076446533, + "num_tokens": 23037557.0, + "step": 899 + }, + { + "epoch": 0.0988359323522952, + "grad_norm": 2.3639822006225586, + "learning_rate": 4.934138309549945e-07, + "loss": 1.1347, + "mean_token_accuracy": 0.6656528115272522, + "num_tokens": 23066028.0, + "step": 900 + }, + { + "epoch": 0.09894575005490885, + "grad_norm": 2.604346752166748, + "learning_rate": 4.939626783754117e-07, + "loss": 1.1054, + "mean_token_accuracy": 0.6733850836753845, + "num_tokens": 23088540.0, + "step": 901 + }, + { + "epoch": 0.09905556775752251, + "grad_norm": 2.1075899600982666, + "learning_rate": 4.945115257958287e-07, + "loss": 1.1078, + "mean_token_accuracy": 0.6802260875701904, + "num_tokens": 23120670.0, + "step": 902 + }, + { + "epoch": 0.09916538546013617, + "grad_norm": 2.339346408843994, + "learning_rate": 4.950603732162459e-07, + "loss": 0.9556, + "mean_token_accuracy": 0.7142061591148376, + "num_tokens": 23145834.0, + "step": 903 + }, + { + "epoch": 0.09927520316274983, + "grad_norm": 2.5230889320373535, + "learning_rate": 4.95609220636663e-07, + "loss": 1.059, + "mean_token_accuracy": 0.6926707029342651, + "num_tokens": 23168891.0, + "step": 904 + }, + { + "epoch": 0.0993850208653635, + "grad_norm": 2.155362367630005, + "learning_rate": 4.961580680570801e-07, + "loss": 1.1225, + "mean_token_accuracy": 0.6733286380767822, + "num_tokens": 23199133.0, + "step": 905 + }, + { + "epoch": 0.09949483856797715, + "grad_norm": 2.3403918743133545, + "learning_rate": 4.967069154774972e-07, + "loss": 1.0947, + "mean_token_accuracy": 0.6765083074569702, + "num_tokens": 23226516.0, + "step": 906 + }, + { + "epoch": 0.09960465627059081, + "grad_norm": 2.292768716812134, + "learning_rate": 4.972557628979143e-07, + "loss": 1.106, + "mean_token_accuracy": 0.675833523273468, + "num_tokens": 23252476.0, + "step": 907 + }, + { + "epoch": 0.09971447397320449, + "grad_norm": 2.222883939743042, + "learning_rate": 4.978046103183315e-07, + "loss": 1.1054, + "mean_token_accuracy": 0.6761691570281982, + "num_tokens": 23280114.0, + "step": 908 + }, + { + "epoch": 0.09982429167581815, + "grad_norm": 2.1525518894195557, + "learning_rate": 4.983534577387486e-07, + "loss": 1.0796, + "mean_token_accuracy": 0.6853739619255066, + "num_tokens": 23309108.0, + "step": 909 + }, + { + "epoch": 0.0999341093784318, + "grad_norm": 2.0570242404937744, + "learning_rate": 4.989023051591657e-07, + "loss": 1.1129, + "mean_token_accuracy": 0.6812817454338074, + "num_tokens": 23339017.0, + "step": 910 + }, + { + "epoch": 0.10004392708104547, + "grad_norm": 2.475557327270508, + "learning_rate": 4.994511525795829e-07, + "loss": 1.0653, + "mean_token_accuracy": 0.6804721355438232, + "num_tokens": 23362701.0, + "step": 911 + }, + { + "epoch": 0.10015374478365913, + "grad_norm": 2.057441473007202, + "learning_rate": 5e-07, + "loss": 1.1686, + "mean_token_accuracy": 0.6625397205352783, + "num_tokens": 23395784.0, + "step": 912 + }, + { + "epoch": 0.10026356248627279, + "grad_norm": 2.6569442749023438, + "learning_rate": 5.005488474204171e-07, + "loss": 1.1409, + "mean_token_accuracy": 0.6693272590637207, + "num_tokens": 23417120.0, + "step": 913 + }, + { + "epoch": 0.10037338018888645, + "grad_norm": 2.527179479598999, + "learning_rate": 5.010976948408342e-07, + "loss": 1.0115, + "mean_token_accuracy": 0.696742832660675, + "num_tokens": 23440120.0, + "step": 914 + }, + { + "epoch": 0.1004831978915001, + "grad_norm": 2.078267812728882, + "learning_rate": 5.016465422612514e-07, + "loss": 1.1128, + "mean_token_accuracy": 0.6749831438064575, + "num_tokens": 23471856.0, + "step": 915 + }, + { + "epoch": 0.10059301559411377, + "grad_norm": 2.3626644611358643, + "learning_rate": 5.021953896816685e-07, + "loss": 1.0251, + "mean_token_accuracy": 0.6915255784988403, + "num_tokens": 23496577.0, + "step": 916 + }, + { + "epoch": 0.10070283329672743, + "grad_norm": 2.4543299674987793, + "learning_rate": 5.027442371020856e-07, + "loss": 1.0848, + "mean_token_accuracy": 0.6831414699554443, + "num_tokens": 23521503.0, + "step": 917 + }, + { + "epoch": 0.1008126509993411, + "grad_norm": 2.384726047515869, + "learning_rate": 5.032930845225028e-07, + "loss": 0.9905, + "mean_token_accuracy": 0.6999648809432983, + "num_tokens": 23547783.0, + "step": 918 + }, + { + "epoch": 0.10092246870195476, + "grad_norm": 2.1878864765167236, + "learning_rate": 5.038419319429198e-07, + "loss": 1.0679, + "mean_token_accuracy": 0.6878508925437927, + "num_tokens": 23576208.0, + "step": 919 + }, + { + "epoch": 0.10103228640456842, + "grad_norm": 2.175597906112671, + "learning_rate": 5.04390779363337e-07, + "loss": 1.0132, + "mean_token_accuracy": 0.6964726448059082, + "num_tokens": 23604352.0, + "step": 920 + }, + { + "epoch": 0.10114210410718208, + "grad_norm": 2.4206933975219727, + "learning_rate": 5.049396267837542e-07, + "loss": 0.9998, + "mean_token_accuracy": 0.7024960517883301, + "num_tokens": 23629114.0, + "step": 921 + }, + { + "epoch": 0.10125192180979574, + "grad_norm": 2.336848497390747, + "learning_rate": 5.054884742041711e-07, + "loss": 1.0768, + "mean_token_accuracy": 0.6780378818511963, + "num_tokens": 23653482.0, + "step": 922 + }, + { + "epoch": 0.1013617395124094, + "grad_norm": 2.3878395557403564, + "learning_rate": 5.060373216245883e-07, + "loss": 1.0541, + "mean_token_accuracy": 0.6867701411247253, + "num_tokens": 23678163.0, + "step": 923 + }, + { + "epoch": 0.10147155721502306, + "grad_norm": 2.0520358085632324, + "learning_rate": 5.065861690450055e-07, + "loss": 1.1441, + "mean_token_accuracy": 0.6652159690856934, + "num_tokens": 23709190.0, + "step": 924 + }, + { + "epoch": 0.10158137491763672, + "grad_norm": 2.346191167831421, + "learning_rate": 5.071350164654225e-07, + "loss": 1.1462, + "mean_token_accuracy": 0.6628869771957397, + "num_tokens": 23737902.0, + "step": 925 + }, + { + "epoch": 0.10169119262025038, + "grad_norm": 2.956902027130127, + "learning_rate": 5.076838638858397e-07, + "loss": 0.9964, + "mean_token_accuracy": 0.7055399417877197, + "num_tokens": 23756060.0, + "step": 926 + }, + { + "epoch": 0.10180101032286405, + "grad_norm": 2.494645833969116, + "learning_rate": 5.082327113062569e-07, + "loss": 1.0221, + "mean_token_accuracy": 0.6921915411949158, + "num_tokens": 23777382.0, + "step": 927 + }, + { + "epoch": 0.10191082802547771, + "grad_norm": 2.1587600708007812, + "learning_rate": 5.087815587266739e-07, + "loss": 1.0394, + "mean_token_accuracy": 0.6888429522514343, + "num_tokens": 23807032.0, + "step": 928 + }, + { + "epoch": 0.10202064572809137, + "grad_norm": 2.5123229026794434, + "learning_rate": 5.093304061470911e-07, + "loss": 1.0725, + "mean_token_accuracy": 0.6785244941711426, + "num_tokens": 23831572.0, + "step": 929 + }, + { + "epoch": 0.10213046343070503, + "grad_norm": 2.5363550186157227, + "learning_rate": 5.098792535675082e-07, + "loss": 1.065, + "mean_token_accuracy": 0.6901184916496277, + "num_tokens": 23855463.0, + "step": 930 + }, + { + "epoch": 0.10224028113331869, + "grad_norm": 2.2865333557128906, + "learning_rate": 5.104281009879253e-07, + "loss": 1.1117, + "mean_token_accuracy": 0.6739166975021362, + "num_tokens": 23885677.0, + "step": 931 + }, + { + "epoch": 0.10235009883593235, + "grad_norm": 2.5664889812469482, + "learning_rate": 5.109769484083425e-07, + "loss": 1.1294, + "mean_token_accuracy": 0.6695092916488647, + "num_tokens": 23910012.0, + "step": 932 + }, + { + "epoch": 0.10245991653854601, + "grad_norm": 2.430516481399536, + "learning_rate": 5.115257958287596e-07, + "loss": 1.0381, + "mean_token_accuracy": 0.6923041939735413, + "num_tokens": 23934853.0, + "step": 933 + }, + { + "epoch": 0.10256973424115967, + "grad_norm": 2.40915846824646, + "learning_rate": 5.120746432491767e-07, + "loss": 1.0765, + "mean_token_accuracy": 0.6880675554275513, + "num_tokens": 23959025.0, + "step": 934 + }, + { + "epoch": 0.10267955194377333, + "grad_norm": 2.3205862045288086, + "learning_rate": 5.126234906695939e-07, + "loss": 1.129, + "mean_token_accuracy": 0.6743189096450806, + "num_tokens": 23985706.0, + "step": 935 + }, + { + "epoch": 0.10278936964638699, + "grad_norm": 2.4630141258239746, + "learning_rate": 5.13172338090011e-07, + "loss": 1.0823, + "mean_token_accuracy": 0.6800808906555176, + "num_tokens": 24011184.0, + "step": 936 + }, + { + "epoch": 0.10289918734900066, + "grad_norm": 2.377927780151367, + "learning_rate": 5.137211855104281e-07, + "loss": 1.0889, + "mean_token_accuracy": 0.6788791418075562, + "num_tokens": 24038477.0, + "step": 937 + }, + { + "epoch": 0.10300900505161432, + "grad_norm": 2.6071372032165527, + "learning_rate": 5.142700329308453e-07, + "loss": 1.0154, + "mean_token_accuracy": 0.6956440210342407, + "num_tokens": 24059675.0, + "step": 938 + }, + { + "epoch": 0.10311882275422798, + "grad_norm": 2.915151357650757, + "learning_rate": 5.148188803512624e-07, + "loss": 0.9256, + "mean_token_accuracy": 0.7206993699073792, + "num_tokens": 24078583.0, + "step": 939 + }, + { + "epoch": 0.10322864045684164, + "grad_norm": 2.2979910373687744, + "learning_rate": 5.153677277716795e-07, + "loss": 1.0357, + "mean_token_accuracy": 0.6942461133003235, + "num_tokens": 24104104.0, + "step": 940 + }, + { + "epoch": 0.1033384581594553, + "grad_norm": 2.7150979042053223, + "learning_rate": 5.159165751920965e-07, + "loss": 0.9804, + "mean_token_accuracy": 0.7039165496826172, + "num_tokens": 24126454.0, + "step": 941 + }, + { + "epoch": 0.10344827586206896, + "grad_norm": 2.1790578365325928, + "learning_rate": 5.164654226125136e-07, + "loss": 1.1566, + "mean_token_accuracy": 0.658362090587616, + "num_tokens": 24157670.0, + "step": 942 + }, + { + "epoch": 0.10355809356468262, + "grad_norm": 2.262561082839966, + "learning_rate": 5.170142700329308e-07, + "loss": 1.115, + "mean_token_accuracy": 0.6714126467704773, + "num_tokens": 24185905.0, + "step": 943 + }, + { + "epoch": 0.10366791126729628, + "grad_norm": 2.4207234382629395, + "learning_rate": 5.175631174533479e-07, + "loss": 1.0825, + "mean_token_accuracy": 0.6761706471443176, + "num_tokens": 24210267.0, + "step": 944 + }, + { + "epoch": 0.10377772896990994, + "grad_norm": 2.135063409805298, + "learning_rate": 5.18111964873765e-07, + "loss": 1.021, + "mean_token_accuracy": 0.6956731081008911, + "num_tokens": 24239667.0, + "step": 945 + }, + { + "epoch": 0.10388754667252362, + "grad_norm": 2.332367181777954, + "learning_rate": 5.186608122941822e-07, + "loss": 0.9707, + "mean_token_accuracy": 0.703732967376709, + "num_tokens": 24266708.0, + "step": 946 + }, + { + "epoch": 0.10399736437513728, + "grad_norm": 2.408708095550537, + "learning_rate": 5.192096597145993e-07, + "loss": 1.0739, + "mean_token_accuracy": 0.6828545331954956, + "num_tokens": 24291185.0, + "step": 947 + }, + { + "epoch": 0.10410718207775094, + "grad_norm": 2.5779595375061035, + "learning_rate": 5.197585071350164e-07, + "loss": 0.997, + "mean_token_accuracy": 0.7012358903884888, + "num_tokens": 24312721.0, + "step": 948 + }, + { + "epoch": 0.1042169997803646, + "grad_norm": 2.3889265060424805, + "learning_rate": 5.203073545554336e-07, + "loss": 1.1267, + "mean_token_accuracy": 0.6604620814323425, + "num_tokens": 24336747.0, + "step": 949 + }, + { + "epoch": 0.10432681748297826, + "grad_norm": 2.159794569015503, + "learning_rate": 5.208562019758507e-07, + "loss": 1.0972, + "mean_token_accuracy": 0.6725834608078003, + "num_tokens": 24366256.0, + "step": 950 + }, + { + "epoch": 0.10443663518559192, + "grad_norm": 2.301758050918579, + "learning_rate": 5.214050493962678e-07, + "loss": 1.1134, + "mean_token_accuracy": 0.6737526059150696, + "num_tokens": 24392636.0, + "step": 951 + }, + { + "epoch": 0.10454645288820558, + "grad_norm": 2.789222478866577, + "learning_rate": 5.219538968166849e-07, + "loss": 1.065, + "mean_token_accuracy": 0.687549889087677, + "num_tokens": 24412949.0, + "step": 952 + }, + { + "epoch": 0.10465627059081924, + "grad_norm": 2.223177194595337, + "learning_rate": 5.225027442371021e-07, + "loss": 1.0711, + "mean_token_accuracy": 0.6921195983886719, + "num_tokens": 24439584.0, + "step": 953 + }, + { + "epoch": 0.1047660882934329, + "grad_norm": 2.4272704124450684, + "learning_rate": 5.230515916575192e-07, + "loss": 0.8866, + "mean_token_accuracy": 0.7204927802085876, + "num_tokens": 24461757.0, + "step": 954 + }, + { + "epoch": 0.10487590599604656, + "grad_norm": 2.707770586013794, + "learning_rate": 5.236004390779363e-07, + "loss": 1.0016, + "mean_token_accuracy": 0.6976996660232544, + "num_tokens": 24481800.0, + "step": 955 + }, + { + "epoch": 0.10498572369866023, + "grad_norm": 2.3550350666046143, + "learning_rate": 5.241492864983535e-07, + "loss": 1.0676, + "mean_token_accuracy": 0.6879793405532837, + "num_tokens": 24506893.0, + "step": 956 + }, + { + "epoch": 0.10509554140127389, + "grad_norm": 2.3738350868225098, + "learning_rate": 5.246981339187706e-07, + "loss": 1.0865, + "mean_token_accuracy": 0.6755526661872864, + "num_tokens": 24534300.0, + "step": 957 + }, + { + "epoch": 0.10520535910388755, + "grad_norm": 2.2322165966033936, + "learning_rate": 5.252469813391877e-07, + "loss": 1.0353, + "mean_token_accuracy": 0.689692497253418, + "num_tokens": 24561583.0, + "step": 958 + }, + { + "epoch": 0.10531517680650121, + "grad_norm": 2.414199113845825, + "learning_rate": 5.257958287596049e-07, + "loss": 1.039, + "mean_token_accuracy": 0.6917953491210938, + "num_tokens": 24585797.0, + "step": 959 + }, + { + "epoch": 0.10542499450911487, + "grad_norm": 2.1763570308685303, + "learning_rate": 5.263446761800219e-07, + "loss": 1.1183, + "mean_token_accuracy": 0.6698108911514282, + "num_tokens": 24612514.0, + "step": 960 + }, + { + "epoch": 0.10553481221172853, + "grad_norm": 2.124251365661621, + "learning_rate": 5.26893523600439e-07, + "loss": 1.0028, + "mean_token_accuracy": 0.6990473866462708, + "num_tokens": 24642745.0, + "step": 961 + }, + { + "epoch": 0.10564462991434219, + "grad_norm": 2.2875208854675293, + "learning_rate": 5.274423710208562e-07, + "loss": 1.0098, + "mean_token_accuracy": 0.7039997577667236, + "num_tokens": 24669571.0, + "step": 962 + }, + { + "epoch": 0.10575444761695585, + "grad_norm": 2.3891451358795166, + "learning_rate": 5.279912184412732e-07, + "loss": 1.083, + "mean_token_accuracy": 0.685522198677063, + "num_tokens": 24694106.0, + "step": 963 + }, + { + "epoch": 0.10586426531956951, + "grad_norm": 2.1358938217163086, + "learning_rate": 5.285400658616904e-07, + "loss": 1.0739, + "mean_token_accuracy": 0.6816637516021729, + "num_tokens": 24724129.0, + "step": 964 + }, + { + "epoch": 0.10597408302218318, + "grad_norm": 2.3745601177215576, + "learning_rate": 5.290889132821076e-07, + "loss": 1.0642, + "mean_token_accuracy": 0.6947343349456787, + "num_tokens": 24748147.0, + "step": 965 + }, + { + "epoch": 0.10608390072479684, + "grad_norm": 2.2350800037384033, + "learning_rate": 5.296377607025246e-07, + "loss": 1.0666, + "mean_token_accuracy": 0.696893036365509, + "num_tokens": 24774489.0, + "step": 966 + }, + { + "epoch": 0.1061937184274105, + "grad_norm": 2.0501437187194824, + "learning_rate": 5.301866081229418e-07, + "loss": 1.064, + "mean_token_accuracy": 0.679713785648346, + "num_tokens": 24804971.0, + "step": 967 + }, + { + "epoch": 0.10630353613002416, + "grad_norm": 2.346466302871704, + "learning_rate": 5.30735455543359e-07, + "loss": 1.0405, + "mean_token_accuracy": 0.6923303008079529, + "num_tokens": 24831067.0, + "step": 968 + }, + { + "epoch": 0.10641335383263782, + "grad_norm": 2.565016269683838, + "learning_rate": 5.31284302963776e-07, + "loss": 1.0269, + "mean_token_accuracy": 0.6945549249649048, + "num_tokens": 24851980.0, + "step": 969 + }, + { + "epoch": 0.10652317153525148, + "grad_norm": 2.381258964538574, + "learning_rate": 5.318331503841932e-07, + "loss": 0.961, + "mean_token_accuracy": 0.7095081806182861, + "num_tokens": 24876329.0, + "step": 970 + }, + { + "epoch": 0.10663298923786514, + "grad_norm": 2.1844558715820312, + "learning_rate": 5.323819978046103e-07, + "loss": 1.0336, + "mean_token_accuracy": 0.6919140815734863, + "num_tokens": 24905750.0, + "step": 971 + }, + { + "epoch": 0.1067428069404788, + "grad_norm": 2.5642073154449463, + "learning_rate": 5.329308452250274e-07, + "loss": 1.0331, + "mean_token_accuracy": 0.6980897188186646, + "num_tokens": 24927744.0, + "step": 972 + }, + { + "epoch": 0.10685262464309246, + "grad_norm": 2.0830209255218506, + "learning_rate": 5.334796926454446e-07, + "loss": 1.0103, + "mean_token_accuracy": 0.7002950310707092, + "num_tokens": 24957243.0, + "step": 973 + }, + { + "epoch": 0.10696244234570612, + "grad_norm": 2.1603569984436035, + "learning_rate": 5.340285400658617e-07, + "loss": 1.1033, + "mean_token_accuracy": 0.6832617521286011, + "num_tokens": 24985451.0, + "step": 974 + }, + { + "epoch": 0.1070722600483198, + "grad_norm": 2.5760796070098877, + "learning_rate": 5.345773874862788e-07, + "loss": 1.0149, + "mean_token_accuracy": 0.6887383460998535, + "num_tokens": 25005722.0, + "step": 975 + }, + { + "epoch": 0.10718207775093345, + "grad_norm": 2.5403568744659424, + "learning_rate": 5.35126234906696e-07, + "loss": 0.966, + "mean_token_accuracy": 0.7056225538253784, + "num_tokens": 25027321.0, + "step": 976 + }, + { + "epoch": 0.10729189545354711, + "grad_norm": 2.191742420196533, + "learning_rate": 5.35675082327113e-07, + "loss": 1.0816, + "mean_token_accuracy": 0.679741382598877, + "num_tokens": 25055892.0, + "step": 977 + }, + { + "epoch": 0.10740171315616077, + "grad_norm": 2.4613516330718994, + "learning_rate": 5.362239297475302e-07, + "loss": 1.1035, + "mean_token_accuracy": 0.6730610728263855, + "num_tokens": 25078099.0, + "step": 978 + }, + { + "epoch": 0.10751153085877443, + "grad_norm": 2.404207706451416, + "learning_rate": 5.367727771679473e-07, + "loss": 1.0301, + "mean_token_accuracy": 0.6949062347412109, + "num_tokens": 25104225.0, + "step": 979 + }, + { + "epoch": 0.1076213485613881, + "grad_norm": 2.4213907718658447, + "learning_rate": 5.373216245883643e-07, + "loss": 0.9859, + "mean_token_accuracy": 0.7010424137115479, + "num_tokens": 25128130.0, + "step": 980 + }, + { + "epoch": 0.10773116626400175, + "grad_norm": 2.1897027492523193, + "learning_rate": 5.378704720087815e-07, + "loss": 1.136, + "mean_token_accuracy": 0.6746217012405396, + "num_tokens": 25158531.0, + "step": 981 + }, + { + "epoch": 0.10784098396661541, + "grad_norm": 2.0647239685058594, + "learning_rate": 5.384193194291986e-07, + "loss": 1.0605, + "mean_token_accuracy": 0.6948001384735107, + "num_tokens": 25188456.0, + "step": 982 + }, + { + "epoch": 0.10795080166922907, + "grad_norm": 2.2559001445770264, + "learning_rate": 5.389681668496157e-07, + "loss": 1.1696, + "mean_token_accuracy": 0.6640419363975525, + "num_tokens": 25218216.0, + "step": 983 + }, + { + "epoch": 0.10806061937184275, + "grad_norm": 2.1173818111419678, + "learning_rate": 5.395170142700329e-07, + "loss": 1.069, + "mean_token_accuracy": 0.6780548095703125, + "num_tokens": 25247132.0, + "step": 984 + }, + { + "epoch": 0.10817043707445641, + "grad_norm": 2.426323413848877, + "learning_rate": 5.4006586169045e-07, + "loss": 1.0437, + "mean_token_accuracy": 0.692703366279602, + "num_tokens": 25272375.0, + "step": 985 + }, + { + "epoch": 0.10828025477707007, + "grad_norm": 2.7142221927642822, + "learning_rate": 5.406147091108671e-07, + "loss": 1.072, + "mean_token_accuracy": 0.6825641393661499, + "num_tokens": 25295034.0, + "step": 986 + }, + { + "epoch": 0.10839007247968373, + "grad_norm": 3.355034112930298, + "learning_rate": 5.411635565312843e-07, + "loss": 1.0387, + "mean_token_accuracy": 0.694412350654602, + "num_tokens": 25318012.0, + "step": 987 + }, + { + "epoch": 0.10849989018229739, + "grad_norm": 2.2047107219696045, + "learning_rate": 5.417124039517014e-07, + "loss": 1.1517, + "mean_token_accuracy": 0.667366623878479, + "num_tokens": 25347601.0, + "step": 988 + }, + { + "epoch": 0.10860970788491105, + "grad_norm": 2.328479290008545, + "learning_rate": 5.422612513721185e-07, + "loss": 1.0717, + "mean_token_accuracy": 0.6843074560165405, + "num_tokens": 25374254.0, + "step": 989 + }, + { + "epoch": 0.1087195255875247, + "grad_norm": 2.318492889404297, + "learning_rate": 5.428100987925357e-07, + "loss": 1.0434, + "mean_token_accuracy": 0.6919843554496765, + "num_tokens": 25400219.0, + "step": 990 + }, + { + "epoch": 0.10882934329013837, + "grad_norm": 2.5740303993225098, + "learning_rate": 5.433589462129528e-07, + "loss": 1.0543, + "mean_token_accuracy": 0.7018737196922302, + "num_tokens": 25423616.0, + "step": 991 + }, + { + "epoch": 0.10893916099275203, + "grad_norm": 2.3414688110351562, + "learning_rate": 5.439077936333699e-07, + "loss": 1.0043, + "mean_token_accuracy": 0.6965492963790894, + "num_tokens": 25449074.0, + "step": 992 + }, + { + "epoch": 0.10904897869536569, + "grad_norm": 2.55647611618042, + "learning_rate": 5.44456641053787e-07, + "loss": 1.0731, + "mean_token_accuracy": 0.6933451294898987, + "num_tokens": 25471161.0, + "step": 993 + }, + { + "epoch": 0.10915879639797936, + "grad_norm": 2.054414749145508, + "learning_rate": 5.450054884742042e-07, + "loss": 1.1306, + "mean_token_accuracy": 0.6612062454223633, + "num_tokens": 25503983.0, + "step": 994 + }, + { + "epoch": 0.10926861410059302, + "grad_norm": 2.5204780101776123, + "learning_rate": 5.455543358946213e-07, + "loss": 0.9984, + "mean_token_accuracy": 0.7032999396324158, + "num_tokens": 25526329.0, + "step": 995 + }, + { + "epoch": 0.10937843180320668, + "grad_norm": 2.4399986267089844, + "learning_rate": 5.461031833150384e-07, + "loss": 0.9569, + "mean_token_accuracy": 0.7095093727111816, + "num_tokens": 25549683.0, + "step": 996 + }, + { + "epoch": 0.10948824950582034, + "grad_norm": 2.4259531497955322, + "learning_rate": 5.466520307354556e-07, + "loss": 1.0524, + "mean_token_accuracy": 0.6864868402481079, + "num_tokens": 25573988.0, + "step": 997 + }, + { + "epoch": 0.109598067208434, + "grad_norm": 2.4914143085479736, + "learning_rate": 5.472008781558726e-07, + "loss": 1.0348, + "mean_token_accuracy": 0.6938261389732361, + "num_tokens": 25596525.0, + "step": 998 + }, + { + "epoch": 0.10970788491104766, + "grad_norm": 2.453819990158081, + "learning_rate": 5.477497255762897e-07, + "loss": 1.1723, + "mean_token_accuracy": 0.669230043888092, + "num_tokens": 25623826.0, + "step": 999 + }, + { + "epoch": 0.10981770261366132, + "grad_norm": 2.104861259460449, + "learning_rate": 5.482985729967069e-07, + "loss": 0.985, + "mean_token_accuracy": 0.7023930549621582, + "num_tokens": 25651301.0, + "step": 1000 + }, + { + "epoch": 0.10992752031627498, + "grad_norm": 2.368288516998291, + "learning_rate": 5.48847420417124e-07, + "loss": 1.0427, + "mean_token_accuracy": 0.6854298114776611, + "num_tokens": 25677511.0, + "step": 1001 + }, + { + "epoch": 0.11003733801888864, + "grad_norm": 2.1500868797302246, + "learning_rate": 5.493962678375411e-07, + "loss": 1.1568, + "mean_token_accuracy": 0.6641703844070435, + "num_tokens": 25707743.0, + "step": 1002 + }, + { + "epoch": 0.11014715572150231, + "grad_norm": 2.2428293228149414, + "learning_rate": 5.499451152579583e-07, + "loss": 1.0926, + "mean_token_accuracy": 0.6813926696777344, + "num_tokens": 25735448.0, + "step": 1003 + }, + { + "epoch": 0.11025697342411597, + "grad_norm": 2.21346378326416, + "learning_rate": 5.504939626783753e-07, + "loss": 1.0133, + "mean_token_accuracy": 0.69447261095047, + "num_tokens": 25762476.0, + "step": 1004 + }, + { + "epoch": 0.11036679112672963, + "grad_norm": 2.3779382705688477, + "learning_rate": 5.510428100987925e-07, + "loss": 1.1559, + "mean_token_accuracy": 0.6560201048851013, + "num_tokens": 25787744.0, + "step": 1005 + }, + { + "epoch": 0.11047660882934329, + "grad_norm": 2.7456791400909424, + "learning_rate": 5.515916575192097e-07, + "loss": 0.9926, + "mean_token_accuracy": 0.7015072107315063, + "num_tokens": 25805881.0, + "step": 1006 + }, + { + "epoch": 0.11058642653195695, + "grad_norm": 2.4101333618164062, + "learning_rate": 5.521405049396267e-07, + "loss": 1.0748, + "mean_token_accuracy": 0.6852802038192749, + "num_tokens": 25831261.0, + "step": 1007 + }, + { + "epoch": 0.11069624423457061, + "grad_norm": 2.4068470001220703, + "learning_rate": 5.526893523600439e-07, + "loss": 0.966, + "mean_token_accuracy": 0.706538200378418, + "num_tokens": 25854559.0, + "step": 1008 + }, + { + "epoch": 0.11080606193718427, + "grad_norm": 2.851227283477783, + "learning_rate": 5.532381997804611e-07, + "loss": 0.9555, + "mean_token_accuracy": 0.7104785442352295, + "num_tokens": 25871973.0, + "step": 1009 + }, + { + "epoch": 0.11091587963979793, + "grad_norm": 2.3245604038238525, + "learning_rate": 5.537870472008781e-07, + "loss": 1.0891, + "mean_token_accuracy": 0.6770222187042236, + "num_tokens": 25896845.0, + "step": 1010 + }, + { + "epoch": 0.11102569734241159, + "grad_norm": 2.4637258052825928, + "learning_rate": 5.543358946212953e-07, + "loss": 1.0041, + "mean_token_accuracy": 0.7050315141677856, + "num_tokens": 25919439.0, + "step": 1011 + }, + { + "epoch": 0.11113551504502525, + "grad_norm": 2.327885866165161, + "learning_rate": 5.548847420417125e-07, + "loss": 1.098, + "mean_token_accuracy": 0.6721364855766296, + "num_tokens": 25946103.0, + "step": 1012 + }, + { + "epoch": 0.11124533274763893, + "grad_norm": 2.0441620349884033, + "learning_rate": 5.554335894621295e-07, + "loss": 1.0348, + "mean_token_accuracy": 0.6931048035621643, + "num_tokens": 25978150.0, + "step": 1013 + }, + { + "epoch": 0.11135515045025259, + "grad_norm": 2.183525323867798, + "learning_rate": 5.559824368825467e-07, + "loss": 1.1141, + "mean_token_accuracy": 0.6741337180137634, + "num_tokens": 26008346.0, + "step": 1014 + }, + { + "epoch": 0.11146496815286625, + "grad_norm": 2.2752342224121094, + "learning_rate": 5.565312843029637e-07, + "loss": 1.0845, + "mean_token_accuracy": 0.6859410405158997, + "num_tokens": 26034528.0, + "step": 1015 + }, + { + "epoch": 0.1115747858554799, + "grad_norm": 2.6491031646728516, + "learning_rate": 5.570801317233809e-07, + "loss": 0.9781, + "mean_token_accuracy": 0.7067182064056396, + "num_tokens": 26053689.0, + "step": 1016 + }, + { + "epoch": 0.11168460355809356, + "grad_norm": 2.3286521434783936, + "learning_rate": 5.57628979143798e-07, + "loss": 1.0595, + "mean_token_accuracy": 0.6862612366676331, + "num_tokens": 26080392.0, + "step": 1017 + }, + { + "epoch": 0.11179442126070722, + "grad_norm": 2.402963638305664, + "learning_rate": 5.58177826564215e-07, + "loss": 0.9948, + "mean_token_accuracy": 0.6949711441993713, + "num_tokens": 26104164.0, + "step": 1018 + }, + { + "epoch": 0.11190423896332088, + "grad_norm": 2.352715253829956, + "learning_rate": 5.587266739846322e-07, + "loss": 0.9894, + "mean_token_accuracy": 0.7018221616744995, + "num_tokens": 26128456.0, + "step": 1019 + }, + { + "epoch": 0.11201405666593454, + "grad_norm": 2.214827060699463, + "learning_rate": 5.592755214050494e-07, + "loss": 1.123, + "mean_token_accuracy": 0.668956458568573, + "num_tokens": 26157411.0, + "step": 1020 + }, + { + "epoch": 0.1121238743685482, + "grad_norm": 2.513094186782837, + "learning_rate": 5.598243688254664e-07, + "loss": 1.0841, + "mean_token_accuracy": 0.6863056421279907, + "num_tokens": 26184089.0, + "step": 1021 + }, + { + "epoch": 0.11223369207116188, + "grad_norm": 2.719562530517578, + "learning_rate": 5.603732162458836e-07, + "loss": 1.0425, + "mean_token_accuracy": 0.6898708343505859, + "num_tokens": 26204190.0, + "step": 1022 + }, + { + "epoch": 0.11234350977377554, + "grad_norm": 2.1728341579437256, + "learning_rate": 5.609220636663008e-07, + "loss": 0.964, + "mean_token_accuracy": 0.7089875936508179, + "num_tokens": 26230819.0, + "step": 1023 + }, + { + "epoch": 0.1124533274763892, + "grad_norm": 2.307481050491333, + "learning_rate": 5.614709110867178e-07, + "loss": 1.0806, + "mean_token_accuracy": 0.676567018032074, + "num_tokens": 26256558.0, + "step": 1024 + }, + { + "epoch": 0.11256314517900286, + "grad_norm": 2.3932111263275146, + "learning_rate": 5.62019758507135e-07, + "loss": 1.0313, + "mean_token_accuracy": 0.6917822360992432, + "num_tokens": 26281451.0, + "step": 1025 + }, + { + "epoch": 0.11267296288161652, + "grad_norm": 2.25593638420105, + "learning_rate": 5.625686059275521e-07, + "loss": 1.0835, + "mean_token_accuracy": 0.679121732711792, + "num_tokens": 26309209.0, + "step": 1026 + }, + { + "epoch": 0.11278278058423018, + "grad_norm": 2.434828758239746, + "learning_rate": 5.631174533479692e-07, + "loss": 1.0949, + "mean_token_accuracy": 0.6832535266876221, + "num_tokens": 26333070.0, + "step": 1027 + }, + { + "epoch": 0.11289259828684384, + "grad_norm": 2.526679754257202, + "learning_rate": 5.636663007683864e-07, + "loss": 1.1226, + "mean_token_accuracy": 0.6670439839363098, + "num_tokens": 26355910.0, + "step": 1028 + }, + { + "epoch": 0.1130024159894575, + "grad_norm": 2.3124046325683594, + "learning_rate": 5.642151481888035e-07, + "loss": 1.1774, + "mean_token_accuracy": 0.65590900182724, + "num_tokens": 26384400.0, + "step": 1029 + }, + { + "epoch": 0.11311223369207116, + "grad_norm": 2.224581718444824, + "learning_rate": 5.647639956092206e-07, + "loss": 1.0876, + "mean_token_accuracy": 0.677108883857727, + "num_tokens": 26411692.0, + "step": 1030 + }, + { + "epoch": 0.11322205139468482, + "grad_norm": 2.4885566234588623, + "learning_rate": 5.653128430296378e-07, + "loss": 1.0319, + "mean_token_accuracy": 0.6983217000961304, + "num_tokens": 26434230.0, + "step": 1031 + }, + { + "epoch": 0.11333186909729849, + "grad_norm": 2.333879232406616, + "learning_rate": 5.658616904500549e-07, + "loss": 1.08, + "mean_token_accuracy": 0.6782866716384888, + "num_tokens": 26461274.0, + "step": 1032 + }, + { + "epoch": 0.11344168679991215, + "grad_norm": 2.1287612915039062, + "learning_rate": 5.66410537870472e-07, + "loss": 1.1271, + "mean_token_accuracy": 0.6717129945755005, + "num_tokens": 26492137.0, + "step": 1033 + }, + { + "epoch": 0.11355150450252581, + "grad_norm": 2.1669535636901855, + "learning_rate": 5.669593852908892e-07, + "loss": 1.1472, + "mean_token_accuracy": 0.663504421710968, + "num_tokens": 26521031.0, + "step": 1034 + }, + { + "epoch": 0.11366132220513947, + "grad_norm": 2.3938584327697754, + "learning_rate": 5.675082327113063e-07, + "loss": 1.0104, + "mean_token_accuracy": 0.6957409977912903, + "num_tokens": 26544837.0, + "step": 1035 + }, + { + "epoch": 0.11377113990775313, + "grad_norm": 2.649188280105591, + "learning_rate": 5.680570801317233e-07, + "loss": 1.0792, + "mean_token_accuracy": 0.6911489963531494, + "num_tokens": 26566035.0, + "step": 1036 + }, + { + "epoch": 0.11388095761036679, + "grad_norm": 2.3228676319122314, + "learning_rate": 5.686059275521404e-07, + "loss": 1.0324, + "mean_token_accuracy": 0.6925108432769775, + "num_tokens": 26591517.0, + "step": 1037 + }, + { + "epoch": 0.11399077531298045, + "grad_norm": 2.534559488296509, + "learning_rate": 5.691547749725576e-07, + "loss": 1.036, + "mean_token_accuracy": 0.6927147507667542, + "num_tokens": 26614591.0, + "step": 1038 + }, + { + "epoch": 0.11410059301559411, + "grad_norm": 2.352809429168701, + "learning_rate": 5.697036223929747e-07, + "loss": 1.1552, + "mean_token_accuracy": 0.6607364416122437, + "num_tokens": 26642033.0, + "step": 1039 + }, + { + "epoch": 0.11421041071820777, + "grad_norm": 2.2648024559020996, + "learning_rate": 5.702524698133918e-07, + "loss": 1.0073, + "mean_token_accuracy": 0.6998096108436584, + "num_tokens": 26666960.0, + "step": 1040 + }, + { + "epoch": 0.11432022842082144, + "grad_norm": 2.1124885082244873, + "learning_rate": 5.70801317233809e-07, + "loss": 1.1703, + "mean_token_accuracy": 0.6502271890640259, + "num_tokens": 26700478.0, + "step": 1041 + }, + { + "epoch": 0.1144300461234351, + "grad_norm": 2.263082504272461, + "learning_rate": 5.713501646542261e-07, + "loss": 1.0943, + "mean_token_accuracy": 0.6714516282081604, + "num_tokens": 26728398.0, + "step": 1042 + }, + { + "epoch": 0.11453986382604876, + "grad_norm": 1.999738097190857, + "learning_rate": 5.718990120746432e-07, + "loss": 1.0863, + "mean_token_accuracy": 0.6764565110206604, + "num_tokens": 26760802.0, + "step": 1043 + }, + { + "epoch": 0.11464968152866242, + "grad_norm": 2.714129686355591, + "learning_rate": 5.724478594950604e-07, + "loss": 0.9454, + "mean_token_accuracy": 0.7107506394386292, + "num_tokens": 26779286.0, + "step": 1044 + }, + { + "epoch": 0.11475949923127608, + "grad_norm": 2.5428833961486816, + "learning_rate": 5.729967069154775e-07, + "loss": 1.1106, + "mean_token_accuracy": 0.6716949343681335, + "num_tokens": 26803555.0, + "step": 1045 + }, + { + "epoch": 0.11486931693388974, + "grad_norm": 2.3323607444763184, + "learning_rate": 5.735455543358946e-07, + "loss": 0.9983, + "mean_token_accuracy": 0.7008629441261292, + "num_tokens": 26828254.0, + "step": 1046 + }, + { + "epoch": 0.1149791346365034, + "grad_norm": 2.2197439670562744, + "learning_rate": 5.740944017563118e-07, + "loss": 1.0223, + "mean_token_accuracy": 0.6904281377792358, + "num_tokens": 26855395.0, + "step": 1047 + }, + { + "epoch": 0.11508895233911706, + "grad_norm": 2.126365900039673, + "learning_rate": 5.746432491767288e-07, + "loss": 1.1351, + "mean_token_accuracy": 0.6627999544143677, + "num_tokens": 26884496.0, + "step": 1048 + }, + { + "epoch": 0.11519877004173072, + "grad_norm": 2.2202718257904053, + "learning_rate": 5.75192096597146e-07, + "loss": 1.0958, + "mean_token_accuracy": 0.6723947525024414, + "num_tokens": 26914020.0, + "step": 1049 + }, + { + "epoch": 0.11530858774434438, + "grad_norm": 2.297539710998535, + "learning_rate": 5.757409440175632e-07, + "loss": 1.0245, + "mean_token_accuracy": 0.689984142780304, + "num_tokens": 26940388.0, + "step": 1050 + }, + { + "epoch": 0.11541840544695806, + "grad_norm": 2.340196371078491, + "learning_rate": 5.762897914379802e-07, + "loss": 1.059, + "mean_token_accuracy": 0.6947457194328308, + "num_tokens": 26966311.0, + "step": 1051 + }, + { + "epoch": 0.11552822314957172, + "grad_norm": 2.549572229385376, + "learning_rate": 5.768386388583974e-07, + "loss": 1.0052, + "mean_token_accuracy": 0.7019009590148926, + "num_tokens": 26991539.0, + "step": 1052 + }, + { + "epoch": 0.11563804085218538, + "grad_norm": 2.414240837097168, + "learning_rate": 5.773874862788145e-07, + "loss": 1.1643, + "mean_token_accuracy": 0.6624428033828735, + "num_tokens": 27017836.0, + "step": 1053 + }, + { + "epoch": 0.11574785855479904, + "grad_norm": 2.2150723934173584, + "learning_rate": 5.779363336992316e-07, + "loss": 1.1104, + "mean_token_accuracy": 0.6644688844680786, + "num_tokens": 27047333.0, + "step": 1054 + }, + { + "epoch": 0.1158576762574127, + "grad_norm": 2.204375982284546, + "learning_rate": 5.784851811196487e-07, + "loss": 1.1025, + "mean_token_accuracy": 0.6714568138122559, + "num_tokens": 27077017.0, + "step": 1055 + }, + { + "epoch": 0.11596749396002635, + "grad_norm": 1.947204351425171, + "learning_rate": 5.790340285400658e-07, + "loss": 1.0728, + "mean_token_accuracy": 0.6877516508102417, + "num_tokens": 27110927.0, + "step": 1056 + }, + { + "epoch": 0.11607731166264001, + "grad_norm": 2.5462141036987305, + "learning_rate": 5.795828759604829e-07, + "loss": 1.1901, + "mean_token_accuracy": 0.6544251441955566, + "num_tokens": 27133267.0, + "step": 1057 + }, + { + "epoch": 0.11618712936525367, + "grad_norm": 2.3248965740203857, + "learning_rate": 5.801317233809001e-07, + "loss": 1.1039, + "mean_token_accuracy": 0.6859447956085205, + "num_tokens": 27159155.0, + "step": 1058 + }, + { + "epoch": 0.11629694706786733, + "grad_norm": 2.608325958251953, + "learning_rate": 5.806805708013171e-07, + "loss": 1.0613, + "mean_token_accuracy": 0.686715841293335, + "num_tokens": 27181710.0, + "step": 1059 + }, + { + "epoch": 0.11640676477048101, + "grad_norm": 2.4183547496795654, + "learning_rate": 5.812294182217343e-07, + "loss": 1.0512, + "mean_token_accuracy": 0.6902159452438354, + "num_tokens": 27207820.0, + "step": 1060 + }, + { + "epoch": 0.11651658247309467, + "grad_norm": 2.3369014263153076, + "learning_rate": 5.817782656421515e-07, + "loss": 1.0626, + "mean_token_accuracy": 0.6875239610671997, + "num_tokens": 27232849.0, + "step": 1061 + }, + { + "epoch": 0.11662640017570833, + "grad_norm": 2.7119455337524414, + "learning_rate": 5.823271130625685e-07, + "loss": 1.043, + "mean_token_accuracy": 0.6919645071029663, + "num_tokens": 27254366.0, + "step": 1062 + }, + { + "epoch": 0.11673621787832199, + "grad_norm": 2.3980422019958496, + "learning_rate": 5.828759604829857e-07, + "loss": 1.03, + "mean_token_accuracy": 0.6931745409965515, + "num_tokens": 27279662.0, + "step": 1063 + }, + { + "epoch": 0.11684603558093565, + "grad_norm": 2.415640354156494, + "learning_rate": 5.834248079034029e-07, + "loss": 1.0652, + "mean_token_accuracy": 0.6862539649009705, + "num_tokens": 27304298.0, + "step": 1064 + }, + { + "epoch": 0.11695585328354931, + "grad_norm": 2.14017391204834, + "learning_rate": 5.839736553238199e-07, + "loss": 1.1516, + "mean_token_accuracy": 0.6754164099693298, + "num_tokens": 27335238.0, + "step": 1065 + }, + { + "epoch": 0.11706567098616297, + "grad_norm": 2.117936134338379, + "learning_rate": 5.845225027442371e-07, + "loss": 1.1532, + "mean_token_accuracy": 0.6593514680862427, + "num_tokens": 27365343.0, + "step": 1066 + }, + { + "epoch": 0.11717548868877663, + "grad_norm": 2.733630657196045, + "learning_rate": 5.850713501646543e-07, + "loss": 1.0082, + "mean_token_accuracy": 0.6952943801879883, + "num_tokens": 27387654.0, + "step": 1067 + }, + { + "epoch": 0.11728530639139029, + "grad_norm": 2.13838791847229, + "learning_rate": 5.856201975850713e-07, + "loss": 1.1, + "mean_token_accuracy": 0.6792137622833252, + "num_tokens": 27417160.0, + "step": 1068 + }, + { + "epoch": 0.11739512409400395, + "grad_norm": 2.069648027420044, + "learning_rate": 5.861690450054885e-07, + "loss": 1.16, + "mean_token_accuracy": 0.6572467088699341, + "num_tokens": 27449507.0, + "step": 1069 + }, + { + "epoch": 0.11750494179661762, + "grad_norm": 2.2511284351348877, + "learning_rate": 5.867178924259056e-07, + "loss": 1.0399, + "mean_token_accuracy": 0.6944482922554016, + "num_tokens": 27475706.0, + "step": 1070 + }, + { + "epoch": 0.11761475949923128, + "grad_norm": 2.613856315612793, + "learning_rate": 5.872667398463227e-07, + "loss": 1.0078, + "mean_token_accuracy": 0.6952537894248962, + "num_tokens": 27496246.0, + "step": 1071 + }, + { + "epoch": 0.11772457720184494, + "grad_norm": 2.3582892417907715, + "learning_rate": 5.878155872667399e-07, + "loss": 1.0807, + "mean_token_accuracy": 0.6804640889167786, + "num_tokens": 27525228.0, + "step": 1072 + }, + { + "epoch": 0.1178343949044586, + "grad_norm": 2.6869428157806396, + "learning_rate": 5.88364434687157e-07, + "loss": 1.0461, + "mean_token_accuracy": 0.6882572174072266, + "num_tokens": 27546118.0, + "step": 1073 + }, + { + "epoch": 0.11794421260707226, + "grad_norm": 2.644705057144165, + "learning_rate": 5.88913282107574e-07, + "loss": 1.1294, + "mean_token_accuracy": 0.6832899451255798, + "num_tokens": 27569284.0, + "step": 1074 + }, + { + "epoch": 0.11805403030968592, + "grad_norm": 2.2151684761047363, + "learning_rate": 5.894621295279912e-07, + "loss": 1.038, + "mean_token_accuracy": 0.6911913156509399, + "num_tokens": 27597112.0, + "step": 1075 + }, + { + "epoch": 0.11816384801229958, + "grad_norm": 2.1751012802124023, + "learning_rate": 5.900109769484083e-07, + "loss": 1.064, + "mean_token_accuracy": 0.681155800819397, + "num_tokens": 27625031.0, + "step": 1076 + }, + { + "epoch": 0.11827366571491324, + "grad_norm": 2.3627071380615234, + "learning_rate": 5.905598243688254e-07, + "loss": 1.0678, + "mean_token_accuracy": 0.6781623959541321, + "num_tokens": 27650949.0, + "step": 1077 + }, + { + "epoch": 0.1183834834175269, + "grad_norm": 2.619168996810913, + "learning_rate": 5.911086717892426e-07, + "loss": 1.1544, + "mean_token_accuracy": 0.6670091152191162, + "num_tokens": 27674363.0, + "step": 1078 + }, + { + "epoch": 0.11849330112014057, + "grad_norm": 2.450515031814575, + "learning_rate": 5.916575192096597e-07, + "loss": 1.1501, + "mean_token_accuracy": 0.66014164686203, + "num_tokens": 27699236.0, + "step": 1079 + }, + { + "epoch": 0.11860311882275423, + "grad_norm": 2.066211223602295, + "learning_rate": 5.922063666300768e-07, + "loss": 1.0829, + "mean_token_accuracy": 0.6765162944793701, + "num_tokens": 27731716.0, + "step": 1080 + }, + { + "epoch": 0.1187129365253679, + "grad_norm": 2.303473711013794, + "learning_rate": 5.927552140504939e-07, + "loss": 1.0181, + "mean_token_accuracy": 0.6994929313659668, + "num_tokens": 27756852.0, + "step": 1081 + }, + { + "epoch": 0.11882275422798155, + "grad_norm": 2.556297779083252, + "learning_rate": 5.93304061470911e-07, + "loss": 1.064, + "mean_token_accuracy": 0.6823696494102478, + "num_tokens": 27779814.0, + "step": 1082 + }, + { + "epoch": 0.11893257193059521, + "grad_norm": 2.4718575477600098, + "learning_rate": 5.938529088913282e-07, + "loss": 1.0857, + "mean_token_accuracy": 0.6826986074447632, + "num_tokens": 27803416.0, + "step": 1083 + }, + { + "epoch": 0.11904238963320887, + "grad_norm": 2.2966628074645996, + "learning_rate": 5.944017563117453e-07, + "loss": 1.0961, + "mean_token_accuracy": 0.6777862310409546, + "num_tokens": 27830743.0, + "step": 1084 + }, + { + "epoch": 0.11915220733582253, + "grad_norm": 2.5486958026885986, + "learning_rate": 5.949506037321624e-07, + "loss": 1.1392, + "mean_token_accuracy": 0.6731153726577759, + "num_tokens": 27854983.0, + "step": 1085 + }, + { + "epoch": 0.11926202503843619, + "grad_norm": 2.4577770233154297, + "learning_rate": 5.954994511525796e-07, + "loss": 1.0612, + "mean_token_accuracy": 0.6964850425720215, + "num_tokens": 27879596.0, + "step": 1086 + }, + { + "epoch": 0.11937184274104985, + "grad_norm": 2.21065616607666, + "learning_rate": 5.960482985729967e-07, + "loss": 1.1052, + "mean_token_accuracy": 0.6718060970306396, + "num_tokens": 27907749.0, + "step": 1087 + }, + { + "epoch": 0.11948166044366351, + "grad_norm": 2.3104677200317383, + "learning_rate": 5.965971459934138e-07, + "loss": 1.0885, + "mean_token_accuracy": 0.6735804080963135, + "num_tokens": 27934307.0, + "step": 1088 + }, + { + "epoch": 0.11959147814627719, + "grad_norm": 2.569373846054077, + "learning_rate": 5.971459934138309e-07, + "loss": 0.9696, + "mean_token_accuracy": 0.7042970657348633, + "num_tokens": 27956614.0, + "step": 1089 + }, + { + "epoch": 0.11970129584889085, + "grad_norm": 2.2086360454559326, + "learning_rate": 5.976948408342481e-07, + "loss": 1.0839, + "mean_token_accuracy": 0.6773380041122437, + "num_tokens": 27985768.0, + "step": 1090 + }, + { + "epoch": 0.1198111135515045, + "grad_norm": 2.5468311309814453, + "learning_rate": 5.982436882546652e-07, + "loss": 0.9924, + "mean_token_accuracy": 0.7000941634178162, + "num_tokens": 28008643.0, + "step": 1091 + }, + { + "epoch": 0.11992093125411817, + "grad_norm": 2.2942748069763184, + "learning_rate": 5.987925356750822e-07, + "loss": 1.1257, + "mean_token_accuracy": 0.6684376001358032, + "num_tokens": 28037803.0, + "step": 1092 + }, + { + "epoch": 0.12003074895673183, + "grad_norm": 2.3639581203460693, + "learning_rate": 5.993413830954994e-07, + "loss": 1.0234, + "mean_token_accuracy": 0.6973055601119995, + "num_tokens": 28064166.0, + "step": 1093 + }, + { + "epoch": 0.12014056665934549, + "grad_norm": 2.3030786514282227, + "learning_rate": 5.998902305159165e-07, + "loss": 1.0676, + "mean_token_accuracy": 0.6922204494476318, + "num_tokens": 28087689.0, + "step": 1094 + }, + { + "epoch": 0.12025038436195915, + "grad_norm": 2.23632550239563, + "learning_rate": 6.004390779363336e-07, + "loss": 1.0865, + "mean_token_accuracy": 0.6803595423698425, + "num_tokens": 28117388.0, + "step": 1095 + }, + { + "epoch": 0.1203602020645728, + "grad_norm": 2.535780668258667, + "learning_rate": 6.009879253567508e-07, + "loss": 1.042, + "mean_token_accuracy": 0.6980804204940796, + "num_tokens": 28141285.0, + "step": 1096 + }, + { + "epoch": 0.12047001976718646, + "grad_norm": 2.4178640842437744, + "learning_rate": 6.015367727771679e-07, + "loss": 1.0709, + "mean_token_accuracy": 0.6948413848876953, + "num_tokens": 28166155.0, + "step": 1097 + }, + { + "epoch": 0.12057983746980014, + "grad_norm": 2.3600242137908936, + "learning_rate": 6.02085620197585e-07, + "loss": 1.029, + "mean_token_accuracy": 0.6971770524978638, + "num_tokens": 28192612.0, + "step": 1098 + }, + { + "epoch": 0.1206896551724138, + "grad_norm": 2.2094216346740723, + "learning_rate": 6.026344676180022e-07, + "loss": 1.1103, + "mean_token_accuracy": 0.6815259456634521, + "num_tokens": 28221352.0, + "step": 1099 + }, + { + "epoch": 0.12079947287502746, + "grad_norm": 2.3113889694213867, + "learning_rate": 6.031833150384192e-07, + "loss": 1.0669, + "mean_token_accuracy": 0.6893618106842041, + "num_tokens": 28247588.0, + "step": 1100 + }, + { + "epoch": 0.12090929057764112, + "grad_norm": 2.4775164127349854, + "learning_rate": 6.037321624588364e-07, + "loss": 1.1371, + "mean_token_accuracy": 0.6694380640983582, + "num_tokens": 28273352.0, + "step": 1101 + }, + { + "epoch": 0.12101910828025478, + "grad_norm": 2.299145460128784, + "learning_rate": 6.042810098792536e-07, + "loss": 1.0785, + "mean_token_accuracy": 0.6831938028335571, + "num_tokens": 28299990.0, + "step": 1102 + }, + { + "epoch": 0.12112892598286844, + "grad_norm": 2.7056875228881836, + "learning_rate": 6.048298572996706e-07, + "loss": 0.9983, + "mean_token_accuracy": 0.7008225917816162, + "num_tokens": 28320065.0, + "step": 1103 + }, + { + "epoch": 0.1212387436854821, + "grad_norm": 2.486482858657837, + "learning_rate": 6.053787047200878e-07, + "loss": 1.027, + "mean_token_accuracy": 0.6942160129547119, + "num_tokens": 28342130.0, + "step": 1104 + }, + { + "epoch": 0.12134856138809576, + "grad_norm": 2.6738760471343994, + "learning_rate": 6.05927552140505e-07, + "loss": 0.9583, + "mean_token_accuracy": 0.7188361883163452, + "num_tokens": 28361305.0, + "step": 1105 + }, + { + "epoch": 0.12145837909070942, + "grad_norm": 2.321826696395874, + "learning_rate": 6.06476399560922e-07, + "loss": 1.0479, + "mean_token_accuracy": 0.6891440153121948, + "num_tokens": 28386604.0, + "step": 1106 + }, + { + "epoch": 0.12156819679332308, + "grad_norm": 2.5878071784973145, + "learning_rate": 6.070252469813392e-07, + "loss": 1.0546, + "mean_token_accuracy": 0.6806243658065796, + "num_tokens": 28409431.0, + "step": 1107 + }, + { + "epoch": 0.12167801449593675, + "grad_norm": 2.441490411758423, + "learning_rate": 6.075740944017564e-07, + "loss": 1.102, + "mean_token_accuracy": 0.6811205148696899, + "num_tokens": 28435739.0, + "step": 1108 + }, + { + "epoch": 0.12178783219855041, + "grad_norm": 2.3066225051879883, + "learning_rate": 6.081229418221734e-07, + "loss": 1.052, + "mean_token_accuracy": 0.6850881576538086, + "num_tokens": 28461215.0, + "step": 1109 + }, + { + "epoch": 0.12189764990116407, + "grad_norm": 2.4843499660491943, + "learning_rate": 6.086717892425906e-07, + "loss": 1.0733, + "mean_token_accuracy": 0.6848229765892029, + "num_tokens": 28487570.0, + "step": 1110 + }, + { + "epoch": 0.12200746760377773, + "grad_norm": 2.341872453689575, + "learning_rate": 6.092206366630076e-07, + "loss": 1.0136, + "mean_token_accuracy": 0.6999138593673706, + "num_tokens": 28510181.0, + "step": 1111 + }, + { + "epoch": 0.12211728530639139, + "grad_norm": 2.395625352859497, + "learning_rate": 6.097694840834247e-07, + "loss": 1.0119, + "mean_token_accuracy": 0.6936061978340149, + "num_tokens": 28533387.0, + "step": 1112 + }, + { + "epoch": 0.12222710300900505, + "grad_norm": 2.214881658554077, + "learning_rate": 6.103183315038419e-07, + "loss": 1.0349, + "mean_token_accuracy": 0.7030299305915833, + "num_tokens": 28559657.0, + "step": 1113 + }, + { + "epoch": 0.12233692071161871, + "grad_norm": 2.3607337474823, + "learning_rate": 6.10867178924259e-07, + "loss": 1.1006, + "mean_token_accuracy": 0.6762726902961731, + "num_tokens": 28585885.0, + "step": 1114 + }, + { + "epoch": 0.12244673841423237, + "grad_norm": 2.6292243003845215, + "learning_rate": 6.114160263446761e-07, + "loss": 0.9846, + "mean_token_accuracy": 0.7062618136405945, + "num_tokens": 28606721.0, + "step": 1115 + }, + { + "epoch": 0.12255655611684603, + "grad_norm": 2.215590476989746, + "learning_rate": 6.119648737650933e-07, + "loss": 1.0266, + "mean_token_accuracy": 0.6970869302749634, + "num_tokens": 28633321.0, + "step": 1116 + }, + { + "epoch": 0.1226663738194597, + "grad_norm": 2.4296085834503174, + "learning_rate": 6.125137211855103e-07, + "loss": 1.0037, + "mean_token_accuracy": 0.6934735178947449, + "num_tokens": 28656477.0, + "step": 1117 + }, + { + "epoch": 0.12277619152207336, + "grad_norm": 2.6818795204162598, + "learning_rate": 6.130625686059275e-07, + "loss": 1.1205, + "mean_token_accuracy": 0.6686362624168396, + "num_tokens": 28678010.0, + "step": 1118 + }, + { + "epoch": 0.12288600922468702, + "grad_norm": 2.2570364475250244, + "learning_rate": 6.136114160263447e-07, + "loss": 0.9844, + "mean_token_accuracy": 0.7033611536026001, + "num_tokens": 28702737.0, + "step": 1119 + }, + { + "epoch": 0.12299582692730068, + "grad_norm": 2.2448554039001465, + "learning_rate": 6.141602634467617e-07, + "loss": 1.0296, + "mean_token_accuracy": 0.6937024593353271, + "num_tokens": 28729153.0, + "step": 1120 + }, + { + "epoch": 0.12310564462991434, + "grad_norm": 2.1462910175323486, + "learning_rate": 6.147091108671789e-07, + "loss": 1.0868, + "mean_token_accuracy": 0.679275393486023, + "num_tokens": 28756976.0, + "step": 1121 + }, + { + "epoch": 0.123215462332528, + "grad_norm": 2.4824438095092773, + "learning_rate": 6.15257958287596e-07, + "loss": 1.0331, + "mean_token_accuracy": 0.6889564990997314, + "num_tokens": 28780559.0, + "step": 1122 + }, + { + "epoch": 0.12332528003514166, + "grad_norm": 2.404554843902588, + "learning_rate": 6.158068057080131e-07, + "loss": 1.0618, + "mean_token_accuracy": 0.6762468814849854, + "num_tokens": 28806778.0, + "step": 1123 + }, + { + "epoch": 0.12343509773775532, + "grad_norm": 2.376955509185791, + "learning_rate": 6.163556531284303e-07, + "loss": 1.0369, + "mean_token_accuracy": 0.6841171383857727, + "num_tokens": 28831516.0, + "step": 1124 + }, + { + "epoch": 0.12354491544036898, + "grad_norm": 2.274768590927124, + "learning_rate": 6.169045005488474e-07, + "loss": 1.0492, + "mean_token_accuracy": 0.6863953471183777, + "num_tokens": 28858807.0, + "step": 1125 + }, + { + "epoch": 0.12365473314298264, + "grad_norm": 2.3801181316375732, + "learning_rate": 6.174533479692645e-07, + "loss": 0.9956, + "mean_token_accuracy": 0.7044143676757812, + "num_tokens": 28881859.0, + "step": 1126 + }, + { + "epoch": 0.12376455084559632, + "grad_norm": 2.279102087020874, + "learning_rate": 6.180021953896817e-07, + "loss": 1.0618, + "mean_token_accuracy": 0.6846882104873657, + "num_tokens": 28908039.0, + "step": 1127 + }, + { + "epoch": 0.12387436854820998, + "grad_norm": 2.1424295902252197, + "learning_rate": 6.185510428100988e-07, + "loss": 1.1364, + "mean_token_accuracy": 0.6602370738983154, + "num_tokens": 28937876.0, + "step": 1128 + }, + { + "epoch": 0.12398418625082364, + "grad_norm": 2.145653486251831, + "learning_rate": 6.190998902305159e-07, + "loss": 1.0475, + "mean_token_accuracy": 0.6912420988082886, + "num_tokens": 28965848.0, + "step": 1129 + }, + { + "epoch": 0.1240940039534373, + "grad_norm": 2.3337504863739014, + "learning_rate": 6.196487376509331e-07, + "loss": 1.0214, + "mean_token_accuracy": 0.694859504699707, + "num_tokens": 28988861.0, + "step": 1130 + }, + { + "epoch": 0.12420382165605096, + "grad_norm": 2.3633601665496826, + "learning_rate": 6.201975850713501e-07, + "loss": 1.1132, + "mean_token_accuracy": 0.6678733229637146, + "num_tokens": 29013171.0, + "step": 1131 + }, + { + "epoch": 0.12431363935866462, + "grad_norm": 2.547398328781128, + "learning_rate": 6.207464324917672e-07, + "loss": 1.0875, + "mean_token_accuracy": 0.6843721866607666, + "num_tokens": 29038455.0, + "step": 1132 + }, + { + "epoch": 0.12442345706127828, + "grad_norm": 2.368436336517334, + "learning_rate": 6.212952799121843e-07, + "loss": 1.0539, + "mean_token_accuracy": 0.685149610042572, + "num_tokens": 29062404.0, + "step": 1133 + }, + { + "epoch": 0.12453327476389194, + "grad_norm": 2.4282541275024414, + "learning_rate": 6.218441273326015e-07, + "loss": 1.0641, + "mean_token_accuracy": 0.6891733407974243, + "num_tokens": 29086761.0, + "step": 1134 + }, + { + "epoch": 0.1246430924665056, + "grad_norm": 2.3886876106262207, + "learning_rate": 6.223929747530186e-07, + "loss": 1.0718, + "mean_token_accuracy": 0.680880069732666, + "num_tokens": 29112216.0, + "step": 1135 + }, + { + "epoch": 0.12475291016911927, + "grad_norm": 2.4347763061523438, + "learning_rate": 6.229418221734357e-07, + "loss": 0.9878, + "mean_token_accuracy": 0.6993449330329895, + "num_tokens": 29134007.0, + "step": 1136 + }, + { + "epoch": 0.12486272787173293, + "grad_norm": 2.445854902267456, + "learning_rate": 6.234906695938529e-07, + "loss": 1.1139, + "mean_token_accuracy": 0.670561671257019, + "num_tokens": 29158367.0, + "step": 1137 + }, + { + "epoch": 0.12497254557434659, + "grad_norm": 2.50978684425354, + "learning_rate": 6.2403951701427e-07, + "loss": 1.0441, + "mean_token_accuracy": 0.6940225958824158, + "num_tokens": 29181510.0, + "step": 1138 + }, + { + "epoch": 0.12508236327696023, + "grad_norm": 2.279857873916626, + "learning_rate": 6.245883644346871e-07, + "loss": 1.0026, + "mean_token_accuracy": 0.6988011598587036, + "num_tokens": 29207615.0, + "step": 1139 + }, + { + "epoch": 0.1251921809795739, + "grad_norm": 2.1935713291168213, + "learning_rate": 6.251372118551043e-07, + "loss": 1.0297, + "mean_token_accuracy": 0.6952390670776367, + "num_tokens": 29234389.0, + "step": 1140 + }, + { + "epoch": 0.12530199868218758, + "grad_norm": 2.260809898376465, + "learning_rate": 6.256860592755214e-07, + "loss": 1.1226, + "mean_token_accuracy": 0.6680383086204529, + "num_tokens": 29261140.0, + "step": 1141 + }, + { + "epoch": 0.12541181638480123, + "grad_norm": 2.6248669624328613, + "learning_rate": 6.262349066959385e-07, + "loss": 1.1077, + "mean_token_accuracy": 0.6793994307518005, + "num_tokens": 29284031.0, + "step": 1142 + }, + { + "epoch": 0.1255216340874149, + "grad_norm": 2.340116500854492, + "learning_rate": 6.267837541163557e-07, + "loss": 1.0354, + "mean_token_accuracy": 0.6817914247512817, + "num_tokens": 29309352.0, + "step": 1143 + }, + { + "epoch": 0.12563145179002855, + "grad_norm": 2.557497024536133, + "learning_rate": 6.273326015367727e-07, + "loss": 1.0239, + "mean_token_accuracy": 0.6928813457489014, + "num_tokens": 29330645.0, + "step": 1144 + }, + { + "epoch": 0.12574126949264222, + "grad_norm": 2.5423758029937744, + "learning_rate": 6.278814489571899e-07, + "loss": 1.0545, + "mean_token_accuracy": 0.6902070641517639, + "num_tokens": 29352727.0, + "step": 1145 + }, + { + "epoch": 0.12585108719525587, + "grad_norm": 2.3103039264678955, + "learning_rate": 6.284302963776071e-07, + "loss": 1.1214, + "mean_token_accuracy": 0.6732255220413208, + "num_tokens": 29378146.0, + "step": 1146 + }, + { + "epoch": 0.12596090489786954, + "grad_norm": 2.3044469356536865, + "learning_rate": 6.289791437980241e-07, + "loss": 0.9092, + "mean_token_accuracy": 0.727246880531311, + "num_tokens": 29405264.0, + "step": 1147 + }, + { + "epoch": 0.1260707226004832, + "grad_norm": 2.3740549087524414, + "learning_rate": 6.295279912184413e-07, + "loss": 1.073, + "mean_token_accuracy": 0.6796823143959045, + "num_tokens": 29428897.0, + "step": 1148 + }, + { + "epoch": 0.12618054030309686, + "grad_norm": 2.728165864944458, + "learning_rate": 6.300768386388585e-07, + "loss": 1.0403, + "mean_token_accuracy": 0.6900435090065002, + "num_tokens": 29448891.0, + "step": 1149 + }, + { + "epoch": 0.1262903580057105, + "grad_norm": 1.9019871950149536, + "learning_rate": 6.306256860592754e-07, + "loss": 1.0036, + "mean_token_accuracy": 0.6976126432418823, + "num_tokens": 29484551.0, + "step": 1150 + }, + { + "epoch": 0.12640017570832418, + "grad_norm": 2.1866466999053955, + "learning_rate": 6.311745334796926e-07, + "loss": 1.1156, + "mean_token_accuracy": 0.6710710525512695, + "num_tokens": 29514233.0, + "step": 1151 + }, + { + "epoch": 0.12650999341093785, + "grad_norm": 2.391373634338379, + "learning_rate": 6.317233809001098e-07, + "loss": 1.0439, + "mean_token_accuracy": 0.7003644704818726, + "num_tokens": 29539544.0, + "step": 1152 + }, + { + "epoch": 0.1266198111135515, + "grad_norm": 2.2190680503845215, + "learning_rate": 6.322722283205268e-07, + "loss": 1.116, + "mean_token_accuracy": 0.6703087091445923, + "num_tokens": 29567333.0, + "step": 1153 + }, + { + "epoch": 0.12672962881616517, + "grad_norm": 2.810387372970581, + "learning_rate": 6.32821075740944e-07, + "loss": 0.9552, + "mean_token_accuracy": 0.706829845905304, + "num_tokens": 29586758.0, + "step": 1154 + }, + { + "epoch": 0.12683944651877882, + "grad_norm": 2.1734368801116943, + "learning_rate": 6.33369923161361e-07, + "loss": 1.0355, + "mean_token_accuracy": 0.6878481507301331, + "num_tokens": 29615572.0, + "step": 1155 + }, + { + "epoch": 0.1269492642213925, + "grad_norm": 2.1286890506744385, + "learning_rate": 6.339187705817782e-07, + "loss": 1.0491, + "mean_token_accuracy": 0.6910337805747986, + "num_tokens": 29644348.0, + "step": 1156 + }, + { + "epoch": 0.12705908192400614, + "grad_norm": 2.3013901710510254, + "learning_rate": 6.344676180021954e-07, + "loss": 0.9426, + "mean_token_accuracy": 0.714634358882904, + "num_tokens": 29669279.0, + "step": 1157 + }, + { + "epoch": 0.1271688996266198, + "grad_norm": 2.3741583824157715, + "learning_rate": 6.350164654226124e-07, + "loss": 1.0969, + "mean_token_accuracy": 0.6743388772010803, + "num_tokens": 29693900.0, + "step": 1158 + }, + { + "epoch": 0.12727871732923346, + "grad_norm": 2.3591418266296387, + "learning_rate": 6.355653128430296e-07, + "loss": 1.0333, + "mean_token_accuracy": 0.6909908652305603, + "num_tokens": 29717770.0, + "step": 1159 + }, + { + "epoch": 0.12738853503184713, + "grad_norm": 2.2663512229919434, + "learning_rate": 6.361141602634468e-07, + "loss": 1.0255, + "mean_token_accuracy": 0.6927522420883179, + "num_tokens": 29743832.0, + "step": 1160 + }, + { + "epoch": 0.1274983527344608, + "grad_norm": 2.139529228210449, + "learning_rate": 6.366630076838638e-07, + "loss": 0.9826, + "mean_token_accuracy": 0.7020789384841919, + "num_tokens": 29771859.0, + "step": 1161 + }, + { + "epoch": 0.12760817043707445, + "grad_norm": 2.2380800247192383, + "learning_rate": 6.37211855104281e-07, + "loss": 1.0873, + "mean_token_accuracy": 0.6713510751724243, + "num_tokens": 29800084.0, + "step": 1162 + }, + { + "epoch": 0.12771798813968813, + "grad_norm": 2.3009610176086426, + "learning_rate": 6.377607025246982e-07, + "loss": 1.0826, + "mean_token_accuracy": 0.6766561269760132, + "num_tokens": 29826130.0, + "step": 1163 + }, + { + "epoch": 0.12782780584230177, + "grad_norm": 2.277845621109009, + "learning_rate": 6.383095499451152e-07, + "loss": 1.095, + "mean_token_accuracy": 0.6855454444885254, + "num_tokens": 29854917.0, + "step": 1164 + }, + { + "epoch": 0.12793762354491545, + "grad_norm": 2.1621556282043457, + "learning_rate": 6.388583973655324e-07, + "loss": 1.0573, + "mean_token_accuracy": 0.6882429122924805, + "num_tokens": 29884342.0, + "step": 1165 + }, + { + "epoch": 0.1280474412475291, + "grad_norm": 2.3232555389404297, + "learning_rate": 6.394072447859495e-07, + "loss": 1.0452, + "mean_token_accuracy": 0.6870830059051514, + "num_tokens": 29909337.0, + "step": 1166 + }, + { + "epoch": 0.12815725895014277, + "grad_norm": 2.1992266178131104, + "learning_rate": 6.399560922063666e-07, + "loss": 0.9624, + "mean_token_accuracy": 0.710338294506073, + "num_tokens": 29936579.0, + "step": 1167 + }, + { + "epoch": 0.1282670766527564, + "grad_norm": 2.350184679031372, + "learning_rate": 6.405049396267838e-07, + "loss": 1.005, + "mean_token_accuracy": 0.7020090818405151, + "num_tokens": 29962667.0, + "step": 1168 + }, + { + "epoch": 0.12837689435537009, + "grad_norm": 2.4396731853485107, + "learning_rate": 6.410537870472008e-07, + "loss": 1.0325, + "mean_token_accuracy": 0.6918807625770569, + "num_tokens": 29986017.0, + "step": 1169 + }, + { + "epoch": 0.12848671205798376, + "grad_norm": 2.7227773666381836, + "learning_rate": 6.416026344676179e-07, + "loss": 1.0544, + "mean_token_accuracy": 0.6827831268310547, + "num_tokens": 30006250.0, + "step": 1170 + }, + { + "epoch": 0.1285965297605974, + "grad_norm": 2.1527576446533203, + "learning_rate": 6.421514818880351e-07, + "loss": 1.0044, + "mean_token_accuracy": 0.703477144241333, + "num_tokens": 30034716.0, + "step": 1171 + }, + { + "epoch": 0.12870634746321108, + "grad_norm": 2.4183804988861084, + "learning_rate": 6.427003293084522e-07, + "loss": 1.0041, + "mean_token_accuracy": 0.6920932531356812, + "num_tokens": 30057423.0, + "step": 1172 + }, + { + "epoch": 0.12881616516582473, + "grad_norm": 2.2188503742218018, + "learning_rate": 6.432491767288693e-07, + "loss": 1.0569, + "mean_token_accuracy": 0.6849051117897034, + "num_tokens": 30083922.0, + "step": 1173 + }, + { + "epoch": 0.1289259828684384, + "grad_norm": 2.0148727893829346, + "learning_rate": 6.437980241492865e-07, + "loss": 1.0897, + "mean_token_accuracy": 0.6765892505645752, + "num_tokens": 30115588.0, + "step": 1174 + }, + { + "epoch": 0.12903580057105205, + "grad_norm": 2.489393711090088, + "learning_rate": 6.443468715697036e-07, + "loss": 1.0507, + "mean_token_accuracy": 0.6973329782485962, + "num_tokens": 30139020.0, + "step": 1175 + }, + { + "epoch": 0.12914561827366572, + "grad_norm": 2.3309476375579834, + "learning_rate": 6.448957189901207e-07, + "loss": 1.0488, + "mean_token_accuracy": 0.7040923833847046, + "num_tokens": 30163652.0, + "step": 1176 + }, + { + "epoch": 0.12925543597627936, + "grad_norm": 2.4431798458099365, + "learning_rate": 6.454445664105378e-07, + "loss": 1.0805, + "mean_token_accuracy": 0.6861797571182251, + "num_tokens": 30188861.0, + "step": 1177 + }, + { + "epoch": 0.12936525367889304, + "grad_norm": 2.4993972778320312, + "learning_rate": 6.45993413830955e-07, + "loss": 1.0304, + "mean_token_accuracy": 0.7000340223312378, + "num_tokens": 30211087.0, + "step": 1178 + }, + { + "epoch": 0.1294750713815067, + "grad_norm": 2.66683030128479, + "learning_rate": 6.465422612513721e-07, + "loss": 1.0392, + "mean_token_accuracy": 0.6878511905670166, + "num_tokens": 30231714.0, + "step": 1179 + }, + { + "epoch": 0.12958488908412036, + "grad_norm": 2.555952310562134, + "learning_rate": 6.470911086717892e-07, + "loss": 1.0802, + "mean_token_accuracy": 0.682161808013916, + "num_tokens": 30254580.0, + "step": 1180 + }, + { + "epoch": 0.12969470678673403, + "grad_norm": 2.158823251724243, + "learning_rate": 6.476399560922064e-07, + "loss": 1.1472, + "mean_token_accuracy": 0.6653010845184326, + "num_tokens": 30285628.0, + "step": 1181 + }, + { + "epoch": 0.12980452448934768, + "grad_norm": 2.7250847816467285, + "learning_rate": 6.481888035126235e-07, + "loss": 1.0131, + "mean_token_accuracy": 0.7023658156394958, + "num_tokens": 30306478.0, + "step": 1182 + }, + { + "epoch": 0.12991434219196135, + "grad_norm": 2.497318983078003, + "learning_rate": 6.487376509330406e-07, + "loss": 0.9914, + "mean_token_accuracy": 0.7070196866989136, + "num_tokens": 30329407.0, + "step": 1183 + }, + { + "epoch": 0.130024159894575, + "grad_norm": 2.191632032394409, + "learning_rate": 6.492864983534578e-07, + "loss": 1.1629, + "mean_token_accuracy": 0.6684154272079468, + "num_tokens": 30359821.0, + "step": 1184 + }, + { + "epoch": 0.13013397759718867, + "grad_norm": 2.257852554321289, + "learning_rate": 6.498353457738749e-07, + "loss": 1.0058, + "mean_token_accuracy": 0.6969531774520874, + "num_tokens": 30385032.0, + "step": 1185 + }, + { + "epoch": 0.13024379529980232, + "grad_norm": 2.121271848678589, + "learning_rate": 6.50384193194292e-07, + "loss": 1.0475, + "mean_token_accuracy": 0.6948682069778442, + "num_tokens": 30413971.0, + "step": 1186 + }, + { + "epoch": 0.130353613002416, + "grad_norm": 2.4922213554382324, + "learning_rate": 6.509330406147092e-07, + "loss": 1.0486, + "mean_token_accuracy": 0.6828778982162476, + "num_tokens": 30438097.0, + "step": 1187 + }, + { + "epoch": 0.13046343070502964, + "grad_norm": 3.2422914505004883, + "learning_rate": 6.514818880351261e-07, + "loss": 0.9341, + "mean_token_accuracy": 0.7162164449691772, + "num_tokens": 30453632.0, + "step": 1188 + }, + { + "epoch": 0.1305732484076433, + "grad_norm": 2.724647283554077, + "learning_rate": 6.520307354555433e-07, + "loss": 1.1205, + "mean_token_accuracy": 0.6804147362709045, + "num_tokens": 30474120.0, + "step": 1189 + }, + { + "epoch": 0.13068306611025698, + "grad_norm": 2.443549394607544, + "learning_rate": 6.525795828759604e-07, + "loss": 1.0029, + "mean_token_accuracy": 0.6969353556632996, + "num_tokens": 30496657.0, + "step": 1190 + }, + { + "epoch": 0.13079288381287063, + "grad_norm": 2.4659087657928467, + "learning_rate": 6.531284302963775e-07, + "loss": 1.0603, + "mean_token_accuracy": 0.6943995952606201, + "num_tokens": 30519216.0, + "step": 1191 + }, + { + "epoch": 0.1309027015154843, + "grad_norm": 2.467712879180908, + "learning_rate": 6.536772777167947e-07, + "loss": 0.9896, + "mean_token_accuracy": 0.7035229206085205, + "num_tokens": 30542033.0, + "step": 1192 + }, + { + "epoch": 0.13101251921809795, + "grad_norm": 2.356890916824341, + "learning_rate": 6.542261251372118e-07, + "loss": 1.0581, + "mean_token_accuracy": 0.6912974119186401, + "num_tokens": 30566985.0, + "step": 1193 + }, + { + "epoch": 0.13112233692071162, + "grad_norm": 2.3694472312927246, + "learning_rate": 6.547749725576289e-07, + "loss": 1.0402, + "mean_token_accuracy": 0.6968672871589661, + "num_tokens": 30591276.0, + "step": 1194 + }, + { + "epoch": 0.13123215462332527, + "grad_norm": 2.2745423316955566, + "learning_rate": 6.553238199780461e-07, + "loss": 1.1118, + "mean_token_accuracy": 0.6786168813705444, + "num_tokens": 30618529.0, + "step": 1195 + }, + { + "epoch": 0.13134197232593894, + "grad_norm": 2.46765398979187, + "learning_rate": 6.558726673984632e-07, + "loss": 1.0223, + "mean_token_accuracy": 0.696221113204956, + "num_tokens": 30642884.0, + "step": 1196 + }, + { + "epoch": 0.1314517900285526, + "grad_norm": 2.359302282333374, + "learning_rate": 6.564215148188803e-07, + "loss": 1.0489, + "mean_token_accuracy": 0.6893817186355591, + "num_tokens": 30667104.0, + "step": 1197 + }, + { + "epoch": 0.13156160773116626, + "grad_norm": 2.516847610473633, + "learning_rate": 6.569703622392975e-07, + "loss": 1.0114, + "mean_token_accuracy": 0.6974983215332031, + "num_tokens": 30689856.0, + "step": 1198 + }, + { + "epoch": 0.13167142543377994, + "grad_norm": 2.3151869773864746, + "learning_rate": 6.575192096597145e-07, + "loss": 0.9652, + "mean_token_accuracy": 0.7092156410217285, + "num_tokens": 30713419.0, + "step": 1199 + }, + { + "epoch": 0.13178124313639358, + "grad_norm": 2.4539895057678223, + "learning_rate": 6.580680570801317e-07, + "loss": 1.0073, + "mean_token_accuracy": 0.6939480900764465, + "num_tokens": 30736472.0, + "step": 1200 + }, + { + "epoch": 0.13189106083900726, + "grad_norm": 2.107311725616455, + "learning_rate": 6.586169045005489e-07, + "loss": 0.9822, + "mean_token_accuracy": 0.701869547367096, + "num_tokens": 30763400.0, + "step": 1201 + }, + { + "epoch": 0.1320008785416209, + "grad_norm": 2.26216459274292, + "learning_rate": 6.591657519209659e-07, + "loss": 1.0126, + "mean_token_accuracy": 0.7050834894180298, + "num_tokens": 30788515.0, + "step": 1202 + }, + { + "epoch": 0.13211069624423458, + "grad_norm": 2.122319459915161, + "learning_rate": 6.597145993413831e-07, + "loss": 1.1209, + "mean_token_accuracy": 0.6670966148376465, + "num_tokens": 30818964.0, + "step": 1203 + }, + { + "epoch": 0.13222051394684822, + "grad_norm": 2.263537645339966, + "learning_rate": 6.602634467618003e-07, + "loss": 1.03, + "mean_token_accuracy": 0.6926409006118774, + "num_tokens": 30847236.0, + "step": 1204 + }, + { + "epoch": 0.1323303316494619, + "grad_norm": 1.9456621408462524, + "learning_rate": 6.608122941822173e-07, + "loss": 1.0712, + "mean_token_accuracy": 0.6771968603134155, + "num_tokens": 30884867.0, + "step": 1205 + }, + { + "epoch": 0.13244014935207554, + "grad_norm": 2.3187942504882812, + "learning_rate": 6.613611416026345e-07, + "loss": 1.0273, + "mean_token_accuracy": 0.6858818531036377, + "num_tokens": 30910054.0, + "step": 1206 + }, + { + "epoch": 0.13254996705468922, + "grad_norm": 2.2265825271606445, + "learning_rate": 6.619099890230515e-07, + "loss": 1.0538, + "mean_token_accuracy": 0.6831690669059753, + "num_tokens": 30938338.0, + "step": 1207 + }, + { + "epoch": 0.1326597847573029, + "grad_norm": 2.167180299758911, + "learning_rate": 6.624588364434686e-07, + "loss": 1.1079, + "mean_token_accuracy": 0.6779705286026001, + "num_tokens": 30966821.0, + "step": 1208 + }, + { + "epoch": 0.13276960245991654, + "grad_norm": 2.4933037757873535, + "learning_rate": 6.630076838638858e-07, + "loss": 1.0525, + "mean_token_accuracy": 0.6869426965713501, + "num_tokens": 30990752.0, + "step": 1209 + }, + { + "epoch": 0.1328794201625302, + "grad_norm": 2.1802754402160645, + "learning_rate": 6.635565312843029e-07, + "loss": 1.0555, + "mean_token_accuracy": 0.6975319385528564, + "num_tokens": 31018790.0, + "step": 1210 + }, + { + "epoch": 0.13298923786514386, + "grad_norm": 2.045381784439087, + "learning_rate": 6.6410537870472e-07, + "loss": 1.0824, + "mean_token_accuracy": 0.6828473806381226, + "num_tokens": 31051191.0, + "step": 1211 + }, + { + "epoch": 0.13309905556775753, + "grad_norm": 2.0762126445770264, + "learning_rate": 6.646542261251372e-07, + "loss": 1.0872, + "mean_token_accuracy": 0.6789512634277344, + "num_tokens": 31083200.0, + "step": 1212 + }, + { + "epoch": 0.13320887327037118, + "grad_norm": 2.293579578399658, + "learning_rate": 6.652030735455543e-07, + "loss": 1.0053, + "mean_token_accuracy": 0.6969612240791321, + "num_tokens": 31109385.0, + "step": 1213 + }, + { + "epoch": 0.13331869097298485, + "grad_norm": 2.280874490737915, + "learning_rate": 6.657519209659714e-07, + "loss": 1.0732, + "mean_token_accuracy": 0.6865526437759399, + "num_tokens": 31135629.0, + "step": 1214 + }, + { + "epoch": 0.1334285086755985, + "grad_norm": 2.5817861557006836, + "learning_rate": 6.663007683863886e-07, + "loss": 1.0318, + "mean_token_accuracy": 0.6977328062057495, + "num_tokens": 31157325.0, + "step": 1215 + }, + { + "epoch": 0.13353832637821217, + "grad_norm": 2.558431625366211, + "learning_rate": 6.668496158068057e-07, + "loss": 1.0702, + "mean_token_accuracy": 0.6821802854537964, + "num_tokens": 31180458.0, + "step": 1216 + }, + { + "epoch": 0.13364814408082584, + "grad_norm": 2.422055244445801, + "learning_rate": 6.673984632272228e-07, + "loss": 0.9757, + "mean_token_accuracy": 0.6993889212608337, + "num_tokens": 31204332.0, + "step": 1217 + }, + { + "epoch": 0.1337579617834395, + "grad_norm": 2.406681537628174, + "learning_rate": 6.679473106476399e-07, + "loss": 0.9991, + "mean_token_accuracy": 0.7011187076568604, + "num_tokens": 31227380.0, + "step": 1218 + }, + { + "epoch": 0.13386777948605316, + "grad_norm": 2.1787428855895996, + "learning_rate": 6.684961580680571e-07, + "loss": 1.066, + "mean_token_accuracy": 0.6766520142555237, + "num_tokens": 31255910.0, + "step": 1219 + }, + { + "epoch": 0.1339775971886668, + "grad_norm": 2.1791155338287354, + "learning_rate": 6.690450054884742e-07, + "loss": 0.963, + "mean_token_accuracy": 0.7110203504562378, + "num_tokens": 31284213.0, + "step": 1220 + }, + { + "epoch": 0.13408741489128048, + "grad_norm": 2.2566425800323486, + "learning_rate": 6.695938529088913e-07, + "loss": 1.0314, + "mean_token_accuracy": 0.6899492740631104, + "num_tokens": 31310469.0, + "step": 1221 + }, + { + "epoch": 0.13419723259389413, + "grad_norm": 2.48553729057312, + "learning_rate": 6.701427003293085e-07, + "loss": 1.0361, + "mean_token_accuracy": 0.6926579475402832, + "num_tokens": 31334212.0, + "step": 1222 + }, + { + "epoch": 0.1343070502965078, + "grad_norm": 2.173583745956421, + "learning_rate": 6.706915477497256e-07, + "loss": 0.9854, + "mean_token_accuracy": 0.7012600898742676, + "num_tokens": 31364305.0, + "step": 1223 + }, + { + "epoch": 0.13441686799912145, + "grad_norm": 2.492323875427246, + "learning_rate": 6.712403951701427e-07, + "loss": 1.1344, + "mean_token_accuracy": 0.6672207713127136, + "num_tokens": 31388387.0, + "step": 1224 + }, + { + "epoch": 0.13452668570173512, + "grad_norm": 2.2663514614105225, + "learning_rate": 6.717892425905599e-07, + "loss": 1.1364, + "mean_token_accuracy": 0.6622843146324158, + "num_tokens": 31417529.0, + "step": 1225 + }, + { + "epoch": 0.13463650340434877, + "grad_norm": 2.7237212657928467, + "learning_rate": 6.723380900109769e-07, + "loss": 0.9238, + "mean_token_accuracy": 0.7260220050811768, + "num_tokens": 31436373.0, + "step": 1226 + }, + { + "epoch": 0.13474632110696244, + "grad_norm": 2.681654691696167, + "learning_rate": 6.72886937431394e-07, + "loss": 1.0524, + "mean_token_accuracy": 0.6911544799804688, + "num_tokens": 31459161.0, + "step": 1227 + }, + { + "epoch": 0.13485613880957611, + "grad_norm": 2.0639588832855225, + "learning_rate": 6.734357848518111e-07, + "loss": 1.0542, + "mean_token_accuracy": 0.6886347532272339, + "num_tokens": 31489162.0, + "step": 1228 + }, + { + "epoch": 0.13496595651218976, + "grad_norm": 2.221560478210449, + "learning_rate": 6.739846322722282e-07, + "loss": 1.0371, + "mean_token_accuracy": 0.6849765181541443, + "num_tokens": 31519012.0, + "step": 1229 + }, + { + "epoch": 0.13507577421480343, + "grad_norm": 2.2783408164978027, + "learning_rate": 6.745334796926454e-07, + "loss": 1.0497, + "mean_token_accuracy": 0.6864590644836426, + "num_tokens": 31545684.0, + "step": 1230 + }, + { + "epoch": 0.13518559191741708, + "grad_norm": 2.02730655670166, + "learning_rate": 6.750823271130625e-07, + "loss": 1.1203, + "mean_token_accuracy": 0.6656005382537842, + "num_tokens": 31577498.0, + "step": 1231 + }, + { + "epoch": 0.13529540962003075, + "grad_norm": 2.030059814453125, + "learning_rate": 6.756311745334796e-07, + "loss": 1.045, + "mean_token_accuracy": 0.6873692274093628, + "num_tokens": 31608332.0, + "step": 1232 + }, + { + "epoch": 0.1354052273226444, + "grad_norm": 2.4638831615448, + "learning_rate": 6.761800219538968e-07, + "loss": 1.1145, + "mean_token_accuracy": 0.6798506379127502, + "num_tokens": 31632609.0, + "step": 1233 + }, + { + "epoch": 0.13551504502525807, + "grad_norm": 2.1777231693267822, + "learning_rate": 6.767288693743139e-07, + "loss": 1.0726, + "mean_token_accuracy": 0.6879253387451172, + "num_tokens": 31660210.0, + "step": 1234 + }, + { + "epoch": 0.13562486272787172, + "grad_norm": 2.326463222503662, + "learning_rate": 6.77277716794731e-07, + "loss": 0.9682, + "mean_token_accuracy": 0.7131405472755432, + "num_tokens": 31685620.0, + "step": 1235 + }, + { + "epoch": 0.1357346804304854, + "grad_norm": 2.3447670936584473, + "learning_rate": 6.778265642151482e-07, + "loss": 1.0719, + "mean_token_accuracy": 0.6867733001708984, + "num_tokens": 31710603.0, + "step": 1236 + }, + { + "epoch": 0.13584449813309907, + "grad_norm": 2.1837780475616455, + "learning_rate": 6.783754116355653e-07, + "loss": 1.0753, + "mean_token_accuracy": 0.6834267377853394, + "num_tokens": 31739621.0, + "step": 1237 + }, + { + "epoch": 0.1359543158357127, + "grad_norm": 2.2774291038513184, + "learning_rate": 6.789242590559824e-07, + "loss": 1.1219, + "mean_token_accuracy": 0.6695222854614258, + "num_tokens": 31766992.0, + "step": 1238 + }, + { + "epoch": 0.1360641335383264, + "grad_norm": 2.4622340202331543, + "learning_rate": 6.794731064763996e-07, + "loss": 1.033, + "mean_token_accuracy": 0.6884952783584595, + "num_tokens": 31790562.0, + "step": 1239 + }, + { + "epoch": 0.13617395124094003, + "grad_norm": 2.4723916053771973, + "learning_rate": 6.800219538968166e-07, + "loss": 0.9901, + "mean_token_accuracy": 0.6981740593910217, + "num_tokens": 31812835.0, + "step": 1240 + }, + { + "epoch": 0.1362837689435537, + "grad_norm": 2.2014567852020264, + "learning_rate": 6.805708013172338e-07, + "loss": 0.979, + "mean_token_accuracy": 0.7019795775413513, + "num_tokens": 31840435.0, + "step": 1241 + }, + { + "epoch": 0.13639358664616735, + "grad_norm": 2.578221559524536, + "learning_rate": 6.81119648737651e-07, + "loss": 1.0317, + "mean_token_accuracy": 0.6918296813964844, + "num_tokens": 31861641.0, + "step": 1242 + }, + { + "epoch": 0.13650340434878103, + "grad_norm": 2.2955446243286133, + "learning_rate": 6.81668496158068e-07, + "loss": 1.0498, + "mean_token_accuracy": 0.6848989725112915, + "num_tokens": 31888330.0, + "step": 1243 + }, + { + "epoch": 0.13661322205139467, + "grad_norm": 2.2408699989318848, + "learning_rate": 6.822173435784852e-07, + "loss": 1.0391, + "mean_token_accuracy": 0.6901640892028809, + "num_tokens": 31914859.0, + "step": 1244 + }, + { + "epoch": 0.13672303975400835, + "grad_norm": 2.3588709831237793, + "learning_rate": 6.827661909989023e-07, + "loss": 1.02, + "mean_token_accuracy": 0.6940352320671082, + "num_tokens": 31938439.0, + "step": 1245 + }, + { + "epoch": 0.13683285745662202, + "grad_norm": 2.3310463428497314, + "learning_rate": 6.833150384193193e-07, + "loss": 1.1294, + "mean_token_accuracy": 0.6793626546859741, + "num_tokens": 31967549.0, + "step": 1246 + }, + { + "epoch": 0.13694267515923567, + "grad_norm": 2.211894989013672, + "learning_rate": 6.838638858397365e-07, + "loss": 1.0923, + "mean_token_accuracy": 0.6785785555839539, + "num_tokens": 31997107.0, + "step": 1247 + }, + { + "epoch": 0.13705249286184934, + "grad_norm": 2.874880313873291, + "learning_rate": 6.844127332601537e-07, + "loss": 0.9492, + "mean_token_accuracy": 0.7110393643379211, + "num_tokens": 32015937.0, + "step": 1248 + }, + { + "epoch": 0.13716231056446299, + "grad_norm": 2.133932113647461, + "learning_rate": 6.849615806805707e-07, + "loss": 1.151, + "mean_token_accuracy": 0.6697031855583191, + "num_tokens": 32049074.0, + "step": 1249 + }, + { + "epoch": 0.13727212826707666, + "grad_norm": 2.3087332248687744, + "learning_rate": 6.855104281009879e-07, + "loss": 1.0005, + "mean_token_accuracy": 0.6983277797698975, + "num_tokens": 32076793.0, + "step": 1250 + }, + { + "epoch": 0.1373819459696903, + "grad_norm": 2.5433554649353027, + "learning_rate": 6.86059275521405e-07, + "loss": 1.0821, + "mean_token_accuracy": 0.6712799072265625, + "num_tokens": 32099841.0, + "step": 1251 + }, + { + "epoch": 0.13749176367230398, + "grad_norm": 2.387833595275879, + "learning_rate": 6.866081229418221e-07, + "loss": 1.0704, + "mean_token_accuracy": 0.6828113794326782, + "num_tokens": 32124811.0, + "step": 1252 + }, + { + "epoch": 0.13760158137491763, + "grad_norm": 2.3087801933288574, + "learning_rate": 6.871569703622393e-07, + "loss": 0.9422, + "mean_token_accuracy": 0.7117609977722168, + "num_tokens": 32148409.0, + "step": 1253 + }, + { + "epoch": 0.1377113990775313, + "grad_norm": 2.110504388809204, + "learning_rate": 6.877058177826564e-07, + "loss": 1.0878, + "mean_token_accuracy": 0.683347761631012, + "num_tokens": 32179973.0, + "step": 1254 + }, + { + "epoch": 0.13782121678014497, + "grad_norm": 2.3281402587890625, + "learning_rate": 6.882546652030735e-07, + "loss": 0.947, + "mean_token_accuracy": 0.7177460193634033, + "num_tokens": 32206143.0, + "step": 1255 + }, + { + "epoch": 0.13793103448275862, + "grad_norm": 2.2350056171417236, + "learning_rate": 6.888035126234907e-07, + "loss": 1.0912, + "mean_token_accuracy": 0.6834779381752014, + "num_tokens": 32232991.0, + "step": 1256 + }, + { + "epoch": 0.1380408521853723, + "grad_norm": 1.9286788702011108, + "learning_rate": 6.893523600439078e-07, + "loss": 1.0479, + "mean_token_accuracy": 0.6880221366882324, + "num_tokens": 32265527.0, + "step": 1257 + }, + { + "epoch": 0.13815066988798594, + "grad_norm": 2.3424251079559326, + "learning_rate": 6.899012074643249e-07, + "loss": 1.0657, + "mean_token_accuracy": 0.6845696568489075, + "num_tokens": 32292457.0, + "step": 1258 + }, + { + "epoch": 0.1382604875905996, + "grad_norm": 2.3778605461120605, + "learning_rate": 6.904500548847421e-07, + "loss": 1.0579, + "mean_token_accuracy": 0.6866952776908875, + "num_tokens": 32317401.0, + "step": 1259 + }, + { + "epoch": 0.13837030529321326, + "grad_norm": 2.411813735961914, + "learning_rate": 6.909989023051592e-07, + "loss": 1.1154, + "mean_token_accuracy": 0.6665362119674683, + "num_tokens": 32342179.0, + "step": 1260 + }, + { + "epoch": 0.13848012299582693, + "grad_norm": 2.5779290199279785, + "learning_rate": 6.915477497255763e-07, + "loss": 1.0683, + "mean_token_accuracy": 0.6788812279701233, + "num_tokens": 32364710.0, + "step": 1261 + }, + { + "epoch": 0.13858994069844058, + "grad_norm": 2.5808117389678955, + "learning_rate": 6.920965971459934e-07, + "loss": 1.094, + "mean_token_accuracy": 0.6725931167602539, + "num_tokens": 32388842.0, + "step": 1262 + }, + { + "epoch": 0.13869975840105425, + "grad_norm": 2.219615936279297, + "learning_rate": 6.926454445664105e-07, + "loss": 1.0294, + "mean_token_accuracy": 0.6893891096115112, + "num_tokens": 32416728.0, + "step": 1263 + }, + { + "epoch": 0.1388095761036679, + "grad_norm": 2.422741651535034, + "learning_rate": 6.931942919868276e-07, + "loss": 0.9885, + "mean_token_accuracy": 0.698641300201416, + "num_tokens": 32440106.0, + "step": 1264 + }, + { + "epoch": 0.13891939380628157, + "grad_norm": 2.2732107639312744, + "learning_rate": 6.937431394072447e-07, + "loss": 1.0848, + "mean_token_accuracy": 0.6890630125999451, + "num_tokens": 32468200.0, + "step": 1265 + }, + { + "epoch": 0.13902921150889525, + "grad_norm": 2.367624521255493, + "learning_rate": 6.942919868276618e-07, + "loss": 1.0025, + "mean_token_accuracy": 0.703421950340271, + "num_tokens": 32492767.0, + "step": 1266 + }, + { + "epoch": 0.1391390292115089, + "grad_norm": 2.5698728561401367, + "learning_rate": 6.94840834248079e-07, + "loss": 1.0784, + "mean_token_accuracy": 0.6804579496383667, + "num_tokens": 32515903.0, + "step": 1267 + }, + { + "epoch": 0.13924884691412256, + "grad_norm": 2.114193916320801, + "learning_rate": 6.953896816684961e-07, + "loss": 1.0643, + "mean_token_accuracy": 0.6876977682113647, + "num_tokens": 32546230.0, + "step": 1268 + }, + { + "epoch": 0.1393586646167362, + "grad_norm": 2.3465490341186523, + "learning_rate": 6.959385290889132e-07, + "loss": 0.9202, + "mean_token_accuracy": 0.7209793329238892, + "num_tokens": 32570960.0, + "step": 1269 + }, + { + "epoch": 0.13946848231934988, + "grad_norm": 2.262593984603882, + "learning_rate": 6.964873765093304e-07, + "loss": 1.1014, + "mean_token_accuracy": 0.6741670370101929, + "num_tokens": 32600299.0, + "step": 1270 + }, + { + "epoch": 0.13957830002196353, + "grad_norm": 2.3399930000305176, + "learning_rate": 6.970362239297475e-07, + "loss": 0.9461, + "mean_token_accuracy": 0.7211796045303345, + "num_tokens": 32623018.0, + "step": 1271 + }, + { + "epoch": 0.1396881177245772, + "grad_norm": 2.3430161476135254, + "learning_rate": 6.975850713501646e-07, + "loss": 1.0933, + "mean_token_accuracy": 0.6826972961425781, + "num_tokens": 32650474.0, + "step": 1272 + }, + { + "epoch": 0.13979793542719085, + "grad_norm": 2.1600911617279053, + "learning_rate": 6.981339187705817e-07, + "loss": 1.0421, + "mean_token_accuracy": 0.6962224245071411, + "num_tokens": 32678084.0, + "step": 1273 + }, + { + "epoch": 0.13990775312980452, + "grad_norm": 1.968942403793335, + "learning_rate": 6.986827661909989e-07, + "loss": 1.1063, + "mean_token_accuracy": 0.6667652726173401, + "num_tokens": 32714757.0, + "step": 1274 + }, + { + "epoch": 0.1400175708324182, + "grad_norm": 2.582707166671753, + "learning_rate": 6.99231613611416e-07, + "loss": 1.0238, + "mean_token_accuracy": 0.697520911693573, + "num_tokens": 32735850.0, + "step": 1275 + }, + { + "epoch": 0.14012738853503184, + "grad_norm": 2.4916553497314453, + "learning_rate": 6.997804610318331e-07, + "loss": 1.0608, + "mean_token_accuracy": 0.6987147331237793, + "num_tokens": 32758541.0, + "step": 1276 + }, + { + "epoch": 0.14023720623764552, + "grad_norm": 2.2654457092285156, + "learning_rate": 7.003293084522503e-07, + "loss": 1.1385, + "mean_token_accuracy": 0.6656633615493774, + "num_tokens": 32786543.0, + "step": 1277 + }, + { + "epoch": 0.14034702394025916, + "grad_norm": 2.6803064346313477, + "learning_rate": 7.008781558726674e-07, + "loss": 1.0506, + "mean_token_accuracy": 0.6833924055099487, + "num_tokens": 32805964.0, + "step": 1278 + }, + { + "epoch": 0.14045684164287284, + "grad_norm": 2.2966949939727783, + "learning_rate": 7.014270032930845e-07, + "loss": 1.0513, + "mean_token_accuracy": 0.6912661790847778, + "num_tokens": 32831864.0, + "step": 1279 + }, + { + "epoch": 0.14056665934548648, + "grad_norm": 2.4881513118743896, + "learning_rate": 7.019758507135017e-07, + "loss": 1.0074, + "mean_token_accuracy": 0.6935230493545532, + "num_tokens": 32853860.0, + "step": 1280 + }, + { + "epoch": 0.14067647704810016, + "grad_norm": 2.297023296356201, + "learning_rate": 7.025246981339188e-07, + "loss": 0.9891, + "mean_token_accuracy": 0.7070841789245605, + "num_tokens": 32878950.0, + "step": 1281 + }, + { + "epoch": 0.1407862947507138, + "grad_norm": 2.197071075439453, + "learning_rate": 7.030735455543358e-07, + "loss": 0.9733, + "mean_token_accuracy": 0.7093228101730347, + "num_tokens": 32904121.0, + "step": 1282 + }, + { + "epoch": 0.14089611245332748, + "grad_norm": 2.449148654937744, + "learning_rate": 7.03622392974753e-07, + "loss": 1.0111, + "mean_token_accuracy": 0.7007176876068115, + "num_tokens": 32927491.0, + "step": 1283 + }, + { + "epoch": 0.14100593015594115, + "grad_norm": 2.392733573913574, + "learning_rate": 7.0417124039517e-07, + "loss": 1.019, + "mean_token_accuracy": 0.6994675397872925, + "num_tokens": 32950817.0, + "step": 1284 + }, + { + "epoch": 0.1411157478585548, + "grad_norm": 2.5570015907287598, + "learning_rate": 7.047200878155872e-07, + "loss": 1.0954, + "mean_token_accuracy": 0.6865234971046448, + "num_tokens": 32972055.0, + "step": 1285 + }, + { + "epoch": 0.14122556556116847, + "grad_norm": 2.5650391578674316, + "learning_rate": 7.052689352360044e-07, + "loss": 0.9793, + "mean_token_accuracy": 0.7008194923400879, + "num_tokens": 32994788.0, + "step": 1286 + }, + { + "epoch": 0.14133538326378212, + "grad_norm": 2.2773196697235107, + "learning_rate": 7.058177826564214e-07, + "loss": 1.0771, + "mean_token_accuracy": 0.6893271207809448, + "num_tokens": 33021155.0, + "step": 1287 + }, + { + "epoch": 0.1414452009663958, + "grad_norm": 2.394624710083008, + "learning_rate": 7.063666300768386e-07, + "loss": 1.144, + "mean_token_accuracy": 0.6617296934127808, + "num_tokens": 33050024.0, + "step": 1288 + }, + { + "epoch": 0.14155501866900944, + "grad_norm": 2.181563377380371, + "learning_rate": 7.069154774972558e-07, + "loss": 1.0894, + "mean_token_accuracy": 0.6781411170959473, + "num_tokens": 33080047.0, + "step": 1289 + }, + { + "epoch": 0.1416648363716231, + "grad_norm": 2.300414800643921, + "learning_rate": 7.074643249176728e-07, + "loss": 0.9981, + "mean_token_accuracy": 0.6983599662780762, + "num_tokens": 33104814.0, + "step": 1290 + }, + { + "epoch": 0.14177465407423676, + "grad_norm": 2.4549129009246826, + "learning_rate": 7.0801317233809e-07, + "loss": 0.9596, + "mean_token_accuracy": 0.7162258625030518, + "num_tokens": 33126328.0, + "step": 1291 + }, + { + "epoch": 0.14188447177685043, + "grad_norm": 2.5219016075134277, + "learning_rate": 7.085620197585072e-07, + "loss": 1.1067, + "mean_token_accuracy": 0.6768922805786133, + "num_tokens": 33152175.0, + "step": 1292 + }, + { + "epoch": 0.1419942894794641, + "grad_norm": 1.932201862335205, + "learning_rate": 7.091108671789242e-07, + "loss": 1.106, + "mean_token_accuracy": 0.6739040613174438, + "num_tokens": 33186300.0, + "step": 1293 + }, + { + "epoch": 0.14210410718207775, + "grad_norm": 2.2429873943328857, + "learning_rate": 7.096597145993414e-07, + "loss": 1.0135, + "mean_token_accuracy": 0.6916505098342896, + "num_tokens": 33212528.0, + "step": 1294 + }, + { + "epoch": 0.14221392488469142, + "grad_norm": 2.4889612197875977, + "learning_rate": 7.102085620197584e-07, + "loss": 1.0724, + "mean_token_accuracy": 0.6717678308486938, + "num_tokens": 33235487.0, + "step": 1295 + }, + { + "epoch": 0.14232374258730507, + "grad_norm": 2.199612617492676, + "learning_rate": 7.107574094401756e-07, + "loss": 1.0112, + "mean_token_accuracy": 0.7062577605247498, + "num_tokens": 33260361.0, + "step": 1296 + }, + { + "epoch": 0.14243356028991874, + "grad_norm": 2.3673439025878906, + "learning_rate": 7.113062568605928e-07, + "loss": 1.0154, + "mean_token_accuracy": 0.6900631189346313, + "num_tokens": 33283781.0, + "step": 1297 + }, + { + "epoch": 0.1425433779925324, + "grad_norm": 2.1483640670776367, + "learning_rate": 7.118551042810098e-07, + "loss": 1.057, + "mean_token_accuracy": 0.685052216053009, + "num_tokens": 33310624.0, + "step": 1298 + }, + { + "epoch": 0.14265319569514606, + "grad_norm": 2.3366684913635254, + "learning_rate": 7.12403951701427e-07, + "loss": 1.1407, + "mean_token_accuracy": 0.6620305776596069, + "num_tokens": 33336325.0, + "step": 1299 + }, + { + "epoch": 0.1427630133977597, + "grad_norm": 2.249648094177246, + "learning_rate": 7.129527991218442e-07, + "loss": 1.0424, + "mean_token_accuracy": 0.6851078867912292, + "num_tokens": 33361714.0, + "step": 1300 + }, + { + "epoch": 0.14287283110037338, + "grad_norm": 2.3001108169555664, + "learning_rate": 7.135016465422611e-07, + "loss": 1.0373, + "mean_token_accuracy": 0.7026200294494629, + "num_tokens": 33387069.0, + "step": 1301 + }, + { + "epoch": 0.14298264880298703, + "grad_norm": 2.326556444168091, + "learning_rate": 7.140504939626783e-07, + "loss": 0.9107, + "mean_token_accuracy": 0.7232853770256042, + "num_tokens": 33410806.0, + "step": 1302 + }, + { + "epoch": 0.1430924665056007, + "grad_norm": 2.1379594802856445, + "learning_rate": 7.145993413830955e-07, + "loss": 1.03, + "mean_token_accuracy": 0.6981021761894226, + "num_tokens": 33439584.0, + "step": 1303 + }, + { + "epoch": 0.14320228420821438, + "grad_norm": 2.136491537094116, + "learning_rate": 7.151481888035125e-07, + "loss": 0.9977, + "mean_token_accuracy": 0.7014040946960449, + "num_tokens": 33468178.0, + "step": 1304 + }, + { + "epoch": 0.14331210191082802, + "grad_norm": 2.2695746421813965, + "learning_rate": 7.156970362239297e-07, + "loss": 0.9859, + "mean_token_accuracy": 0.7056534886360168, + "num_tokens": 33492780.0, + "step": 1305 + }, + { + "epoch": 0.1434219196134417, + "grad_norm": 1.9874473810195923, + "learning_rate": 7.162458836443468e-07, + "loss": 1.0766, + "mean_token_accuracy": 0.6854550838470459, + "num_tokens": 33526258.0, + "step": 1306 + }, + { + "epoch": 0.14353173731605534, + "grad_norm": 2.3314476013183594, + "learning_rate": 7.167947310647639e-07, + "loss": 0.9743, + "mean_token_accuracy": 0.7279733419418335, + "num_tokens": 33550227.0, + "step": 1307 + }, + { + "epoch": 0.14364155501866901, + "grad_norm": 2.0030055046081543, + "learning_rate": 7.173435784851811e-07, + "loss": 1.0969, + "mean_token_accuracy": 0.6792635917663574, + "num_tokens": 33581461.0, + "step": 1308 + }, + { + "epoch": 0.14375137272128266, + "grad_norm": 2.2964425086975098, + "learning_rate": 7.178924259055982e-07, + "loss": 1.057, + "mean_token_accuracy": 0.6846024394035339, + "num_tokens": 33607027.0, + "step": 1309 + }, + { + "epoch": 0.14386119042389633, + "grad_norm": 2.230947971343994, + "learning_rate": 7.184412733260153e-07, + "loss": 0.9182, + "mean_token_accuracy": 0.7230355739593506, + "num_tokens": 33634935.0, + "step": 1310 + }, + { + "epoch": 0.14397100812650998, + "grad_norm": 2.32314395904541, + "learning_rate": 7.189901207464325e-07, + "loss": 1.0359, + "mean_token_accuracy": 0.6896568536758423, + "num_tokens": 33661177.0, + "step": 1311 + }, + { + "epoch": 0.14408082582912365, + "grad_norm": 2.2770159244537354, + "learning_rate": 7.195389681668496e-07, + "loss": 1.0697, + "mean_token_accuracy": 0.6804457902908325, + "num_tokens": 33685976.0, + "step": 1312 + }, + { + "epoch": 0.14419064353173733, + "grad_norm": 2.6060791015625, + "learning_rate": 7.200878155872667e-07, + "loss": 0.9096, + "mean_token_accuracy": 0.7217901349067688, + "num_tokens": 33704123.0, + "step": 1313 + }, + { + "epoch": 0.14430046123435097, + "grad_norm": 2.409308433532715, + "learning_rate": 7.206366630076838e-07, + "loss": 1.0539, + "mean_token_accuracy": 0.6854755282402039, + "num_tokens": 33727727.0, + "step": 1314 + }, + { + "epoch": 0.14441027893696465, + "grad_norm": 2.362048625946045, + "learning_rate": 7.21185510428101e-07, + "loss": 1.0209, + "mean_token_accuracy": 0.6966896653175354, + "num_tokens": 33754844.0, + "step": 1315 + }, + { + "epoch": 0.1445200966395783, + "grad_norm": 2.0281424522399902, + "learning_rate": 7.217343578485181e-07, + "loss": 0.9951, + "mean_token_accuracy": 0.7024397253990173, + "num_tokens": 33785644.0, + "step": 1316 + }, + { + "epoch": 0.14462991434219197, + "grad_norm": 2.2028141021728516, + "learning_rate": 7.222832052689352e-07, + "loss": 1.0024, + "mean_token_accuracy": 0.7056156396865845, + "num_tokens": 33812440.0, + "step": 1317 + }, + { + "epoch": 0.1447397320448056, + "grad_norm": 2.3704617023468018, + "learning_rate": 7.228320526893524e-07, + "loss": 1.0487, + "mean_token_accuracy": 0.6965635418891907, + "num_tokens": 33837409.0, + "step": 1318 + }, + { + "epoch": 0.1448495497474193, + "grad_norm": 2.0866079330444336, + "learning_rate": 7.233809001097695e-07, + "loss": 1.0728, + "mean_token_accuracy": 0.6805024743080139, + "num_tokens": 33867282.0, + "step": 1319 + }, + { + "epoch": 0.14495936745003293, + "grad_norm": 2.301236629486084, + "learning_rate": 7.239297475301865e-07, + "loss": 1.1187, + "mean_token_accuracy": 0.6703392267227173, + "num_tokens": 33892774.0, + "step": 1320 + }, + { + "epoch": 0.1450691851526466, + "grad_norm": 1.8548765182495117, + "learning_rate": 7.244785949506037e-07, + "loss": 1.1335, + "mean_token_accuracy": 0.667043924331665, + "num_tokens": 33930487.0, + "step": 1321 + }, + { + "epoch": 0.14517900285526028, + "grad_norm": 2.5848207473754883, + "learning_rate": 7.250274423710208e-07, + "loss": 1.1075, + "mean_token_accuracy": 0.6765713691711426, + "num_tokens": 33953725.0, + "step": 1322 + }, + { + "epoch": 0.14528882055787393, + "grad_norm": 2.239722967147827, + "learning_rate": 7.255762897914379e-07, + "loss": 1.0658, + "mean_token_accuracy": 0.6851575374603271, + "num_tokens": 33980517.0, + "step": 1323 + }, + { + "epoch": 0.1453986382604876, + "grad_norm": 1.9996733665466309, + "learning_rate": 7.261251372118551e-07, + "loss": 1.1172, + "mean_token_accuracy": 0.6793633699417114, + "num_tokens": 34014194.0, + "step": 1324 + }, + { + "epoch": 0.14550845596310125, + "grad_norm": 2.2542037963867188, + "learning_rate": 7.266739846322721e-07, + "loss": 1.1096, + "mean_token_accuracy": 0.6669776439666748, + "num_tokens": 34041936.0, + "step": 1325 + }, + { + "epoch": 0.14561827366571492, + "grad_norm": 2.3681392669677734, + "learning_rate": 7.272228320526893e-07, + "loss": 0.8964, + "mean_token_accuracy": 0.7260633111000061, + "num_tokens": 34065564.0, + "step": 1326 + }, + { + "epoch": 0.14572809136832857, + "grad_norm": 2.505321741104126, + "learning_rate": 7.277716794731065e-07, + "loss": 0.9762, + "mean_token_accuracy": 0.7037985324859619, + "num_tokens": 34086954.0, + "step": 1327 + }, + { + "epoch": 0.14583790907094224, + "grad_norm": 2.174700975418091, + "learning_rate": 7.283205268935235e-07, + "loss": 1.0106, + "mean_token_accuracy": 0.696319580078125, + "num_tokens": 34113772.0, + "step": 1328 + }, + { + "epoch": 0.14594772677355589, + "grad_norm": 2.50898814201355, + "learning_rate": 7.288693743139407e-07, + "loss": 0.9978, + "mean_token_accuracy": 0.6957257390022278, + "num_tokens": 34135837.0, + "step": 1329 + }, + { + "epoch": 0.14605754447616956, + "grad_norm": 2.3737375736236572, + "learning_rate": 7.294182217343579e-07, + "loss": 0.9962, + "mean_token_accuracy": 0.6961583495140076, + "num_tokens": 34159439.0, + "step": 1330 + }, + { + "epoch": 0.14616736217878323, + "grad_norm": 2.2161667346954346, + "learning_rate": 7.299670691547749e-07, + "loss": 1.0857, + "mean_token_accuracy": 0.6839134097099304, + "num_tokens": 34185123.0, + "step": 1331 + }, + { + "epoch": 0.14627717988139688, + "grad_norm": 2.516294002532959, + "learning_rate": 7.305159165751921e-07, + "loss": 1.0163, + "mean_token_accuracy": 0.6973220109939575, + "num_tokens": 34207803.0, + "step": 1332 + }, + { + "epoch": 0.14638699758401055, + "grad_norm": 2.333418130874634, + "learning_rate": 7.310647639956093e-07, + "loss": 1.0964, + "mean_token_accuracy": 0.6754635572433472, + "num_tokens": 34235086.0, + "step": 1333 + }, + { + "epoch": 0.1464968152866242, + "grad_norm": 2.4046716690063477, + "learning_rate": 7.316136114160263e-07, + "loss": 1.0619, + "mean_token_accuracy": 0.6842087507247925, + "num_tokens": 34259617.0, + "step": 1334 + }, + { + "epoch": 0.14660663298923787, + "grad_norm": 2.1285181045532227, + "learning_rate": 7.321624588364435e-07, + "loss": 0.992, + "mean_token_accuracy": 0.7043927907943726, + "num_tokens": 34287427.0, + "step": 1335 + }, + { + "epoch": 0.14671645069185152, + "grad_norm": 2.1970417499542236, + "learning_rate": 7.327113062568605e-07, + "loss": 1.1346, + "mean_token_accuracy": 0.6679275631904602, + "num_tokens": 34315740.0, + "step": 1336 + }, + { + "epoch": 0.1468262683944652, + "grad_norm": 2.646646738052368, + "learning_rate": 7.332601536772777e-07, + "loss": 1.1024, + "mean_token_accuracy": 0.6809260845184326, + "num_tokens": 34338113.0, + "step": 1337 + }, + { + "epoch": 0.14693608609707884, + "grad_norm": 2.2833411693573, + "learning_rate": 7.338090010976949e-07, + "loss": 1.038, + "mean_token_accuracy": 0.6907508969306946, + "num_tokens": 34364122.0, + "step": 1338 + }, + { + "epoch": 0.1470459037996925, + "grad_norm": 2.3605122566223145, + "learning_rate": 7.343578485181118e-07, + "loss": 1.088, + "mean_token_accuracy": 0.6782320737838745, + "num_tokens": 34388611.0, + "step": 1339 + }, + { + "epoch": 0.14715572150230616, + "grad_norm": 2.5310590267181396, + "learning_rate": 7.34906695938529e-07, + "loss": 1.0764, + "mean_token_accuracy": 0.6754786968231201, + "num_tokens": 34409127.0, + "step": 1340 + }, + { + "epoch": 0.14726553920491983, + "grad_norm": 2.3131251335144043, + "learning_rate": 7.354555433589462e-07, + "loss": 0.9304, + "mean_token_accuracy": 0.7123652100563049, + "num_tokens": 34431061.0, + "step": 1341 + }, + { + "epoch": 0.1473753569075335, + "grad_norm": 2.218961477279663, + "learning_rate": 7.360043907793632e-07, + "loss": 1.0521, + "mean_token_accuracy": 0.6840171813964844, + "num_tokens": 34457075.0, + "step": 1342 + }, + { + "epoch": 0.14748517461014715, + "grad_norm": 2.2338786125183105, + "learning_rate": 7.365532381997804e-07, + "loss": 1.0994, + "mean_token_accuracy": 0.6685579419136047, + "num_tokens": 34485297.0, + "step": 1343 + }, + { + "epoch": 0.14759499231276083, + "grad_norm": 2.4365503787994385, + "learning_rate": 7.371020856201976e-07, + "loss": 0.9301, + "mean_token_accuracy": 0.7105551958084106, + "num_tokens": 34506887.0, + "step": 1344 + }, + { + "epoch": 0.14770481001537447, + "grad_norm": 2.236522674560547, + "learning_rate": 7.376509330406146e-07, + "loss": 1.0135, + "mean_token_accuracy": 0.7013297080993652, + "num_tokens": 34533920.0, + "step": 1345 + }, + { + "epoch": 0.14781462771798815, + "grad_norm": 2.3462302684783936, + "learning_rate": 7.381997804610318e-07, + "loss": 1.0808, + "mean_token_accuracy": 0.6882693767547607, + "num_tokens": 34558303.0, + "step": 1346 + }, + { + "epoch": 0.1479244454206018, + "grad_norm": 2.3564248085021973, + "learning_rate": 7.387486278814489e-07, + "loss": 1.0451, + "mean_token_accuracy": 0.6892306804656982, + "num_tokens": 34582613.0, + "step": 1347 + }, + { + "epoch": 0.14803426312321546, + "grad_norm": 2.369875907897949, + "learning_rate": 7.39297475301866e-07, + "loss": 1.0949, + "mean_token_accuracy": 0.6747181415557861, + "num_tokens": 34607605.0, + "step": 1348 + }, + { + "epoch": 0.1481440808258291, + "grad_norm": 2.1856024265289307, + "learning_rate": 7.398463227222832e-07, + "loss": 1.0573, + "mean_token_accuracy": 0.6845035552978516, + "num_tokens": 34635959.0, + "step": 1349 + }, + { + "epoch": 0.14825389852844278, + "grad_norm": 2.5394091606140137, + "learning_rate": 7.403951701427003e-07, + "loss": 1.0149, + "mean_token_accuracy": 0.6925002932548523, + "num_tokens": 34658172.0, + "step": 1350 + }, + { + "epoch": 0.14836371623105646, + "grad_norm": 2.1801517009735107, + "learning_rate": 7.409440175631174e-07, + "loss": 1.0848, + "mean_token_accuracy": 0.6811391115188599, + "num_tokens": 34686809.0, + "step": 1351 + }, + { + "epoch": 0.1484735339336701, + "grad_norm": 2.5486392974853516, + "learning_rate": 7.414928649835346e-07, + "loss": 0.9889, + "mean_token_accuracy": 0.6999415159225464, + "num_tokens": 34706933.0, + "step": 1352 + }, + { + "epoch": 0.14858335163628378, + "grad_norm": 2.174381971359253, + "learning_rate": 7.420417124039517e-07, + "loss": 1.0069, + "mean_token_accuracy": 0.6929332613945007, + "num_tokens": 34734608.0, + "step": 1353 + }, + { + "epoch": 0.14869316933889742, + "grad_norm": 2.2581939697265625, + "learning_rate": 7.425905598243688e-07, + "loss": 0.9532, + "mean_token_accuracy": 0.710367739200592, + "num_tokens": 34758887.0, + "step": 1354 + }, + { + "epoch": 0.1488029870415111, + "grad_norm": 2.4077060222625732, + "learning_rate": 7.43139407244786e-07, + "loss": 1.0881, + "mean_token_accuracy": 0.6788390874862671, + "num_tokens": 34783555.0, + "step": 1355 + }, + { + "epoch": 0.14891280474412474, + "grad_norm": 2.4096968173980713, + "learning_rate": 7.436882546652031e-07, + "loss": 0.9979, + "mean_token_accuracy": 0.6993991136550903, + "num_tokens": 34808035.0, + "step": 1356 + }, + { + "epoch": 0.14902262244673842, + "grad_norm": 2.1786375045776367, + "learning_rate": 7.442371020856202e-07, + "loss": 1.0279, + "mean_token_accuracy": 0.6917158365249634, + "num_tokens": 34836934.0, + "step": 1357 + }, + { + "epoch": 0.14913244014935206, + "grad_norm": 2.4386043548583984, + "learning_rate": 7.447859495060372e-07, + "loss": 1.0591, + "mean_token_accuracy": 0.6893469095230103, + "num_tokens": 34859748.0, + "step": 1358 + }, + { + "epoch": 0.14924225785196574, + "grad_norm": 2.1566762924194336, + "learning_rate": 7.453347969264544e-07, + "loss": 0.9972, + "mean_token_accuracy": 0.6941355466842651, + "num_tokens": 34886281.0, + "step": 1359 + }, + { + "epoch": 0.1493520755545794, + "grad_norm": 2.356282949447632, + "learning_rate": 7.458836443468715e-07, + "loss": 1.0229, + "mean_token_accuracy": 0.6943450570106506, + "num_tokens": 34911314.0, + "step": 1360 + }, + { + "epoch": 0.14946189325719306, + "grad_norm": 2.1993863582611084, + "learning_rate": 7.464324917672886e-07, + "loss": 1.0404, + "mean_token_accuracy": 0.6853691339492798, + "num_tokens": 34940521.0, + "step": 1361 + }, + { + "epoch": 0.14957171095980673, + "grad_norm": 2.372764825820923, + "learning_rate": 7.469813391877058e-07, + "loss": 1.0716, + "mean_token_accuracy": 0.6784408092498779, + "num_tokens": 34966674.0, + "step": 1362 + }, + { + "epoch": 0.14968152866242038, + "grad_norm": 2.0950355529785156, + "learning_rate": 7.475301866081229e-07, + "loss": 1.0929, + "mean_token_accuracy": 0.6749966740608215, + "num_tokens": 34996448.0, + "step": 1363 + }, + { + "epoch": 0.14979134636503405, + "grad_norm": 2.1505796909332275, + "learning_rate": 7.4807903402854e-07, + "loss": 0.9831, + "mean_token_accuracy": 0.7036553025245667, + "num_tokens": 35025510.0, + "step": 1364 + }, + { + "epoch": 0.1499011640676477, + "grad_norm": 2.0684943199157715, + "learning_rate": 7.486278814489572e-07, + "loss": 1.0514, + "mean_token_accuracy": 0.6838526725769043, + "num_tokens": 35056049.0, + "step": 1365 + }, + { + "epoch": 0.15001098177026137, + "grad_norm": 2.1742451190948486, + "learning_rate": 7.491767288693743e-07, + "loss": 1.1185, + "mean_token_accuracy": 0.6791344285011292, + "num_tokens": 35085201.0, + "step": 1366 + }, + { + "epoch": 0.15012079947287502, + "grad_norm": 2.4556140899658203, + "learning_rate": 7.497255762897914e-07, + "loss": 0.9764, + "mean_token_accuracy": 0.7056015729904175, + "num_tokens": 35108225.0, + "step": 1367 + }, + { + "epoch": 0.1502306171754887, + "grad_norm": 2.223243474960327, + "learning_rate": 7.502744237102086e-07, + "loss": 1.0178, + "mean_token_accuracy": 0.6997281908988953, + "num_tokens": 35136030.0, + "step": 1368 + }, + { + "epoch": 0.15034043487810236, + "grad_norm": 2.411648988723755, + "learning_rate": 7.508232711306256e-07, + "loss": 1.0503, + "mean_token_accuracy": 0.685016930103302, + "num_tokens": 35159446.0, + "step": 1369 + }, + { + "epoch": 0.150450252580716, + "grad_norm": 2.3581044673919678, + "learning_rate": 7.513721185510428e-07, + "loss": 1.0714, + "mean_token_accuracy": 0.6791707277297974, + "num_tokens": 35186943.0, + "step": 1370 + }, + { + "epoch": 0.15056007028332968, + "grad_norm": 2.3548622131347656, + "learning_rate": 7.5192096597146e-07, + "loss": 1.0068, + "mean_token_accuracy": 0.702694833278656, + "num_tokens": 35211560.0, + "step": 1371 + }, + { + "epoch": 0.15066988798594333, + "grad_norm": 2.3850247859954834, + "learning_rate": 7.52469813391877e-07, + "loss": 1.0462, + "mean_token_accuracy": 0.6900955438613892, + "num_tokens": 35235314.0, + "step": 1372 + }, + { + "epoch": 0.150779705688557, + "grad_norm": 2.108672857284546, + "learning_rate": 7.530186608122942e-07, + "loss": 1.1681, + "mean_token_accuracy": 0.661410927772522, + "num_tokens": 35267173.0, + "step": 1373 + }, + { + "epoch": 0.15088952339117065, + "grad_norm": 2.665987253189087, + "learning_rate": 7.535675082327113e-07, + "loss": 1.0377, + "mean_token_accuracy": 0.6887472867965698, + "num_tokens": 35288283.0, + "step": 1374 + }, + { + "epoch": 0.15099934109378432, + "grad_norm": 2.4454782009124756, + "learning_rate": 7.541163556531284e-07, + "loss": 0.9994, + "mean_token_accuracy": 0.7019792795181274, + "num_tokens": 35311014.0, + "step": 1375 + }, + { + "epoch": 0.15110915879639797, + "grad_norm": 2.4239468574523926, + "learning_rate": 7.546652030735456e-07, + "loss": 1.1085, + "mean_token_accuracy": 0.6874197721481323, + "num_tokens": 35335156.0, + "step": 1376 + }, + { + "epoch": 0.15121897649901164, + "grad_norm": 2.327040672302246, + "learning_rate": 7.552140504939627e-07, + "loss": 0.9746, + "mean_token_accuracy": 0.702420711517334, + "num_tokens": 35358750.0, + "step": 1377 + }, + { + "epoch": 0.1513287942016253, + "grad_norm": 2.193791627883911, + "learning_rate": 7.557628979143797e-07, + "loss": 1.0815, + "mean_token_accuracy": 0.6830390691757202, + "num_tokens": 35385047.0, + "step": 1378 + }, + { + "epoch": 0.15143861190423896, + "grad_norm": 2.4382519721984863, + "learning_rate": 7.563117453347969e-07, + "loss": 1.0719, + "mean_token_accuracy": 0.6830697059631348, + "num_tokens": 35408335.0, + "step": 1379 + }, + { + "epoch": 0.15154842960685264, + "grad_norm": 2.1339311599731445, + "learning_rate": 7.568605927552139e-07, + "loss": 1.052, + "mean_token_accuracy": 0.6863256096839905, + "num_tokens": 35435959.0, + "step": 1380 + }, + { + "epoch": 0.15165824730946628, + "grad_norm": 2.415844202041626, + "learning_rate": 7.574094401756311e-07, + "loss": 0.9725, + "mean_token_accuracy": 0.7064295411109924, + "num_tokens": 35459715.0, + "step": 1381 + }, + { + "epoch": 0.15176806501207996, + "grad_norm": 2.348931074142456, + "learning_rate": 7.579582875960483e-07, + "loss": 1.1279, + "mean_token_accuracy": 0.6608187556266785, + "num_tokens": 35484064.0, + "step": 1382 + }, + { + "epoch": 0.1518778827146936, + "grad_norm": 2.2701609134674072, + "learning_rate": 7.585071350164653e-07, + "loss": 1.0806, + "mean_token_accuracy": 0.6767642498016357, + "num_tokens": 35511856.0, + "step": 1383 + }, + { + "epoch": 0.15198770041730728, + "grad_norm": 2.2294414043426514, + "learning_rate": 7.590559824368825e-07, + "loss": 0.9669, + "mean_token_accuracy": 0.7104941606521606, + "num_tokens": 35536386.0, + "step": 1384 + }, + { + "epoch": 0.15209751811992092, + "grad_norm": 2.1097705364227295, + "learning_rate": 7.596048298572997e-07, + "loss": 1.1155, + "mean_token_accuracy": 0.6689174175262451, + "num_tokens": 35565993.0, + "step": 1385 + }, + { + "epoch": 0.1522073358225346, + "grad_norm": 2.420114517211914, + "learning_rate": 7.601536772777167e-07, + "loss": 1.0049, + "mean_token_accuracy": 0.6978819370269775, + "num_tokens": 35591338.0, + "step": 1386 + }, + { + "epoch": 0.15231715352514824, + "grad_norm": 2.0195930004119873, + "learning_rate": 7.607025246981339e-07, + "loss": 1.0872, + "mean_token_accuracy": 0.6786079406738281, + "num_tokens": 35622557.0, + "step": 1387 + }, + { + "epoch": 0.15242697122776191, + "grad_norm": 2.1986985206604004, + "learning_rate": 7.612513721185511e-07, + "loss": 1.0121, + "mean_token_accuracy": 0.6966283321380615, + "num_tokens": 35649096.0, + "step": 1388 + }, + { + "epoch": 0.1525367889303756, + "grad_norm": 2.3882856369018555, + "learning_rate": 7.618002195389681e-07, + "loss": 1.1029, + "mean_token_accuracy": 0.6729844808578491, + "num_tokens": 35675523.0, + "step": 1389 + }, + { + "epoch": 0.15264660663298923, + "grad_norm": 2.282721996307373, + "learning_rate": 7.623490669593853e-07, + "loss": 1.0537, + "mean_token_accuracy": 0.6870706081390381, + "num_tokens": 35703371.0, + "step": 1390 + }, + { + "epoch": 0.1527564243356029, + "grad_norm": 2.4693355560302734, + "learning_rate": 7.628979143798024e-07, + "loss": 1.0212, + "mean_token_accuracy": 0.6918149590492249, + "num_tokens": 35726880.0, + "step": 1391 + }, + { + "epoch": 0.15286624203821655, + "grad_norm": 2.374406337738037, + "learning_rate": 7.634467618002195e-07, + "loss": 1.0554, + "mean_token_accuracy": 0.686493992805481, + "num_tokens": 35752556.0, + "step": 1392 + }, + { + "epoch": 0.15297605974083023, + "grad_norm": 2.411684274673462, + "learning_rate": 7.639956092206367e-07, + "loss": 1.0546, + "mean_token_accuracy": 0.6856253147125244, + "num_tokens": 35776494.0, + "step": 1393 + }, + { + "epoch": 0.15308587744344387, + "grad_norm": 2.3408043384552, + "learning_rate": 7.645444566410538e-07, + "loss": 1.095, + "mean_token_accuracy": 0.6728119850158691, + "num_tokens": 35801923.0, + "step": 1394 + }, + { + "epoch": 0.15319569514605755, + "grad_norm": 2.191861152648926, + "learning_rate": 7.650933040614709e-07, + "loss": 0.9379, + "mean_token_accuracy": 0.7277936935424805, + "num_tokens": 35827958.0, + "step": 1395 + }, + { + "epoch": 0.1533055128486712, + "grad_norm": 2.178156852722168, + "learning_rate": 7.656421514818881e-07, + "loss": 1.0499, + "mean_token_accuracy": 0.6893376111984253, + "num_tokens": 35856944.0, + "step": 1396 + }, + { + "epoch": 0.15341533055128487, + "grad_norm": 2.3180387020111084, + "learning_rate": 7.66190998902305e-07, + "loss": 1.0178, + "mean_token_accuracy": 0.7008300423622131, + "num_tokens": 35883350.0, + "step": 1397 + }, + { + "epoch": 0.15352514825389854, + "grad_norm": 1.947755217552185, + "learning_rate": 7.667398463227222e-07, + "loss": 1.0489, + "mean_token_accuracy": 0.6844663619995117, + "num_tokens": 35916815.0, + "step": 1398 + }, + { + "epoch": 0.1536349659565122, + "grad_norm": 2.192426919937134, + "learning_rate": 7.672886937431394e-07, + "loss": 1.1281, + "mean_token_accuracy": 0.6652870774269104, + "num_tokens": 35945810.0, + "step": 1399 + }, + { + "epoch": 0.15374478365912586, + "grad_norm": 2.523332118988037, + "learning_rate": 7.678375411635564e-07, + "loss": 1.0428, + "mean_token_accuracy": 0.6898975372314453, + "num_tokens": 35967840.0, + "step": 1400 + }, + { + "epoch": 0.1538546013617395, + "grad_norm": 2.422727584838867, + "learning_rate": 7.683863885839736e-07, + "loss": 0.9999, + "mean_token_accuracy": 0.7025365829467773, + "num_tokens": 35990175.0, + "step": 1401 + }, + { + "epoch": 0.15396441906435318, + "grad_norm": 2.113978147506714, + "learning_rate": 7.689352360043907e-07, + "loss": 1.0459, + "mean_token_accuracy": 0.6836882829666138, + "num_tokens": 36019016.0, + "step": 1402 + }, + { + "epoch": 0.15407423676696683, + "grad_norm": 2.2963268756866455, + "learning_rate": 7.694840834248078e-07, + "loss": 1.0982, + "mean_token_accuracy": 0.6864619255065918, + "num_tokens": 36044536.0, + "step": 1403 + }, + { + "epoch": 0.1541840544695805, + "grad_norm": 2.4239652156829834, + "learning_rate": 7.70032930845225e-07, + "loss": 1.106, + "mean_token_accuracy": 0.6705945730209351, + "num_tokens": 36067712.0, + "step": 1404 + }, + { + "epoch": 0.15429387217219415, + "grad_norm": 2.1826915740966797, + "learning_rate": 7.705817782656421e-07, + "loss": 0.9666, + "mean_token_accuracy": 0.7078818082809448, + "num_tokens": 36093842.0, + "step": 1405 + }, + { + "epoch": 0.15440368987480782, + "grad_norm": 2.435220718383789, + "learning_rate": 7.711306256860592e-07, + "loss": 1.0399, + "mean_token_accuracy": 0.6905511617660522, + "num_tokens": 36116329.0, + "step": 1406 + }, + { + "epoch": 0.1545135075774215, + "grad_norm": 2.731247663497925, + "learning_rate": 7.716794731064764e-07, + "loss": 1.0402, + "mean_token_accuracy": 0.6861788034439087, + "num_tokens": 36137566.0, + "step": 1407 + }, + { + "epoch": 0.15462332528003514, + "grad_norm": 2.5706493854522705, + "learning_rate": 7.722283205268935e-07, + "loss": 1.0144, + "mean_token_accuracy": 0.6999014019966125, + "num_tokens": 36159934.0, + "step": 1408 + }, + { + "epoch": 0.1547331429826488, + "grad_norm": 2.415222406387329, + "learning_rate": 7.727771679473106e-07, + "loss": 1.0385, + "mean_token_accuracy": 0.6886651515960693, + "num_tokens": 36184254.0, + "step": 1409 + }, + { + "epoch": 0.15484296068526246, + "grad_norm": 2.3050169944763184, + "learning_rate": 7.733260153677278e-07, + "loss": 1.114, + "mean_token_accuracy": 0.678180456161499, + "num_tokens": 36211356.0, + "step": 1410 + }, + { + "epoch": 0.15495277838787613, + "grad_norm": 2.426565170288086, + "learning_rate": 7.738748627881449e-07, + "loss": 1.0733, + "mean_token_accuracy": 0.6884408593177795, + "num_tokens": 36235906.0, + "step": 1411 + }, + { + "epoch": 0.15506259609048978, + "grad_norm": 2.3179662227630615, + "learning_rate": 7.74423710208562e-07, + "loss": 1.0254, + "mean_token_accuracy": 0.7036166191101074, + "num_tokens": 36261995.0, + "step": 1412 + }, + { + "epoch": 0.15517241379310345, + "grad_norm": 2.1950159072875977, + "learning_rate": 7.749725576289791e-07, + "loss": 1.0625, + "mean_token_accuracy": 0.6793489456176758, + "num_tokens": 36288315.0, + "step": 1413 + }, + { + "epoch": 0.1552822314957171, + "grad_norm": 2.4194746017456055, + "learning_rate": 7.755214050493963e-07, + "loss": 1.0482, + "mean_token_accuracy": 0.6876755356788635, + "num_tokens": 36310184.0, + "step": 1414 + }, + { + "epoch": 0.15539204919833077, + "grad_norm": 2.1745238304138184, + "learning_rate": 7.760702524698134e-07, + "loss": 0.9419, + "mean_token_accuracy": 0.7133064270019531, + "num_tokens": 36335340.0, + "step": 1415 + }, + { + "epoch": 0.15550186690094442, + "grad_norm": 2.2556207180023193, + "learning_rate": 7.766190998902304e-07, + "loss": 1.0039, + "mean_token_accuracy": 0.705350399017334, + "num_tokens": 36360635.0, + "step": 1416 + }, + { + "epoch": 0.1556116846035581, + "grad_norm": 2.2613823413848877, + "learning_rate": 7.771679473106476e-07, + "loss": 0.9746, + "mean_token_accuracy": 0.7096534967422485, + "num_tokens": 36384346.0, + "step": 1417 + }, + { + "epoch": 0.15572150230617177, + "grad_norm": 2.232762098312378, + "learning_rate": 7.777167947310647e-07, + "loss": 1.0754, + "mean_token_accuracy": 0.688490092754364, + "num_tokens": 36411226.0, + "step": 1418 + }, + { + "epoch": 0.1558313200087854, + "grad_norm": 2.2246830463409424, + "learning_rate": 7.782656421514818e-07, + "loss": 1.0257, + "mean_token_accuracy": 0.6994016170501709, + "num_tokens": 36437160.0, + "step": 1419 + }, + { + "epoch": 0.15594113771139909, + "grad_norm": 2.0272367000579834, + "learning_rate": 7.78814489571899e-07, + "loss": 1.11, + "mean_token_accuracy": 0.6722016930580139, + "num_tokens": 36471867.0, + "step": 1420 + }, + { + "epoch": 0.15605095541401273, + "grad_norm": 2.149820327758789, + "learning_rate": 7.793633369923161e-07, + "loss": 1.0754, + "mean_token_accuracy": 0.6789872050285339, + "num_tokens": 36501370.0, + "step": 1421 + }, + { + "epoch": 0.1561607731166264, + "grad_norm": 2.5396854877471924, + "learning_rate": 7.799121844127332e-07, + "loss": 1.1314, + "mean_token_accuracy": 0.665706992149353, + "num_tokens": 36524391.0, + "step": 1422 + }, + { + "epoch": 0.15627059081924005, + "grad_norm": 2.186183214187622, + "learning_rate": 7.804610318331504e-07, + "loss": 1.1367, + "mean_token_accuracy": 0.6620551347732544, + "num_tokens": 36554288.0, + "step": 1423 + }, + { + "epoch": 0.15638040852185373, + "grad_norm": 2.360835313796997, + "learning_rate": 7.810098792535674e-07, + "loss": 1.0227, + "mean_token_accuracy": 0.6941133737564087, + "num_tokens": 36578331.0, + "step": 1424 + }, + { + "epoch": 0.15649022622446737, + "grad_norm": 2.3727662563323975, + "learning_rate": 7.815587266739846e-07, + "loss": 1.0343, + "mean_token_accuracy": 0.6969127058982849, + "num_tokens": 36602449.0, + "step": 1425 + }, + { + "epoch": 0.15660004392708105, + "grad_norm": 2.252835273742676, + "learning_rate": 7.821075740944018e-07, + "loss": 1.0702, + "mean_token_accuracy": 0.6893181800842285, + "num_tokens": 36629983.0, + "step": 1426 + }, + { + "epoch": 0.15670986162969472, + "grad_norm": 2.0446155071258545, + "learning_rate": 7.826564215148188e-07, + "loss": 1.0911, + "mean_token_accuracy": 0.6811555027961731, + "num_tokens": 36662613.0, + "step": 1427 + }, + { + "epoch": 0.15681967933230836, + "grad_norm": 2.368311882019043, + "learning_rate": 7.83205268935236e-07, + "loss": 1.0797, + "mean_token_accuracy": 0.6838181614875793, + "num_tokens": 36686530.0, + "step": 1428 + }, + { + "epoch": 0.15692949703492204, + "grad_norm": 2.1679441928863525, + "learning_rate": 7.837541163556532e-07, + "loss": 1.0068, + "mean_token_accuracy": 0.6967319250106812, + "num_tokens": 36714017.0, + "step": 1429 + }, + { + "epoch": 0.15703931473753568, + "grad_norm": 2.1788761615753174, + "learning_rate": 7.843029637760702e-07, + "loss": 0.9735, + "mean_token_accuracy": 0.7176002264022827, + "num_tokens": 36740290.0, + "step": 1430 + }, + { + "epoch": 0.15714913244014936, + "grad_norm": 2.4205715656280518, + "learning_rate": 7.848518111964874e-07, + "loss": 0.9113, + "mean_token_accuracy": 0.7169144749641418, + "num_tokens": 36761197.0, + "step": 1431 + }, + { + "epoch": 0.157258950142763, + "grad_norm": 2.4574952125549316, + "learning_rate": 7.854006586169045e-07, + "loss": 1.0522, + "mean_token_accuracy": 0.6847361326217651, + "num_tokens": 36783565.0, + "step": 1432 + }, + { + "epoch": 0.15736876784537668, + "grad_norm": 2.487067937850952, + "learning_rate": 7.859495060373216e-07, + "loss": 1.1117, + "mean_token_accuracy": 0.6741031408309937, + "num_tokens": 36808092.0, + "step": 1433 + }, + { + "epoch": 0.15747858554799032, + "grad_norm": 2.2578959465026855, + "learning_rate": 7.864983534577388e-07, + "loss": 0.9567, + "mean_token_accuracy": 0.71131432056427, + "num_tokens": 36833868.0, + "step": 1434 + }, + { + "epoch": 0.157588403250604, + "grad_norm": 2.510072708129883, + "learning_rate": 7.870472008781557e-07, + "loss": 1.0324, + "mean_token_accuracy": 0.6960924863815308, + "num_tokens": 36857919.0, + "step": 1435 + }, + { + "epoch": 0.15769822095321767, + "grad_norm": 2.3143301010131836, + "learning_rate": 7.875960482985729e-07, + "loss": 0.9425, + "mean_token_accuracy": 0.7205230593681335, + "num_tokens": 36881643.0, + "step": 1436 + }, + { + "epoch": 0.15780803865583132, + "grad_norm": 2.243581771850586, + "learning_rate": 7.881448957189901e-07, + "loss": 1.0943, + "mean_token_accuracy": 0.6708686947822571, + "num_tokens": 36910063.0, + "step": 1437 + }, + { + "epoch": 0.157917856358445, + "grad_norm": 2.1781179904937744, + "learning_rate": 7.886937431394071e-07, + "loss": 1.0817, + "mean_token_accuracy": 0.6736361980438232, + "num_tokens": 36938996.0, + "step": 1438 + }, + { + "epoch": 0.15802767406105864, + "grad_norm": 2.3750388622283936, + "learning_rate": 7.892425905598243e-07, + "loss": 0.9907, + "mean_token_accuracy": 0.6983785629272461, + "num_tokens": 36962742.0, + "step": 1439 + }, + { + "epoch": 0.1581374917636723, + "grad_norm": 2.4652020931243896, + "learning_rate": 7.897914379802415e-07, + "loss": 0.9353, + "mean_token_accuracy": 0.7155052423477173, + "num_tokens": 36986454.0, + "step": 1440 + }, + { + "epoch": 0.15824730946628596, + "grad_norm": 2.358389139175415, + "learning_rate": 7.903402854006585e-07, + "loss": 0.9755, + "mean_token_accuracy": 0.7068676948547363, + "num_tokens": 37010817.0, + "step": 1441 + }, + { + "epoch": 0.15835712716889963, + "grad_norm": 2.551792860031128, + "learning_rate": 7.908891328210757e-07, + "loss": 1.0607, + "mean_token_accuracy": 0.6878511309623718, + "num_tokens": 37034593.0, + "step": 1442 + }, + { + "epoch": 0.15846694487151328, + "grad_norm": 2.4623711109161377, + "learning_rate": 7.914379802414928e-07, + "loss": 0.9124, + "mean_token_accuracy": 0.7216609716415405, + "num_tokens": 37056796.0, + "step": 1443 + }, + { + "epoch": 0.15857676257412695, + "grad_norm": 2.0576846599578857, + "learning_rate": 7.919868276619099e-07, + "loss": 0.9395, + "mean_token_accuracy": 0.7109856605529785, + "num_tokens": 37086774.0, + "step": 1444 + }, + { + "epoch": 0.15868658027674062, + "grad_norm": 2.2916805744171143, + "learning_rate": 7.925356750823271e-07, + "loss": 1.0407, + "mean_token_accuracy": 0.6960600018501282, + "num_tokens": 37111137.0, + "step": 1445 + }, + { + "epoch": 0.15879639797935427, + "grad_norm": 1.9371598958969116, + "learning_rate": 7.930845225027442e-07, + "loss": 0.9814, + "mean_token_accuracy": 0.6986298561096191, + "num_tokens": 37142189.0, + "step": 1446 + }, + { + "epoch": 0.15890621568196794, + "grad_norm": 2.2077322006225586, + "learning_rate": 7.936333699231613e-07, + "loss": 1.0904, + "mean_token_accuracy": 0.6760100722312927, + "num_tokens": 37170033.0, + "step": 1447 + }, + { + "epoch": 0.1590160333845816, + "grad_norm": 2.5420191287994385, + "learning_rate": 7.941822173435785e-07, + "loss": 0.9983, + "mean_token_accuracy": 0.7011175155639648, + "num_tokens": 37191992.0, + "step": 1448 + }, + { + "epoch": 0.15912585108719526, + "grad_norm": 2.326845169067383, + "learning_rate": 7.947310647639956e-07, + "loss": 1.0467, + "mean_token_accuracy": 0.689427375793457, + "num_tokens": 37216223.0, + "step": 1449 + }, + { + "epoch": 0.1592356687898089, + "grad_norm": 2.2766785621643066, + "learning_rate": 7.952799121844127e-07, + "loss": 0.9662, + "mean_token_accuracy": 0.7079501152038574, + "num_tokens": 37240804.0, + "step": 1450 + }, + { + "epoch": 0.15934548649242258, + "grad_norm": 2.464667558670044, + "learning_rate": 7.958287596048299e-07, + "loss": 1.028, + "mean_token_accuracy": 0.6873897314071655, + "num_tokens": 37263263.0, + "step": 1451 + }, + { + "epoch": 0.15945530419503623, + "grad_norm": 2.102471113204956, + "learning_rate": 7.96377607025247e-07, + "loss": 0.9651, + "mean_token_accuracy": 0.7106063365936279, + "num_tokens": 37291502.0, + "step": 1452 + }, + { + "epoch": 0.1595651218976499, + "grad_norm": 2.298757553100586, + "learning_rate": 7.969264544456641e-07, + "loss": 1.0567, + "mean_token_accuracy": 0.6890108585357666, + "num_tokens": 37315053.0, + "step": 1453 + }, + { + "epoch": 0.15967493960026355, + "grad_norm": 2.4137728214263916, + "learning_rate": 7.974753018660811e-07, + "loss": 1.1463, + "mean_token_accuracy": 0.667669415473938, + "num_tokens": 37341666.0, + "step": 1454 + }, + { + "epoch": 0.15978475730287722, + "grad_norm": 2.099700450897217, + "learning_rate": 7.980241492864983e-07, + "loss": 1.0503, + "mean_token_accuracy": 0.6931610107421875, + "num_tokens": 37369727.0, + "step": 1455 + }, + { + "epoch": 0.1598945750054909, + "grad_norm": 2.2747673988342285, + "learning_rate": 7.985729967069154e-07, + "loss": 1.0315, + "mean_token_accuracy": 0.6869220733642578, + "num_tokens": 37397653.0, + "step": 1456 + }, + { + "epoch": 0.16000439270810454, + "grad_norm": 2.3970582485198975, + "learning_rate": 7.991218441273325e-07, + "loss": 0.9997, + "mean_token_accuracy": 0.6928779482841492, + "num_tokens": 37421986.0, + "step": 1457 + }, + { + "epoch": 0.16011421041071822, + "grad_norm": 2.3653457164764404, + "learning_rate": 7.996706915477497e-07, + "loss": 0.9931, + "mean_token_accuracy": 0.7053875923156738, + "num_tokens": 37445573.0, + "step": 1458 + }, + { + "epoch": 0.16022402811333186, + "grad_norm": 2.0408010482788086, + "learning_rate": 8.002195389681668e-07, + "loss": 1.0255, + "mean_token_accuracy": 0.6957180500030518, + "num_tokens": 37476618.0, + "step": 1459 + }, + { + "epoch": 0.16033384581594554, + "grad_norm": 2.333451509475708, + "learning_rate": 8.007683863885839e-07, + "loss": 1.0701, + "mean_token_accuracy": 0.689022421836853, + "num_tokens": 37502711.0, + "step": 1460 + }, + { + "epoch": 0.16044366351855918, + "grad_norm": 2.169579029083252, + "learning_rate": 8.013172338090011e-07, + "loss": 0.9706, + "mean_token_accuracy": 0.7134848237037659, + "num_tokens": 37529771.0, + "step": 1461 + }, + { + "epoch": 0.16055348122117286, + "grad_norm": 2.551346778869629, + "learning_rate": 8.018660812294182e-07, + "loss": 1.0961, + "mean_token_accuracy": 0.6836872696876526, + "num_tokens": 37551970.0, + "step": 1462 + }, + { + "epoch": 0.1606632989237865, + "grad_norm": 2.2528345584869385, + "learning_rate": 8.024149286498353e-07, + "loss": 0.9901, + "mean_token_accuracy": 0.6991355419158936, + "num_tokens": 37580657.0, + "step": 1463 + }, + { + "epoch": 0.16077311662640018, + "grad_norm": 2.1846871376037598, + "learning_rate": 8.029637760702525e-07, + "loss": 1.0294, + "mean_token_accuracy": 0.6917521953582764, + "num_tokens": 37608424.0, + "step": 1464 + }, + { + "epoch": 0.16088293432901385, + "grad_norm": 2.196384906768799, + "learning_rate": 8.035126234906695e-07, + "loss": 1.0349, + "mean_token_accuracy": 0.6856536269187927, + "num_tokens": 37636343.0, + "step": 1465 + }, + { + "epoch": 0.1609927520316275, + "grad_norm": 2.063547134399414, + "learning_rate": 8.040614709110867e-07, + "loss": 1.0017, + "mean_token_accuracy": 0.706277072429657, + "num_tokens": 37666112.0, + "step": 1466 + }, + { + "epoch": 0.16110256973424117, + "grad_norm": 2.29249906539917, + "learning_rate": 8.046103183315039e-07, + "loss": 0.9665, + "mean_token_accuracy": 0.7065632343292236, + "num_tokens": 37692014.0, + "step": 1467 + }, + { + "epoch": 0.16121238743685481, + "grad_norm": 2.368058919906616, + "learning_rate": 8.051591657519209e-07, + "loss": 0.9928, + "mean_token_accuracy": 0.7021204233169556, + "num_tokens": 37716302.0, + "step": 1468 + }, + { + "epoch": 0.1613222051394685, + "grad_norm": 2.240492582321167, + "learning_rate": 8.057080131723381e-07, + "loss": 1.144, + "mean_token_accuracy": 0.6575920581817627, + "num_tokens": 37745764.0, + "step": 1469 + }, + { + "epoch": 0.16143202284208213, + "grad_norm": 2.455778121948242, + "learning_rate": 8.062568605927553e-07, + "loss": 1.0013, + "mean_token_accuracy": 0.7006898522377014, + "num_tokens": 37768493.0, + "step": 1470 + }, + { + "epoch": 0.1615418405446958, + "grad_norm": 2.460007667541504, + "learning_rate": 8.068057080131723e-07, + "loss": 0.9816, + "mean_token_accuracy": 0.7037140727043152, + "num_tokens": 37789407.0, + "step": 1471 + }, + { + "epoch": 0.16165165824730945, + "grad_norm": 3.1921112537384033, + "learning_rate": 8.073545554335894e-07, + "loss": 0.9822, + "mean_token_accuracy": 0.7076638340950012, + "num_tokens": 37810254.0, + "step": 1472 + }, + { + "epoch": 0.16176147594992313, + "grad_norm": 2.195296049118042, + "learning_rate": 8.079034028540066e-07, + "loss": 1.1224, + "mean_token_accuracy": 0.6666682958602905, + "num_tokens": 37841492.0, + "step": 1473 + }, + { + "epoch": 0.1618712936525368, + "grad_norm": 2.284658670425415, + "learning_rate": 8.084522502744236e-07, + "loss": 0.994, + "mean_token_accuracy": 0.7018691301345825, + "num_tokens": 37869624.0, + "step": 1474 + }, + { + "epoch": 0.16198111135515045, + "grad_norm": 2.0562028884887695, + "learning_rate": 8.090010976948408e-07, + "loss": 1.1184, + "mean_token_accuracy": 0.6795060634613037, + "num_tokens": 37900970.0, + "step": 1475 + }, + { + "epoch": 0.16209092905776412, + "grad_norm": 2.374619483947754, + "learning_rate": 8.095499451152578e-07, + "loss": 1.109, + "mean_token_accuracy": 0.6742105484008789, + "num_tokens": 37925002.0, + "step": 1476 + }, + { + "epoch": 0.16220074676037777, + "grad_norm": 2.6913397312164307, + "learning_rate": 8.10098792535675e-07, + "loss": 0.9614, + "mean_token_accuracy": 0.7103318572044373, + "num_tokens": 37944507.0, + "step": 1477 + }, + { + "epoch": 0.16231056446299144, + "grad_norm": 2.4993855953216553, + "learning_rate": 8.106476399560922e-07, + "loss": 1.0278, + "mean_token_accuracy": 0.6933512687683105, + "num_tokens": 37967081.0, + "step": 1478 + }, + { + "epoch": 0.1624203821656051, + "grad_norm": 2.259850025177002, + "learning_rate": 8.111964873765092e-07, + "loss": 1.0136, + "mean_token_accuracy": 0.6948584318161011, + "num_tokens": 37992341.0, + "step": 1479 + }, + { + "epoch": 0.16253019986821876, + "grad_norm": 2.592531204223633, + "learning_rate": 8.117453347969264e-07, + "loss": 0.9425, + "mean_token_accuracy": 0.713249921798706, + "num_tokens": 38012685.0, + "step": 1480 + }, + { + "epoch": 0.1626400175708324, + "grad_norm": 2.344348192214966, + "learning_rate": 8.122941822173436e-07, + "loss": 1.0027, + "mean_token_accuracy": 0.6956416368484497, + "num_tokens": 38037335.0, + "step": 1481 + }, + { + "epoch": 0.16274983527344608, + "grad_norm": 2.081057071685791, + "learning_rate": 8.128430296377606e-07, + "loss": 1.11, + "mean_token_accuracy": 0.6686311364173889, + "num_tokens": 38069682.0, + "step": 1482 + }, + { + "epoch": 0.16285965297605975, + "grad_norm": 2.235722541809082, + "learning_rate": 8.133918770581778e-07, + "loss": 0.9613, + "mean_token_accuracy": 0.7088592052459717, + "num_tokens": 38096023.0, + "step": 1483 + }, + { + "epoch": 0.1629694706786734, + "grad_norm": 2.2916994094848633, + "learning_rate": 8.13940724478595e-07, + "loss": 0.9867, + "mean_token_accuracy": 0.7000312805175781, + "num_tokens": 38120960.0, + "step": 1484 + }, + { + "epoch": 0.16307928838128707, + "grad_norm": 2.433645009994507, + "learning_rate": 8.14489571899012e-07, + "loss": 1.0459, + "mean_token_accuracy": 0.6915198564529419, + "num_tokens": 38144413.0, + "step": 1485 + }, + { + "epoch": 0.16318910608390072, + "grad_norm": 2.417119264602661, + "learning_rate": 8.150384193194292e-07, + "loss": 1.0865, + "mean_token_accuracy": 0.6851864457130432, + "num_tokens": 38169472.0, + "step": 1486 + }, + { + "epoch": 0.1632989237865144, + "grad_norm": 2.1854448318481445, + "learning_rate": 8.155872667398463e-07, + "loss": 1.1148, + "mean_token_accuracy": 0.6878455877304077, + "num_tokens": 38197532.0, + "step": 1487 + }, + { + "epoch": 0.16340874148912804, + "grad_norm": 2.952998638153076, + "learning_rate": 8.161361141602634e-07, + "loss": 0.8666, + "mean_token_accuracy": 0.7328557968139648, + "num_tokens": 38214676.0, + "step": 1488 + }, + { + "epoch": 0.1635185591917417, + "grad_norm": 2.404900312423706, + "learning_rate": 8.166849615806806e-07, + "loss": 1.069, + "mean_token_accuracy": 0.6818965673446655, + "num_tokens": 38240034.0, + "step": 1489 + }, + { + "epoch": 0.16362837689435536, + "grad_norm": 2.237359046936035, + "learning_rate": 8.172338090010977e-07, + "loss": 1.0969, + "mean_token_accuracy": 0.6772715449333191, + "num_tokens": 38270966.0, + "step": 1490 + }, + { + "epoch": 0.16373819459696903, + "grad_norm": 2.500392198562622, + "learning_rate": 8.177826564215147e-07, + "loss": 1.0895, + "mean_token_accuracy": 0.6887064576148987, + "num_tokens": 38293163.0, + "step": 1491 + }, + { + "epoch": 0.16384801229958268, + "grad_norm": 2.2396178245544434, + "learning_rate": 8.183315038419319e-07, + "loss": 1.0107, + "mean_token_accuracy": 0.7001011371612549, + "num_tokens": 38319479.0, + "step": 1492 + }, + { + "epoch": 0.16395783000219635, + "grad_norm": 2.5234458446502686, + "learning_rate": 8.18880351262349e-07, + "loss": 0.9217, + "mean_token_accuracy": 0.7155214548110962, + "num_tokens": 38339311.0, + "step": 1493 + }, + { + "epoch": 0.16406764770481003, + "grad_norm": 2.4268546104431152, + "learning_rate": 8.194291986827661e-07, + "loss": 1.0524, + "mean_token_accuracy": 0.6899713277816772, + "num_tokens": 38364020.0, + "step": 1494 + }, + { + "epoch": 0.16417746540742367, + "grad_norm": 2.6451315879821777, + "learning_rate": 8.199780461031833e-07, + "loss": 0.9526, + "mean_token_accuracy": 0.7116807103157043, + "num_tokens": 38386331.0, + "step": 1495 + }, + { + "epoch": 0.16428728311003735, + "grad_norm": 2.124600648880005, + "learning_rate": 8.205268935236004e-07, + "loss": 1.0077, + "mean_token_accuracy": 0.6954995393753052, + "num_tokens": 38414855.0, + "step": 1496 + }, + { + "epoch": 0.164397100812651, + "grad_norm": 2.57576322555542, + "learning_rate": 8.210757409440175e-07, + "loss": 1.0695, + "mean_token_accuracy": 0.6799330711364746, + "num_tokens": 38437833.0, + "step": 1497 + }, + { + "epoch": 0.16450691851526467, + "grad_norm": 2.500609874725342, + "learning_rate": 8.216245883644346e-07, + "loss": 1.0578, + "mean_token_accuracy": 0.687585711479187, + "num_tokens": 38461098.0, + "step": 1498 + }, + { + "epoch": 0.1646167362178783, + "grad_norm": 2.3005917072296143, + "learning_rate": 8.221734357848518e-07, + "loss": 1.0706, + "mean_token_accuracy": 0.688995897769928, + "num_tokens": 38486314.0, + "step": 1499 + }, + { + "epoch": 0.16472655392049199, + "grad_norm": 2.303981065750122, + "learning_rate": 8.227222832052689e-07, + "loss": 0.9583, + "mean_token_accuracy": 0.7176511287689209, + "num_tokens": 38509464.0, + "step": 1500 + }, + { + "epoch": 0.16483637162310563, + "grad_norm": 2.107926845550537, + "learning_rate": 8.23271130625686e-07, + "loss": 1.0441, + "mean_token_accuracy": 0.6906611323356628, + "num_tokens": 38538339.0, + "step": 1501 + }, + { + "epoch": 0.1649461893257193, + "grad_norm": 2.2550535202026367, + "learning_rate": 8.238199780461032e-07, + "loss": 0.9672, + "mean_token_accuracy": 0.7067452669143677, + "num_tokens": 38563304.0, + "step": 1502 + }, + { + "epoch": 0.16505600702833298, + "grad_norm": 2.375624418258667, + "learning_rate": 8.243688254665203e-07, + "loss": 0.8397, + "mean_token_accuracy": 0.7363322973251343, + "num_tokens": 38583670.0, + "step": 1503 + }, + { + "epoch": 0.16516582473094663, + "grad_norm": 2.4647462368011475, + "learning_rate": 8.249176728869374e-07, + "loss": 0.9635, + "mean_token_accuracy": 0.7207772135734558, + "num_tokens": 38605282.0, + "step": 1504 + }, + { + "epoch": 0.1652756424335603, + "grad_norm": 2.4879560470581055, + "learning_rate": 8.254665203073546e-07, + "loss": 0.9384, + "mean_token_accuracy": 0.7171170711517334, + "num_tokens": 38627616.0, + "step": 1505 + }, + { + "epoch": 0.16538546013617395, + "grad_norm": 2.4321489334106445, + "learning_rate": 8.260153677277717e-07, + "loss": 0.9623, + "mean_token_accuracy": 0.7083721160888672, + "num_tokens": 38649311.0, + "step": 1506 + }, + { + "epoch": 0.16549527783878762, + "grad_norm": 2.5981082916259766, + "learning_rate": 8.265642151481888e-07, + "loss": 0.997, + "mean_token_accuracy": 0.6980552673339844, + "num_tokens": 38670245.0, + "step": 1507 + }, + { + "epoch": 0.16560509554140126, + "grad_norm": 2.425659656524658, + "learning_rate": 8.27113062568606e-07, + "loss": 0.9521, + "mean_token_accuracy": 0.7115659713745117, + "num_tokens": 38692731.0, + "step": 1508 + }, + { + "epoch": 0.16571491324401494, + "grad_norm": 2.587747812271118, + "learning_rate": 8.27661909989023e-07, + "loss": 1.0433, + "mean_token_accuracy": 0.6992659568786621, + "num_tokens": 38715103.0, + "step": 1509 + }, + { + "epoch": 0.16582473094662858, + "grad_norm": 2.440281629562378, + "learning_rate": 8.282107574094401e-07, + "loss": 1.032, + "mean_token_accuracy": 0.6880885362625122, + "num_tokens": 38738063.0, + "step": 1510 + }, + { + "epoch": 0.16593454864924226, + "grad_norm": 2.427586317062378, + "learning_rate": 8.287596048298572e-07, + "loss": 1.006, + "mean_token_accuracy": 0.6952477097511292, + "num_tokens": 38761678.0, + "step": 1511 + }, + { + "epoch": 0.16604436635185593, + "grad_norm": 2.2549350261688232, + "learning_rate": 8.293084522502743e-07, + "loss": 1.0486, + "mean_token_accuracy": 0.6901805996894836, + "num_tokens": 38785972.0, + "step": 1512 + }, + { + "epoch": 0.16615418405446958, + "grad_norm": 2.087517023086548, + "learning_rate": 8.298572996706915e-07, + "loss": 1.0241, + "mean_token_accuracy": 0.6959050893783569, + "num_tokens": 38815918.0, + "step": 1513 + }, + { + "epoch": 0.16626400175708325, + "grad_norm": 2.252124786376953, + "learning_rate": 8.304061470911086e-07, + "loss": 1.0942, + "mean_token_accuracy": 0.6731599569320679, + "num_tokens": 38842469.0, + "step": 1514 + }, + { + "epoch": 0.1663738194596969, + "grad_norm": 2.472151041030884, + "learning_rate": 8.309549945115257e-07, + "loss": 1.0094, + "mean_token_accuracy": 0.7010284066200256, + "num_tokens": 38865409.0, + "step": 1515 + }, + { + "epoch": 0.16648363716231057, + "grad_norm": 2.079446792602539, + "learning_rate": 8.315038419319429e-07, + "loss": 0.9287, + "mean_token_accuracy": 0.7203820943832397, + "num_tokens": 38892915.0, + "step": 1516 + }, + { + "epoch": 0.16659345486492422, + "grad_norm": 2.0247044563293457, + "learning_rate": 8.3205268935236e-07, + "loss": 1.036, + "mean_token_accuracy": 0.6962776184082031, + "num_tokens": 38923520.0, + "step": 1517 + }, + { + "epoch": 0.1667032725675379, + "grad_norm": 2.507014036178589, + "learning_rate": 8.326015367727771e-07, + "loss": 1.1027, + "mean_token_accuracy": 0.670764684677124, + "num_tokens": 38945996.0, + "step": 1518 + }, + { + "epoch": 0.16681309027015154, + "grad_norm": 2.296037197113037, + "learning_rate": 8.331503841931943e-07, + "loss": 1.0759, + "mean_token_accuracy": 0.6830218434333801, + "num_tokens": 38970821.0, + "step": 1519 + }, + { + "epoch": 0.1669229079727652, + "grad_norm": 2.2515385150909424, + "learning_rate": 8.336992316136113e-07, + "loss": 1.0499, + "mean_token_accuracy": 0.687972903251648, + "num_tokens": 38997581.0, + "step": 1520 + }, + { + "epoch": 0.16703272567537888, + "grad_norm": 2.7354583740234375, + "learning_rate": 8.342480790340285e-07, + "loss": 0.967, + "mean_token_accuracy": 0.7115824222564697, + "num_tokens": 39016381.0, + "step": 1521 + }, + { + "epoch": 0.16714254337799253, + "grad_norm": 2.2540574073791504, + "learning_rate": 8.347969264544457e-07, + "loss": 1.0552, + "mean_token_accuracy": 0.6850624680519104, + "num_tokens": 39043012.0, + "step": 1522 + }, + { + "epoch": 0.1672523610806062, + "grad_norm": 2.3153910636901855, + "learning_rate": 8.353457738748627e-07, + "loss": 1.0262, + "mean_token_accuracy": 0.6963409781455994, + "num_tokens": 39067226.0, + "step": 1523 + }, + { + "epoch": 0.16736217878321985, + "grad_norm": 2.2421462535858154, + "learning_rate": 8.358946212952799e-07, + "loss": 0.9969, + "mean_token_accuracy": 0.7104412317276001, + "num_tokens": 39093696.0, + "step": 1524 + }, + { + "epoch": 0.16747199648583352, + "grad_norm": 2.057833433151245, + "learning_rate": 8.364434687156971e-07, + "loss": 1.052, + "mean_token_accuracy": 0.6845291256904602, + "num_tokens": 39125181.0, + "step": 1525 + }, + { + "epoch": 0.16758181418844717, + "grad_norm": 2.239694833755493, + "learning_rate": 8.369923161361141e-07, + "loss": 1.0126, + "mean_token_accuracy": 0.6971137523651123, + "num_tokens": 39150059.0, + "step": 1526 + }, + { + "epoch": 0.16769163189106084, + "grad_norm": 2.223843812942505, + "learning_rate": 8.375411635565313e-07, + "loss": 1.0108, + "mean_token_accuracy": 0.6964951753616333, + "num_tokens": 39176702.0, + "step": 1527 + }, + { + "epoch": 0.1678014495936745, + "grad_norm": 2.2776875495910645, + "learning_rate": 8.380900109769485e-07, + "loss": 0.9676, + "mean_token_accuracy": 0.7108197808265686, + "num_tokens": 39202411.0, + "step": 1528 + }, + { + "epoch": 0.16791126729628816, + "grad_norm": 2.177589178085327, + "learning_rate": 8.386388583973654e-07, + "loss": 1.0625, + "mean_token_accuracy": 0.6980199813842773, + "num_tokens": 39230269.0, + "step": 1529 + }, + { + "epoch": 0.1680210849989018, + "grad_norm": 2.189424753189087, + "learning_rate": 8.391877058177826e-07, + "loss": 1.0884, + "mean_token_accuracy": 0.6800079345703125, + "num_tokens": 39260161.0, + "step": 1530 + }, + { + "epoch": 0.16813090270151548, + "grad_norm": 2.103694438934326, + "learning_rate": 8.397365532381997e-07, + "loss": 0.9156, + "mean_token_accuracy": 0.7297905087471008, + "num_tokens": 39287044.0, + "step": 1531 + }, + { + "epoch": 0.16824072040412916, + "grad_norm": 2.433988332748413, + "learning_rate": 8.402854006586168e-07, + "loss": 1.0257, + "mean_token_accuracy": 0.6912546157836914, + "num_tokens": 39309271.0, + "step": 1532 + }, + { + "epoch": 0.1683505381067428, + "grad_norm": 2.1832034587860107, + "learning_rate": 8.40834248079034e-07, + "loss": 1.0745, + "mean_token_accuracy": 0.6791394948959351, + "num_tokens": 39339026.0, + "step": 1533 + }, + { + "epoch": 0.16846035580935648, + "grad_norm": 2.4336111545562744, + "learning_rate": 8.413830954994511e-07, + "loss": 1.073, + "mean_token_accuracy": 0.6862690448760986, + "num_tokens": 39364288.0, + "step": 1534 + }, + { + "epoch": 0.16857017351197012, + "grad_norm": 2.6058852672576904, + "learning_rate": 8.419319429198682e-07, + "loss": 1.0443, + "mean_token_accuracy": 0.687309980392456, + "num_tokens": 39386639.0, + "step": 1535 + }, + { + "epoch": 0.1686799912145838, + "grad_norm": 2.2421090602874756, + "learning_rate": 8.424807903402854e-07, + "loss": 0.973, + "mean_token_accuracy": 0.7083072662353516, + "num_tokens": 39412692.0, + "step": 1536 + }, + { + "epoch": 0.16878980891719744, + "grad_norm": 2.3548433780670166, + "learning_rate": 8.430296377607025e-07, + "loss": 1.0227, + "mean_token_accuracy": 0.6937377452850342, + "num_tokens": 39438253.0, + "step": 1537 + }, + { + "epoch": 0.16889962661981112, + "grad_norm": 2.5402729511260986, + "learning_rate": 8.435784851811196e-07, + "loss": 1.065, + "mean_token_accuracy": 0.6789489984512329, + "num_tokens": 39461223.0, + "step": 1538 + }, + { + "epoch": 0.16900944432242476, + "grad_norm": 2.5869293212890625, + "learning_rate": 8.441273326015367e-07, + "loss": 1.0541, + "mean_token_accuracy": 0.6944140195846558, + "num_tokens": 39486840.0, + "step": 1539 + }, + { + "epoch": 0.16911926202503844, + "grad_norm": 2.084749221801758, + "learning_rate": 8.446761800219539e-07, + "loss": 1.1221, + "mean_token_accuracy": 0.6637789607048035, + "num_tokens": 39519019.0, + "step": 1540 + }, + { + "epoch": 0.1692290797276521, + "grad_norm": 2.2795913219451904, + "learning_rate": 8.45225027442371e-07, + "loss": 1.0928, + "mean_token_accuracy": 0.6769983172416687, + "num_tokens": 39544790.0, + "step": 1541 + }, + { + "epoch": 0.16933889743026576, + "grad_norm": 2.3467471599578857, + "learning_rate": 8.457738748627881e-07, + "loss": 1.0371, + "mean_token_accuracy": 0.6921284198760986, + "num_tokens": 39568629.0, + "step": 1542 + }, + { + "epoch": 0.16944871513287943, + "grad_norm": 2.460115671157837, + "learning_rate": 8.463227222832053e-07, + "loss": 0.9805, + "mean_token_accuracy": 0.703757643699646, + "num_tokens": 39591407.0, + "step": 1543 + }, + { + "epoch": 0.16955853283549308, + "grad_norm": 2.277059316635132, + "learning_rate": 8.468715697036224e-07, + "loss": 1.0487, + "mean_token_accuracy": 0.6884174346923828, + "num_tokens": 39615948.0, + "step": 1544 + }, + { + "epoch": 0.16966835053810675, + "grad_norm": 2.4090371131896973, + "learning_rate": 8.474204171240395e-07, + "loss": 1.0195, + "mean_token_accuracy": 0.6974718570709229, + "num_tokens": 39638946.0, + "step": 1545 + }, + { + "epoch": 0.1697781682407204, + "grad_norm": 2.2205588817596436, + "learning_rate": 8.479692645444567e-07, + "loss": 0.9641, + "mean_token_accuracy": 0.7068171501159668, + "num_tokens": 39662942.0, + "step": 1546 + }, + { + "epoch": 0.16988798594333407, + "grad_norm": 2.627084493637085, + "learning_rate": 8.485181119648738e-07, + "loss": 1.0679, + "mean_token_accuracy": 0.6865444183349609, + "num_tokens": 39685283.0, + "step": 1547 + }, + { + "epoch": 0.16999780364594771, + "grad_norm": 2.205289125442505, + "learning_rate": 8.490669593852908e-07, + "loss": 0.9937, + "mean_token_accuracy": 0.7009989023208618, + "num_tokens": 39711614.0, + "step": 1548 + }, + { + "epoch": 0.1701076213485614, + "grad_norm": 2.224844217300415, + "learning_rate": 8.496158068057079e-07, + "loss": 1.0198, + "mean_token_accuracy": 0.700654923915863, + "num_tokens": 39738128.0, + "step": 1549 + }, + { + "epoch": 0.17021743905117506, + "grad_norm": 2.340400457382202, + "learning_rate": 8.50164654226125e-07, + "loss": 0.9847, + "mean_token_accuracy": 0.7025058269500732, + "num_tokens": 39762811.0, + "step": 1550 + }, + { + "epoch": 0.1703272567537887, + "grad_norm": 2.636901378631592, + "learning_rate": 8.507135016465422e-07, + "loss": 1.0145, + "mean_token_accuracy": 0.6977213621139526, + "num_tokens": 39783171.0, + "step": 1551 + }, + { + "epoch": 0.17043707445640238, + "grad_norm": 2.392514944076538, + "learning_rate": 8.512623490669593e-07, + "loss": 1.0176, + "mean_token_accuracy": 0.7018164396286011, + "num_tokens": 39806633.0, + "step": 1552 + }, + { + "epoch": 0.17054689215901603, + "grad_norm": 2.247715950012207, + "learning_rate": 8.518111964873764e-07, + "loss": 1.0559, + "mean_token_accuracy": 0.6821762919425964, + "num_tokens": 39833181.0, + "step": 1553 + }, + { + "epoch": 0.1706567098616297, + "grad_norm": 2.0564587116241455, + "learning_rate": 8.523600439077936e-07, + "loss": 0.9956, + "mean_token_accuracy": 0.6960815191268921, + "num_tokens": 39861343.0, + "step": 1554 + }, + { + "epoch": 0.17076652756424335, + "grad_norm": 2.000586986541748, + "learning_rate": 8.529088913282107e-07, + "loss": 1.0901, + "mean_token_accuracy": 0.6732756495475769, + "num_tokens": 39894468.0, + "step": 1555 + }, + { + "epoch": 0.17087634526685702, + "grad_norm": 2.1239309310913086, + "learning_rate": 8.534577387486278e-07, + "loss": 1.0845, + "mean_token_accuracy": 0.6704649925231934, + "num_tokens": 39924360.0, + "step": 1556 + }, + { + "epoch": 0.17098616296947067, + "grad_norm": 2.1107969284057617, + "learning_rate": 8.54006586169045e-07, + "loss": 0.9879, + "mean_token_accuracy": 0.6985853910446167, + "num_tokens": 39952403.0, + "step": 1557 + }, + { + "epoch": 0.17109598067208434, + "grad_norm": 2.202968120574951, + "learning_rate": 8.545554335894621e-07, + "loss": 1.1235, + "mean_token_accuracy": 0.6692643165588379, + "num_tokens": 39981719.0, + "step": 1558 + }, + { + "epoch": 0.17120579837469801, + "grad_norm": 2.1792144775390625, + "learning_rate": 8.551042810098792e-07, + "loss": 1.0704, + "mean_token_accuracy": 0.6824908256530762, + "num_tokens": 40010708.0, + "step": 1559 + }, + { + "epoch": 0.17131561607731166, + "grad_norm": 2.3027806282043457, + "learning_rate": 8.556531284302964e-07, + "loss": 1.1269, + "mean_token_accuracy": 0.6696293950080872, + "num_tokens": 40038881.0, + "step": 1560 + }, + { + "epoch": 0.17142543377992533, + "grad_norm": 2.379857063293457, + "learning_rate": 8.562019758507134e-07, + "loss": 1.021, + "mean_token_accuracy": 0.6997713446617126, + "num_tokens": 40062372.0, + "step": 1561 + }, + { + "epoch": 0.17153525148253898, + "grad_norm": 2.208815574645996, + "learning_rate": 8.567508232711306e-07, + "loss": 1.0637, + "mean_token_accuracy": 0.683570921421051, + "num_tokens": 40091837.0, + "step": 1562 + }, + { + "epoch": 0.17164506918515265, + "grad_norm": 2.3720831871032715, + "learning_rate": 8.572996706915478e-07, + "loss": 0.9626, + "mean_token_accuracy": 0.7110499143600464, + "num_tokens": 40115438.0, + "step": 1563 + }, + { + "epoch": 0.1717548868877663, + "grad_norm": 2.6327762603759766, + "learning_rate": 8.578485181119648e-07, + "loss": 0.9333, + "mean_token_accuracy": 0.7139830589294434, + "num_tokens": 40134020.0, + "step": 1564 + }, + { + "epoch": 0.17186470459037997, + "grad_norm": 2.218276262283325, + "learning_rate": 8.58397365532382e-07, + "loss": 1.0032, + "mean_token_accuracy": 0.6973327398300171, + "num_tokens": 40164084.0, + "step": 1565 + }, + { + "epoch": 0.17197452229299362, + "grad_norm": 2.2564914226531982, + "learning_rate": 8.589462129527992e-07, + "loss": 1.0768, + "mean_token_accuracy": 0.6752854585647583, + "num_tokens": 40192330.0, + "step": 1566 + }, + { + "epoch": 0.1720843399956073, + "grad_norm": 2.0599255561828613, + "learning_rate": 8.594950603732161e-07, + "loss": 1.0353, + "mean_token_accuracy": 0.6921811103820801, + "num_tokens": 40224089.0, + "step": 1567 + }, + { + "epoch": 0.17219415769822094, + "grad_norm": 2.514724016189575, + "learning_rate": 8.600439077936333e-07, + "loss": 0.957, + "mean_token_accuracy": 0.7093009948730469, + "num_tokens": 40246376.0, + "step": 1568 + }, + { + "epoch": 0.1723039754008346, + "grad_norm": 2.2807722091674805, + "learning_rate": 8.605927552140505e-07, + "loss": 1.0003, + "mean_token_accuracy": 0.699529767036438, + "num_tokens": 40271973.0, + "step": 1569 + }, + { + "epoch": 0.1724137931034483, + "grad_norm": 2.5304911136627197, + "learning_rate": 8.611416026344675e-07, + "loss": 0.9926, + "mean_token_accuracy": 0.7021827697753906, + "num_tokens": 40292559.0, + "step": 1570 + }, + { + "epoch": 0.17252361080606193, + "grad_norm": 2.2180771827697754, + "learning_rate": 8.616904500548847e-07, + "loss": 1.0549, + "mean_token_accuracy": 0.6875752806663513, + "num_tokens": 40319659.0, + "step": 1571 + }, + { + "epoch": 0.1726334285086756, + "grad_norm": 2.191319704055786, + "learning_rate": 8.622392974753018e-07, + "loss": 0.9825, + "mean_token_accuracy": 0.7050296068191528, + "num_tokens": 40345788.0, + "step": 1572 + }, + { + "epoch": 0.17274324621128925, + "grad_norm": 2.1864800453186035, + "learning_rate": 8.627881448957189e-07, + "loss": 1.1036, + "mean_token_accuracy": 0.673043966293335, + "num_tokens": 40377244.0, + "step": 1573 + }, + { + "epoch": 0.17285306391390293, + "grad_norm": 2.4660820960998535, + "learning_rate": 8.633369923161361e-07, + "loss": 1.0038, + "mean_token_accuracy": 0.6954711079597473, + "num_tokens": 40399040.0, + "step": 1574 + }, + { + "epoch": 0.17296288161651657, + "grad_norm": 2.0700817108154297, + "learning_rate": 8.638858397365532e-07, + "loss": 0.9888, + "mean_token_accuracy": 0.7031388282775879, + "num_tokens": 40428634.0, + "step": 1575 + }, + { + "epoch": 0.17307269931913025, + "grad_norm": 2.5283703804016113, + "learning_rate": 8.644346871569703e-07, + "loss": 1.0407, + "mean_token_accuracy": 0.6864623427391052, + "num_tokens": 40451052.0, + "step": 1576 + }, + { + "epoch": 0.1731825170217439, + "grad_norm": 2.294405698776245, + "learning_rate": 8.649835345773875e-07, + "loss": 1.0205, + "mean_token_accuracy": 0.6939899921417236, + "num_tokens": 40475115.0, + "step": 1577 + }, + { + "epoch": 0.17329233472435757, + "grad_norm": 2.3617866039276123, + "learning_rate": 8.655323819978046e-07, + "loss": 1.0134, + "mean_token_accuracy": 0.7037914991378784, + "num_tokens": 40502678.0, + "step": 1578 + }, + { + "epoch": 0.17340215242697124, + "grad_norm": 2.2055633068084717, + "learning_rate": 8.660812294182217e-07, + "loss": 1.014, + "mean_token_accuracy": 0.6868109107017517, + "num_tokens": 40529848.0, + "step": 1579 + }, + { + "epoch": 0.17351197012958489, + "grad_norm": 2.371046543121338, + "learning_rate": 8.666300768386389e-07, + "loss": 0.9361, + "mean_token_accuracy": 0.7158035039901733, + "num_tokens": 40552743.0, + "step": 1580 + }, + { + "epoch": 0.17362178783219856, + "grad_norm": 2.347365379333496, + "learning_rate": 8.67178924259056e-07, + "loss": 1.0589, + "mean_token_accuracy": 0.6841567158699036, + "num_tokens": 40577017.0, + "step": 1581 + }, + { + "epoch": 0.1737316055348122, + "grad_norm": 2.3794901371002197, + "learning_rate": 8.677277716794731e-07, + "loss": 1.0249, + "mean_token_accuracy": 0.6915205121040344, + "num_tokens": 40601330.0, + "step": 1582 + }, + { + "epoch": 0.17384142323742588, + "grad_norm": 2.171071767807007, + "learning_rate": 8.682766190998902e-07, + "loss": 1.1187, + "mean_token_accuracy": 0.6737173795700073, + "num_tokens": 40633592.0, + "step": 1583 + }, + { + "epoch": 0.17395124094003953, + "grad_norm": 2.0219051837921143, + "learning_rate": 8.688254665203073e-07, + "loss": 1.0541, + "mean_token_accuracy": 0.6860887408256531, + "num_tokens": 40666993.0, + "step": 1584 + }, + { + "epoch": 0.1740610586426532, + "grad_norm": 2.3135759830474854, + "learning_rate": 8.693743139407245e-07, + "loss": 1.1363, + "mean_token_accuracy": 0.6651942133903503, + "num_tokens": 40694795.0, + "step": 1585 + }, + { + "epoch": 0.17417087634526685, + "grad_norm": 2.3898613452911377, + "learning_rate": 8.699231613611415e-07, + "loss": 1.0137, + "mean_token_accuracy": 0.6966235637664795, + "num_tokens": 40718297.0, + "step": 1586 + }, + { + "epoch": 0.17428069404788052, + "grad_norm": 2.223151445388794, + "learning_rate": 8.704720087815586e-07, + "loss": 1.0661, + "mean_token_accuracy": 0.691643476486206, + "num_tokens": 40744712.0, + "step": 1587 + }, + { + "epoch": 0.1743905117504942, + "grad_norm": 2.4253458976745605, + "learning_rate": 8.710208562019758e-07, + "loss": 1.0558, + "mean_token_accuracy": 0.6995519399642944, + "num_tokens": 40768910.0, + "step": 1588 + }, + { + "epoch": 0.17450032945310784, + "grad_norm": 2.1548752784729004, + "learning_rate": 8.715697036223929e-07, + "loss": 1.0283, + "mean_token_accuracy": 0.6925768852233887, + "num_tokens": 40795868.0, + "step": 1589 + }, + { + "epoch": 0.1746101471557215, + "grad_norm": 2.3017609119415283, + "learning_rate": 8.7211855104281e-07, + "loss": 1.0184, + "mean_token_accuracy": 0.7016304731369019, + "num_tokens": 40820641.0, + "step": 1590 + }, + { + "epoch": 0.17471996485833516, + "grad_norm": 2.3587396144866943, + "learning_rate": 8.726673984632272e-07, + "loss": 0.9919, + "mean_token_accuracy": 0.7038770914077759, + "num_tokens": 40845837.0, + "step": 1591 + }, + { + "epoch": 0.17482978256094883, + "grad_norm": 2.277177095413208, + "learning_rate": 8.732162458836443e-07, + "loss": 1.1026, + "mean_token_accuracy": 0.6779232025146484, + "num_tokens": 40873741.0, + "step": 1592 + }, + { + "epoch": 0.17493960026356248, + "grad_norm": 2.215975046157837, + "learning_rate": 8.737650933040614e-07, + "loss": 1.0289, + "mean_token_accuracy": 0.6949242353439331, + "num_tokens": 40902548.0, + "step": 1593 + }, + { + "epoch": 0.17504941796617615, + "grad_norm": 2.4377405643463135, + "learning_rate": 8.743139407244785e-07, + "loss": 0.897, + "mean_token_accuracy": 0.7222210168838501, + "num_tokens": 40923592.0, + "step": 1594 + }, + { + "epoch": 0.1751592356687898, + "grad_norm": 2.275129556655884, + "learning_rate": 8.748627881448957e-07, + "loss": 1.0408, + "mean_token_accuracy": 0.6898860335350037, + "num_tokens": 40948949.0, + "step": 1595 + }, + { + "epoch": 0.17526905337140347, + "grad_norm": 2.5701816082000732, + "learning_rate": 8.754116355653128e-07, + "loss": 1.0424, + "mean_token_accuracy": 0.6829131841659546, + "num_tokens": 40970621.0, + "step": 1596 + }, + { + "epoch": 0.17537887107401715, + "grad_norm": 2.425382137298584, + "learning_rate": 8.759604829857299e-07, + "loss": 0.9542, + "mean_token_accuracy": 0.7093929052352905, + "num_tokens": 40994214.0, + "step": 1597 + }, + { + "epoch": 0.1754886887766308, + "grad_norm": 2.306924343109131, + "learning_rate": 8.765093304061471e-07, + "loss": 1.0024, + "mean_token_accuracy": 0.6998305320739746, + "num_tokens": 41019602.0, + "step": 1598 + }, + { + "epoch": 0.17559850647924446, + "grad_norm": 2.063629388809204, + "learning_rate": 8.770581778265642e-07, + "loss": 1.0436, + "mean_token_accuracy": 0.688634991645813, + "num_tokens": 41048712.0, + "step": 1599 + }, + { + "epoch": 0.1757083241818581, + "grad_norm": 1.9670889377593994, + "learning_rate": 8.776070252469813e-07, + "loss": 0.9876, + "mean_token_accuracy": 0.7002075910568237, + "num_tokens": 41078035.0, + "step": 1600 + }, + { + "epoch": 0.17581814188447178, + "grad_norm": 2.2901203632354736, + "learning_rate": 8.781558726673985e-07, + "loss": 1.0236, + "mean_token_accuracy": 0.6960361003875732, + "num_tokens": 41102361.0, + "step": 1601 + }, + { + "epoch": 0.17592795958708543, + "grad_norm": 2.6784677505493164, + "learning_rate": 8.787047200878156e-07, + "loss": 0.9654, + "mean_token_accuracy": 0.708403468132019, + "num_tokens": 41121285.0, + "step": 1602 + }, + { + "epoch": 0.1760377772896991, + "grad_norm": 2.141117811203003, + "learning_rate": 8.792535675082327e-07, + "loss": 1.1429, + "mean_token_accuracy": 0.6656637191772461, + "num_tokens": 41149088.0, + "step": 1603 + }, + { + "epoch": 0.17614759499231275, + "grad_norm": 2.256211042404175, + "learning_rate": 8.798024149286499e-07, + "loss": 1.0318, + "mean_token_accuracy": 0.6940227150917053, + "num_tokens": 41174890.0, + "step": 1604 + }, + { + "epoch": 0.17625741269492642, + "grad_norm": 2.3988983631134033, + "learning_rate": 8.803512623490668e-07, + "loss": 0.9313, + "mean_token_accuracy": 0.7175188064575195, + "num_tokens": 41197828.0, + "step": 1605 + }, + { + "epoch": 0.17636723039754007, + "grad_norm": 2.101802349090576, + "learning_rate": 8.80900109769484e-07, + "loss": 1.0468, + "mean_token_accuracy": 0.6912142038345337, + "num_tokens": 41228291.0, + "step": 1606 + }, + { + "epoch": 0.17647704810015374, + "grad_norm": 2.618912696838379, + "learning_rate": 8.814489571899012e-07, + "loss": 1.0205, + "mean_token_accuracy": 0.7008591294288635, + "num_tokens": 41250043.0, + "step": 1607 + }, + { + "epoch": 0.17658686580276742, + "grad_norm": 2.5853121280670166, + "learning_rate": 8.819978046103182e-07, + "loss": 1.0077, + "mean_token_accuracy": 0.6977077722549438, + "num_tokens": 41271623.0, + "step": 1608 + }, + { + "epoch": 0.17669668350538106, + "grad_norm": 2.3100879192352295, + "learning_rate": 8.825466520307354e-07, + "loss": 1.0917, + "mean_token_accuracy": 0.6839266419410706, + "num_tokens": 41297170.0, + "step": 1609 + }, + { + "epoch": 0.17680650120799474, + "grad_norm": 2.7820918560028076, + "learning_rate": 8.830954994511526e-07, + "loss": 1.0112, + "mean_token_accuracy": 0.6941941380500793, + "num_tokens": 41316036.0, + "step": 1610 + }, + { + "epoch": 0.17691631891060838, + "grad_norm": 2.141089677810669, + "learning_rate": 8.836443468715696e-07, + "loss": 1.0548, + "mean_token_accuracy": 0.6824113130569458, + "num_tokens": 41344544.0, + "step": 1611 + }, + { + "epoch": 0.17702613661322206, + "grad_norm": 2.053852081298828, + "learning_rate": 8.841931942919868e-07, + "loss": 0.9745, + "mean_token_accuracy": 0.703761100769043, + "num_tokens": 41372878.0, + "step": 1612 + }, + { + "epoch": 0.1771359543158357, + "grad_norm": 2.610609292984009, + "learning_rate": 8.84742041712404e-07, + "loss": 0.9705, + "mean_token_accuracy": 0.7050923109054565, + "num_tokens": 41395282.0, + "step": 1613 + }, + { + "epoch": 0.17724577201844938, + "grad_norm": 2.3698654174804688, + "learning_rate": 8.85290889132821e-07, + "loss": 1.0568, + "mean_token_accuracy": 0.6846176385879517, + "num_tokens": 41421399.0, + "step": 1614 + }, + { + "epoch": 0.17735558972106302, + "grad_norm": 2.4287896156311035, + "learning_rate": 8.858397365532382e-07, + "loss": 1.0129, + "mean_token_accuracy": 0.6968954801559448, + "num_tokens": 41443691.0, + "step": 1615 + }, + { + "epoch": 0.1774654074236767, + "grad_norm": 2.3521676063537598, + "learning_rate": 8.863885839736552e-07, + "loss": 1.0839, + "mean_token_accuracy": 0.6789191961288452, + "num_tokens": 41468975.0, + "step": 1616 + }, + { + "epoch": 0.17757522512629037, + "grad_norm": 2.316187620162964, + "learning_rate": 8.869374313940724e-07, + "loss": 0.9293, + "mean_token_accuracy": 0.7184240221977234, + "num_tokens": 41491140.0, + "step": 1617 + }, + { + "epoch": 0.17768504282890402, + "grad_norm": 2.9120514392852783, + "learning_rate": 8.874862788144896e-07, + "loss": 0.8986, + "mean_token_accuracy": 0.7228341698646545, + "num_tokens": 41508165.0, + "step": 1618 + }, + { + "epoch": 0.1777948605315177, + "grad_norm": 2.388038158416748, + "learning_rate": 8.880351262349066e-07, + "loss": 1.0548, + "mean_token_accuracy": 0.6916227340698242, + "num_tokens": 41532212.0, + "step": 1619 + }, + { + "epoch": 0.17790467823413134, + "grad_norm": 2.4221203327178955, + "learning_rate": 8.885839736553238e-07, + "loss": 1.023, + "mean_token_accuracy": 0.6991559863090515, + "num_tokens": 41555404.0, + "step": 1620 + }, + { + "epoch": 0.178014495936745, + "grad_norm": 2.207665205001831, + "learning_rate": 8.89132821075741e-07, + "loss": 1.0575, + "mean_token_accuracy": 0.6844311952590942, + "num_tokens": 41584069.0, + "step": 1621 + }, + { + "epoch": 0.17812431363935866, + "grad_norm": 2.2773051261901855, + "learning_rate": 8.89681668496158e-07, + "loss": 1.0795, + "mean_token_accuracy": 0.6777371168136597, + "num_tokens": 41611755.0, + "step": 1622 + }, + { + "epoch": 0.17823413134197233, + "grad_norm": 2.552604913711548, + "learning_rate": 8.902305159165752e-07, + "loss": 1.0002, + "mean_token_accuracy": 0.6954396963119507, + "num_tokens": 41631806.0, + "step": 1623 + }, + { + "epoch": 0.17834394904458598, + "grad_norm": 2.202993869781494, + "learning_rate": 8.907793633369924e-07, + "loss": 1.0695, + "mean_token_accuracy": 0.6866445541381836, + "num_tokens": 41660051.0, + "step": 1624 + }, + { + "epoch": 0.17845376674719965, + "grad_norm": 2.1622908115386963, + "learning_rate": 8.913282107574093e-07, + "loss": 1.0204, + "mean_token_accuracy": 0.6986954808235168, + "num_tokens": 41688284.0, + "step": 1625 + }, + { + "epoch": 0.17856358444981332, + "grad_norm": 2.7754251956939697, + "learning_rate": 8.918770581778265e-07, + "loss": 0.995, + "mean_token_accuracy": 0.6993615627288818, + "num_tokens": 41707974.0, + "step": 1626 + }, + { + "epoch": 0.17867340215242697, + "grad_norm": 2.4537787437438965, + "learning_rate": 8.924259055982436e-07, + "loss": 1.082, + "mean_token_accuracy": 0.6824201941490173, + "num_tokens": 41731105.0, + "step": 1627 + }, + { + "epoch": 0.17878321985504064, + "grad_norm": 2.6769635677337646, + "learning_rate": 8.929747530186607e-07, + "loss": 0.9699, + "mean_token_accuracy": 0.7062902450561523, + "num_tokens": 41749744.0, + "step": 1628 + }, + { + "epoch": 0.1788930375576543, + "grad_norm": 2.258493423461914, + "learning_rate": 8.935236004390779e-07, + "loss": 1.0644, + "mean_token_accuracy": 0.6853374242782593, + "num_tokens": 41774348.0, + "step": 1629 + }, + { + "epoch": 0.17900285526026796, + "grad_norm": 2.2267584800720215, + "learning_rate": 8.94072447859495e-07, + "loss": 1.0456, + "mean_token_accuracy": 0.6924964189529419, + "num_tokens": 41803342.0, + "step": 1630 + }, + { + "epoch": 0.1791126729628816, + "grad_norm": 2.3262767791748047, + "learning_rate": 8.946212952799121e-07, + "loss": 1.0675, + "mean_token_accuracy": 0.7002981901168823, + "num_tokens": 41828529.0, + "step": 1631 + }, + { + "epoch": 0.17922249066549528, + "grad_norm": 2.5942859649658203, + "learning_rate": 8.951701427003293e-07, + "loss": 1.0154, + "mean_token_accuracy": 0.6922813653945923, + "num_tokens": 41848730.0, + "step": 1632 + }, + { + "epoch": 0.17933230836810893, + "grad_norm": 2.0720999240875244, + "learning_rate": 8.957189901207464e-07, + "loss": 0.9945, + "mean_token_accuracy": 0.7031179666519165, + "num_tokens": 41876431.0, + "step": 1633 + }, + { + "epoch": 0.1794421260707226, + "grad_norm": 2.2985944747924805, + "learning_rate": 8.962678375411635e-07, + "loss": 1.0583, + "mean_token_accuracy": 0.6809114217758179, + "num_tokens": 41901581.0, + "step": 1634 + }, + { + "epoch": 0.17955194377333628, + "grad_norm": 2.3511247634887695, + "learning_rate": 8.968166849615807e-07, + "loss": 0.9472, + "mean_token_accuracy": 0.7138563394546509, + "num_tokens": 41924888.0, + "step": 1635 + }, + { + "epoch": 0.17966176147594992, + "grad_norm": 2.283608913421631, + "learning_rate": 8.973655323819978e-07, + "loss": 1.0029, + "mean_token_accuracy": 0.7019626498222351, + "num_tokens": 41950813.0, + "step": 1636 + }, + { + "epoch": 0.1797715791785636, + "grad_norm": 2.321422576904297, + "learning_rate": 8.979143798024149e-07, + "loss": 0.9292, + "mean_token_accuracy": 0.7173508405685425, + "num_tokens": 41974470.0, + "step": 1637 + }, + { + "epoch": 0.17988139688117724, + "grad_norm": 2.3252906799316406, + "learning_rate": 8.98463227222832e-07, + "loss": 1.0029, + "mean_token_accuracy": 0.7001075148582458, + "num_tokens": 41997850.0, + "step": 1638 + }, + { + "epoch": 0.17999121458379091, + "grad_norm": 2.5172317028045654, + "learning_rate": 8.990120746432492e-07, + "loss": 1.0362, + "mean_token_accuracy": 0.6956506967544556, + "num_tokens": 42023337.0, + "step": 1639 + }, + { + "epoch": 0.18010103228640456, + "grad_norm": 2.3523266315460205, + "learning_rate": 8.995609220636663e-07, + "loss": 0.9372, + "mean_token_accuracy": 0.7170712947845459, + "num_tokens": 42048023.0, + "step": 1640 + }, + { + "epoch": 0.18021084998901823, + "grad_norm": 2.251718521118164, + "learning_rate": 9.001097694840834e-07, + "loss": 1.0551, + "mean_token_accuracy": 0.6862331628799438, + "num_tokens": 42078048.0, + "step": 1641 + }, + { + "epoch": 0.18032066769163188, + "grad_norm": 2.602086305618286, + "learning_rate": 9.006586169045006e-07, + "loss": 0.8878, + "mean_token_accuracy": 0.723825216293335, + "num_tokens": 42098461.0, + "step": 1642 + }, + { + "epoch": 0.18043048539424555, + "grad_norm": 1.9415339231491089, + "learning_rate": 9.012074643249177e-07, + "loss": 0.9687, + "mean_token_accuracy": 0.709740161895752, + "num_tokens": 42132186.0, + "step": 1643 + }, + { + "epoch": 0.1805403030968592, + "grad_norm": 2.144057035446167, + "learning_rate": 9.017563117453347e-07, + "loss": 1.1183, + "mean_token_accuracy": 0.6693044900894165, + "num_tokens": 42163213.0, + "step": 1644 + }, + { + "epoch": 0.18065012079947287, + "grad_norm": 2.285038948059082, + "learning_rate": 9.023051591657519e-07, + "loss": 1.0569, + "mean_token_accuracy": 0.6805509328842163, + "num_tokens": 42187973.0, + "step": 1645 + }, + { + "epoch": 0.18075993850208655, + "grad_norm": 2.231509208679199, + "learning_rate": 9.02854006586169e-07, + "loss": 1.0403, + "mean_token_accuracy": 0.6918728351593018, + "num_tokens": 42214761.0, + "step": 1646 + }, + { + "epoch": 0.1808697562047002, + "grad_norm": 2.445645809173584, + "learning_rate": 9.034028540065861e-07, + "loss": 1.0866, + "mean_token_accuracy": 0.6716196537017822, + "num_tokens": 42238672.0, + "step": 1647 + }, + { + "epoch": 0.18097957390731387, + "grad_norm": 2.4196929931640625, + "learning_rate": 9.039517014270033e-07, + "loss": 1.1371, + "mean_token_accuracy": 0.6611063480377197, + "num_tokens": 42263986.0, + "step": 1648 + }, + { + "epoch": 0.1810893916099275, + "grad_norm": 2.559960126876831, + "learning_rate": 9.045005488474203e-07, + "loss": 1.0263, + "mean_token_accuracy": 0.6921746730804443, + "num_tokens": 42283692.0, + "step": 1649 + }, + { + "epoch": 0.1811992093125412, + "grad_norm": 2.0581681728363037, + "learning_rate": 9.050493962678375e-07, + "loss": 1.0107, + "mean_token_accuracy": 0.6957137584686279, + "num_tokens": 42313302.0, + "step": 1650 + }, + { + "epoch": 0.18130902701515483, + "grad_norm": 2.4803214073181152, + "learning_rate": 9.055982436882547e-07, + "loss": 1.0016, + "mean_token_accuracy": 0.6941118240356445, + "num_tokens": 42334751.0, + "step": 1651 + }, + { + "epoch": 0.1814188447177685, + "grad_norm": 2.1627519130706787, + "learning_rate": 9.061470911086717e-07, + "loss": 1.0714, + "mean_token_accuracy": 0.679439902305603, + "num_tokens": 42361802.0, + "step": 1652 + }, + { + "epoch": 0.18152866242038215, + "grad_norm": 2.2911806106567383, + "learning_rate": 9.066959385290889e-07, + "loss": 1.0825, + "mean_token_accuracy": 0.6782420873641968, + "num_tokens": 42388737.0, + "step": 1653 + }, + { + "epoch": 0.18163848012299583, + "grad_norm": 2.1380348205566406, + "learning_rate": 9.07244785949506e-07, + "loss": 1.07, + "mean_token_accuracy": 0.6841799020767212, + "num_tokens": 42417979.0, + "step": 1654 + }, + { + "epoch": 0.1817482978256095, + "grad_norm": 2.2410974502563477, + "learning_rate": 9.077936333699231e-07, + "loss": 0.9876, + "mean_token_accuracy": 0.7000136971473694, + "num_tokens": 42445095.0, + "step": 1655 + }, + { + "epoch": 0.18185811552822315, + "grad_norm": 2.213229179382324, + "learning_rate": 9.083424807903403e-07, + "loss": 0.9384, + "mean_token_accuracy": 0.7167978286743164, + "num_tokens": 42471016.0, + "step": 1656 + }, + { + "epoch": 0.18196793323083682, + "grad_norm": 2.5045549869537354, + "learning_rate": 9.088913282107573e-07, + "loss": 1.0897, + "mean_token_accuracy": 0.6742047071456909, + "num_tokens": 42496950.0, + "step": 1657 + }, + { + "epoch": 0.18207775093345047, + "grad_norm": 2.2930119037628174, + "learning_rate": 9.094401756311745e-07, + "loss": 1.0312, + "mean_token_accuracy": 0.6897271275520325, + "num_tokens": 42521080.0, + "step": 1658 + }, + { + "epoch": 0.18218756863606414, + "grad_norm": 2.5381970405578613, + "learning_rate": 9.099890230515917e-07, + "loss": 1.0873, + "mean_token_accuracy": 0.6725326180458069, + "num_tokens": 42543593.0, + "step": 1659 + }, + { + "epoch": 0.18229738633867779, + "grad_norm": 2.4748311042785645, + "learning_rate": 9.105378704720087e-07, + "loss": 0.9953, + "mean_token_accuracy": 0.7084068059921265, + "num_tokens": 42565834.0, + "step": 1660 + }, + { + "epoch": 0.18240720404129146, + "grad_norm": 2.3471670150756836, + "learning_rate": 9.110867178924259e-07, + "loss": 1.0865, + "mean_token_accuracy": 0.683125376701355, + "num_tokens": 42588971.0, + "step": 1661 + }, + { + "epoch": 0.1825170217439051, + "grad_norm": 2.008896589279175, + "learning_rate": 9.11635565312843e-07, + "loss": 1.004, + "mean_token_accuracy": 0.6954091787338257, + "num_tokens": 42618157.0, + "step": 1662 + }, + { + "epoch": 0.18262683944651878, + "grad_norm": 2.6956639289855957, + "learning_rate": 9.1218441273326e-07, + "loss": 0.986, + "mean_token_accuracy": 0.7051019668579102, + "num_tokens": 42637343.0, + "step": 1663 + }, + { + "epoch": 0.18273665714913245, + "grad_norm": 2.3204429149627686, + "learning_rate": 9.127332601536772e-07, + "loss": 0.9991, + "mean_token_accuracy": 0.7028645873069763, + "num_tokens": 42661514.0, + "step": 1664 + }, + { + "epoch": 0.1828464748517461, + "grad_norm": 2.41275954246521, + "learning_rate": 9.132821075740944e-07, + "loss": 1.0544, + "mean_token_accuracy": 0.6946877241134644, + "num_tokens": 42686158.0, + "step": 1665 + }, + { + "epoch": 0.18295629255435977, + "grad_norm": 2.5195538997650146, + "learning_rate": 9.138309549945114e-07, + "loss": 1.0446, + "mean_token_accuracy": 0.6886496543884277, + "num_tokens": 42708266.0, + "step": 1666 + }, + { + "epoch": 0.18306611025697342, + "grad_norm": 2.25583553314209, + "learning_rate": 9.143798024149286e-07, + "loss": 0.9584, + "mean_token_accuracy": 0.710178792476654, + "num_tokens": 42734255.0, + "step": 1667 + }, + { + "epoch": 0.1831759279595871, + "grad_norm": 2.1958539485931396, + "learning_rate": 9.149286498353457e-07, + "loss": 1.0183, + "mean_token_accuracy": 0.7076652646064758, + "num_tokens": 42762643.0, + "step": 1668 + }, + { + "epoch": 0.18328574566220074, + "grad_norm": 2.4440548419952393, + "learning_rate": 9.154774972557628e-07, + "loss": 1.0198, + "mean_token_accuracy": 0.6902986168861389, + "num_tokens": 42786777.0, + "step": 1669 + }, + { + "epoch": 0.1833955633648144, + "grad_norm": 2.361617088317871, + "learning_rate": 9.1602634467618e-07, + "loss": 1.0541, + "mean_token_accuracy": 0.6892130374908447, + "num_tokens": 42812036.0, + "step": 1670 + }, + { + "epoch": 0.18350538106742806, + "grad_norm": 2.415225028991699, + "learning_rate": 9.165751920965971e-07, + "loss": 0.9559, + "mean_token_accuracy": 0.7067506313323975, + "num_tokens": 42834603.0, + "step": 1671 + }, + { + "epoch": 0.18361519877004173, + "grad_norm": 2.3336923122406006, + "learning_rate": 9.171240395170142e-07, + "loss": 1.0023, + "mean_token_accuracy": 0.7002763748168945, + "num_tokens": 42858779.0, + "step": 1672 + }, + { + "epoch": 0.1837250164726554, + "grad_norm": 2.2049994468688965, + "learning_rate": 9.176728869374314e-07, + "loss": 1.0796, + "mean_token_accuracy": 0.6767953634262085, + "num_tokens": 42887328.0, + "step": 1673 + }, + { + "epoch": 0.18383483417526905, + "grad_norm": 2.6178596019744873, + "learning_rate": 9.182217343578485e-07, + "loss": 1.068, + "mean_token_accuracy": 0.691777765750885, + "num_tokens": 42908318.0, + "step": 1674 + }, + { + "epoch": 0.18394465187788273, + "grad_norm": 2.2412314414978027, + "learning_rate": 9.187705817782656e-07, + "loss": 1.0103, + "mean_token_accuracy": 0.7038843631744385, + "num_tokens": 42934130.0, + "step": 1675 + }, + { + "epoch": 0.18405446958049637, + "grad_norm": 2.3389692306518555, + "learning_rate": 9.193194291986828e-07, + "loss": 1.0772, + "mean_token_accuracy": 0.6838412284851074, + "num_tokens": 42958428.0, + "step": 1676 + }, + { + "epoch": 0.18416428728311005, + "grad_norm": 2.5723135471343994, + "learning_rate": 9.198682766190999e-07, + "loss": 1.0149, + "mean_token_accuracy": 0.6990461349487305, + "num_tokens": 42979782.0, + "step": 1677 + }, + { + "epoch": 0.1842741049857237, + "grad_norm": 2.148285150527954, + "learning_rate": 9.20417124039517e-07, + "loss": 1.1295, + "mean_token_accuracy": 0.6687197685241699, + "num_tokens": 43008804.0, + "step": 1678 + }, + { + "epoch": 0.18438392268833736, + "grad_norm": 2.2834436893463135, + "learning_rate": 9.209659714599341e-07, + "loss": 1.0776, + "mean_token_accuracy": 0.6818529963493347, + "num_tokens": 43034486.0, + "step": 1679 + }, + { + "epoch": 0.184493740390951, + "grad_norm": 1.9958419799804688, + "learning_rate": 9.215148188803513e-07, + "loss": 1.0839, + "mean_token_accuracy": 0.67557692527771, + "num_tokens": 43066375.0, + "step": 1680 + }, + { + "epoch": 0.18460355809356468, + "grad_norm": 1.860382080078125, + "learning_rate": 9.220636663007683e-07, + "loss": 1.006, + "mean_token_accuracy": 0.7019924521446228, + "num_tokens": 43101586.0, + "step": 1681 + }, + { + "epoch": 0.18471337579617833, + "grad_norm": 2.4648633003234863, + "learning_rate": 9.226125137211854e-07, + "loss": 1.0754, + "mean_token_accuracy": 0.6808419227600098, + "num_tokens": 43125274.0, + "step": 1682 + }, + { + "epoch": 0.184823193498792, + "grad_norm": 2.123202323913574, + "learning_rate": 9.231613611416026e-07, + "loss": 1.0529, + "mean_token_accuracy": 0.6899831295013428, + "num_tokens": 43153636.0, + "step": 1683 + }, + { + "epoch": 0.18493301120140568, + "grad_norm": 2.373389959335327, + "learning_rate": 9.237102085620197e-07, + "loss": 0.978, + "mean_token_accuracy": 0.6986680030822754, + "num_tokens": 43176825.0, + "step": 1684 + }, + { + "epoch": 0.18504282890401932, + "grad_norm": 2.251821279525757, + "learning_rate": 9.242590559824368e-07, + "loss": 0.9405, + "mean_token_accuracy": 0.7162076830863953, + "num_tokens": 43201950.0, + "step": 1685 + }, + { + "epoch": 0.185152646606633, + "grad_norm": 2.534069538116455, + "learning_rate": 9.24807903402854e-07, + "loss": 1.0082, + "mean_token_accuracy": 0.7057222127914429, + "num_tokens": 43224403.0, + "step": 1686 + }, + { + "epoch": 0.18526246430924664, + "grad_norm": 2.1130523681640625, + "learning_rate": 9.253567508232711e-07, + "loss": 1.0495, + "mean_token_accuracy": 0.6891427040100098, + "num_tokens": 43253631.0, + "step": 1687 + }, + { + "epoch": 0.18537228201186032, + "grad_norm": 2.3264505863189697, + "learning_rate": 9.259055982436882e-07, + "loss": 1.0417, + "mean_token_accuracy": 0.6893917322158813, + "num_tokens": 43277748.0, + "step": 1688 + }, + { + "epoch": 0.18548209971447396, + "grad_norm": 2.3414688110351562, + "learning_rate": 9.264544456641053e-07, + "loss": 0.8886, + "mean_token_accuracy": 0.7312352061271667, + "num_tokens": 43300603.0, + "step": 1689 + }, + { + "epoch": 0.18559191741708764, + "grad_norm": 2.53874135017395, + "learning_rate": 9.270032930845224e-07, + "loss": 1.0135, + "mean_token_accuracy": 0.6962611079216003, + "num_tokens": 43321907.0, + "step": 1690 + }, + { + "epoch": 0.18570173511970128, + "grad_norm": 2.262840509414673, + "learning_rate": 9.275521405049396e-07, + "loss": 1.0346, + "mean_token_accuracy": 0.6959080696105957, + "num_tokens": 43346884.0, + "step": 1691 + }, + { + "epoch": 0.18581155282231496, + "grad_norm": 2.5337331295013428, + "learning_rate": 9.281009879253567e-07, + "loss": 1.0117, + "mean_token_accuracy": 0.6922997832298279, + "num_tokens": 43369753.0, + "step": 1692 + }, + { + "epoch": 0.18592137052492863, + "grad_norm": 1.998267412185669, + "learning_rate": 9.286498353457738e-07, + "loss": 0.9662, + "mean_token_accuracy": 0.7090778350830078, + "num_tokens": 43400335.0, + "step": 1693 + }, + { + "epoch": 0.18603118822754228, + "grad_norm": 2.092276096343994, + "learning_rate": 9.29198682766191e-07, + "loss": 1.0163, + "mean_token_accuracy": 0.6941239833831787, + "num_tokens": 43429091.0, + "step": 1694 + }, + { + "epoch": 0.18614100593015595, + "grad_norm": 2.1876816749572754, + "learning_rate": 9.297475301866081e-07, + "loss": 1.0462, + "mean_token_accuracy": 0.6853457093238831, + "num_tokens": 43456815.0, + "step": 1695 + }, + { + "epoch": 0.1862508236327696, + "grad_norm": 2.5151941776275635, + "learning_rate": 9.302963776070252e-07, + "loss": 1.041, + "mean_token_accuracy": 0.6874086856842041, + "num_tokens": 43478892.0, + "step": 1696 + }, + { + "epoch": 0.18636064133538327, + "grad_norm": 2.4735679626464844, + "learning_rate": 9.308452250274424e-07, + "loss": 1.0442, + "mean_token_accuracy": 0.6924055814743042, + "num_tokens": 43501040.0, + "step": 1697 + }, + { + "epoch": 0.18647045903799692, + "grad_norm": 2.3680481910705566, + "learning_rate": 9.313940724478595e-07, + "loss": 1.0795, + "mean_token_accuracy": 0.6745267510414124, + "num_tokens": 43524721.0, + "step": 1698 + }, + { + "epoch": 0.1865802767406106, + "grad_norm": 2.5069973468780518, + "learning_rate": 9.319429198682766e-07, + "loss": 0.9834, + "mean_token_accuracy": 0.7027312517166138, + "num_tokens": 43545429.0, + "step": 1699 + }, + { + "epoch": 0.18669009444322424, + "grad_norm": 2.111858367919922, + "learning_rate": 9.324917672886937e-07, + "loss": 1.0821, + "mean_token_accuracy": 0.672663688659668, + "num_tokens": 43574252.0, + "step": 1700 + }, + { + "epoch": 0.1867999121458379, + "grad_norm": 2.286412239074707, + "learning_rate": 9.330406147091107e-07, + "loss": 1.0429, + "mean_token_accuracy": 0.6817986965179443, + "num_tokens": 43600578.0, + "step": 1701 + }, + { + "epoch": 0.18690972984845158, + "grad_norm": 2.337275743484497, + "learning_rate": 9.335894621295279e-07, + "loss": 1.0484, + "mean_token_accuracy": 0.6987984776496887, + "num_tokens": 43625345.0, + "step": 1702 + }, + { + "epoch": 0.18701954755106523, + "grad_norm": 2.179205894470215, + "learning_rate": 9.341383095499451e-07, + "loss": 1.0175, + "mean_token_accuracy": 0.6976006031036377, + "num_tokens": 43652401.0, + "step": 1703 + }, + { + "epoch": 0.1871293652536789, + "grad_norm": 2.535388231277466, + "learning_rate": 9.346871569703621e-07, + "loss": 0.9771, + "mean_token_accuracy": 0.7018896341323853, + "num_tokens": 43674363.0, + "step": 1704 + }, + { + "epoch": 0.18723918295629255, + "grad_norm": 2.3362255096435547, + "learning_rate": 9.352360043907793e-07, + "loss": 1.0895, + "mean_token_accuracy": 0.6808257699012756, + "num_tokens": 43697902.0, + "step": 1705 + }, + { + "epoch": 0.18734900065890622, + "grad_norm": 2.5230374336242676, + "learning_rate": 9.357848518111965e-07, + "loss": 0.9384, + "mean_token_accuracy": 0.7101998329162598, + "num_tokens": 43723270.0, + "step": 1706 + }, + { + "epoch": 0.18745881836151987, + "grad_norm": 2.2048470973968506, + "learning_rate": 9.363336992316135e-07, + "loss": 1.0329, + "mean_token_accuracy": 0.6850337982177734, + "num_tokens": 43752697.0, + "step": 1707 + }, + { + "epoch": 0.18756863606413354, + "grad_norm": 2.0298073291778564, + "learning_rate": 9.368825466520307e-07, + "loss": 0.9664, + "mean_token_accuracy": 0.7115091681480408, + "num_tokens": 43783625.0, + "step": 1708 + }, + { + "epoch": 0.1876784537667472, + "grad_norm": 2.553994655609131, + "learning_rate": 9.374313940724479e-07, + "loss": 1.0617, + "mean_token_accuracy": 0.6838904619216919, + "num_tokens": 43810488.0, + "step": 1709 + }, + { + "epoch": 0.18778827146936086, + "grad_norm": 2.92557692527771, + "learning_rate": 9.379802414928649e-07, + "loss": 0.9797, + "mean_token_accuracy": 0.7011801600456238, + "num_tokens": 43829053.0, + "step": 1710 + }, + { + "epoch": 0.18789808917197454, + "grad_norm": 2.2688872814178467, + "learning_rate": 9.385290889132821e-07, + "loss": 1.1101, + "mean_token_accuracy": 0.6681824326515198, + "num_tokens": 43859175.0, + "step": 1711 + }, + { + "epoch": 0.18800790687458818, + "grad_norm": 2.322281837463379, + "learning_rate": 9.390779363336992e-07, + "loss": 1.0562, + "mean_token_accuracy": 0.6854133009910583, + "num_tokens": 43885228.0, + "step": 1712 + }, + { + "epoch": 0.18811772457720186, + "grad_norm": 2.278850793838501, + "learning_rate": 9.396267837541163e-07, + "loss": 1.0435, + "mean_token_accuracy": 0.6880121231079102, + "num_tokens": 43911176.0, + "step": 1713 + }, + { + "epoch": 0.1882275422798155, + "grad_norm": 2.0205605030059814, + "learning_rate": 9.401756311745335e-07, + "loss": 1.047, + "mean_token_accuracy": 0.6940406560897827, + "num_tokens": 43944979.0, + "step": 1714 + }, + { + "epoch": 0.18833735998242918, + "grad_norm": 2.2483396530151367, + "learning_rate": 9.407244785949506e-07, + "loss": 1.0193, + "mean_token_accuracy": 0.6958689093589783, + "num_tokens": 43971206.0, + "step": 1715 + }, + { + "epoch": 0.18844717768504282, + "grad_norm": 2.1748554706573486, + "learning_rate": 9.412733260153677e-07, + "loss": 1.0721, + "mean_token_accuracy": 0.6815531849861145, + "num_tokens": 43999454.0, + "step": 1716 + }, + { + "epoch": 0.1885569953876565, + "grad_norm": 2.271761894226074, + "learning_rate": 9.418221734357849e-07, + "loss": 0.9422, + "mean_token_accuracy": 0.717515230178833, + "num_tokens": 44024072.0, + "step": 1717 + }, + { + "epoch": 0.18866681309027014, + "grad_norm": 2.052415132522583, + "learning_rate": 9.42371020856202e-07, + "loss": 1.0156, + "mean_token_accuracy": 0.6995034217834473, + "num_tokens": 44052962.0, + "step": 1718 + }, + { + "epoch": 0.18877663079288381, + "grad_norm": 2.589172840118408, + "learning_rate": 9.42919868276619e-07, + "loss": 1.0245, + "mean_token_accuracy": 0.6870189905166626, + "num_tokens": 44072249.0, + "step": 1719 + }, + { + "epoch": 0.18888644849549746, + "grad_norm": 2.0450432300567627, + "learning_rate": 9.434687156970362e-07, + "loss": 0.9719, + "mean_token_accuracy": 0.6931941509246826, + "num_tokens": 44100832.0, + "step": 1720 + }, + { + "epoch": 0.18899626619811113, + "grad_norm": 2.1628291606903076, + "learning_rate": 9.440175631174532e-07, + "loss": 1.0069, + "mean_token_accuracy": 0.6985822916030884, + "num_tokens": 44128345.0, + "step": 1721 + }, + { + "epoch": 0.1891060839007248, + "grad_norm": 2.186056613922119, + "learning_rate": 9.445664105378704e-07, + "loss": 1.0566, + "mean_token_accuracy": 0.6806696653366089, + "num_tokens": 44156001.0, + "step": 1722 + }, + { + "epoch": 0.18921590160333845, + "grad_norm": 2.0811893939971924, + "learning_rate": 9.451152579582875e-07, + "loss": 1.0325, + "mean_token_accuracy": 0.6942404508590698, + "num_tokens": 44185557.0, + "step": 1723 + }, + { + "epoch": 0.18932571930595213, + "grad_norm": 2.1409549713134766, + "learning_rate": 9.456641053787046e-07, + "loss": 0.9308, + "mean_token_accuracy": 0.7073658108711243, + "num_tokens": 44211467.0, + "step": 1724 + }, + { + "epoch": 0.18943553700856577, + "grad_norm": 2.3605763912200928, + "learning_rate": 9.462129527991218e-07, + "loss": 1.0513, + "mean_token_accuracy": 0.6883121132850647, + "num_tokens": 44235215.0, + "step": 1725 + }, + { + "epoch": 0.18954535471117945, + "grad_norm": 2.135236978530884, + "learning_rate": 9.467618002195389e-07, + "loss": 1.0876, + "mean_token_accuracy": 0.6882803440093994, + "num_tokens": 44264919.0, + "step": 1726 + }, + { + "epoch": 0.1896551724137931, + "grad_norm": 2.3142499923706055, + "learning_rate": 9.47310647639956e-07, + "loss": 0.9399, + "mean_token_accuracy": 0.7143688201904297, + "num_tokens": 44290883.0, + "step": 1727 + }, + { + "epoch": 0.18976499011640677, + "grad_norm": 2.2064826488494873, + "learning_rate": 9.478594950603732e-07, + "loss": 1.052, + "mean_token_accuracy": 0.6829659938812256, + "num_tokens": 44317301.0, + "step": 1728 + }, + { + "epoch": 0.1898748078190204, + "grad_norm": 2.4808883666992188, + "learning_rate": 9.484083424807903e-07, + "loss": 0.9843, + "mean_token_accuracy": 0.7006649374961853, + "num_tokens": 44339514.0, + "step": 1729 + }, + { + "epoch": 0.1899846255216341, + "grad_norm": 2.061530351638794, + "learning_rate": 9.489571899012074e-07, + "loss": 1.0514, + "mean_token_accuracy": 0.6887679696083069, + "num_tokens": 44370744.0, + "step": 1730 + }, + { + "epoch": 0.19009444322424776, + "grad_norm": 2.0067527294158936, + "learning_rate": 9.495060373216246e-07, + "loss": 0.9716, + "mean_token_accuracy": 0.7034304738044739, + "num_tokens": 44401111.0, + "step": 1731 + }, + { + "epoch": 0.1902042609268614, + "grad_norm": 2.254739761352539, + "learning_rate": 9.500548847420417e-07, + "loss": 0.9935, + "mean_token_accuracy": 0.6980063915252686, + "num_tokens": 44425808.0, + "step": 1732 + }, + { + "epoch": 0.19031407862947508, + "grad_norm": 2.4166717529296875, + "learning_rate": 9.506037321624588e-07, + "loss": 0.948, + "mean_token_accuracy": 0.7067198753356934, + "num_tokens": 44448008.0, + "step": 1733 + }, + { + "epoch": 0.19042389633208873, + "grad_norm": 2.4798805713653564, + "learning_rate": 9.511525795828759e-07, + "loss": 1.0377, + "mean_token_accuracy": 0.6912057399749756, + "num_tokens": 44471138.0, + "step": 1734 + }, + { + "epoch": 0.1905337140347024, + "grad_norm": 2.0065953731536865, + "learning_rate": 9.517014270032931e-07, + "loss": 1.1034, + "mean_token_accuracy": 0.6803048849105835, + "num_tokens": 44502965.0, + "step": 1735 + }, + { + "epoch": 0.19064353173731605, + "grad_norm": 2.5049002170562744, + "learning_rate": 9.522502744237102e-07, + "loss": 0.9516, + "mean_token_accuracy": 0.7113103270530701, + "num_tokens": 44524689.0, + "step": 1736 + }, + { + "epoch": 0.19075334943992972, + "grad_norm": 1.993895411491394, + "learning_rate": 9.527991218441273e-07, + "loss": 1.0682, + "mean_token_accuracy": 0.6812386512756348, + "num_tokens": 44558688.0, + "step": 1737 + }, + { + "epoch": 0.19086316714254337, + "grad_norm": 2.0455148220062256, + "learning_rate": 9.533479692645444e-07, + "loss": 1.0846, + "mean_token_accuracy": 0.676051914691925, + "num_tokens": 44591530.0, + "step": 1738 + }, + { + "epoch": 0.19097298484515704, + "grad_norm": 2.397639036178589, + "learning_rate": 9.538968166849616e-07, + "loss": 0.928, + "mean_token_accuracy": 0.7164249420166016, + "num_tokens": 44615107.0, + "step": 1739 + }, + { + "epoch": 0.1910828025477707, + "grad_norm": 2.5631091594696045, + "learning_rate": 9.544456641053787e-07, + "loss": 0.9591, + "mean_token_accuracy": 0.7109991312026978, + "num_tokens": 44634490.0, + "step": 1740 + }, + { + "epoch": 0.19119262025038436, + "grad_norm": 2.2434051036834717, + "learning_rate": 9.549945115257958e-07, + "loss": 1.0901, + "mean_token_accuracy": 0.67569899559021, + "num_tokens": 44663101.0, + "step": 1741 + }, + { + "epoch": 0.19130243795299803, + "grad_norm": 2.2562077045440674, + "learning_rate": 9.55543358946213e-07, + "loss": 0.9795, + "mean_token_accuracy": 0.7013290524482727, + "num_tokens": 44688867.0, + "step": 1742 + }, + { + "epoch": 0.19141225565561168, + "grad_norm": 2.007253646850586, + "learning_rate": 9.5609220636663e-07, + "loss": 1.0365, + "mean_token_accuracy": 0.6839882731437683, + "num_tokens": 44720902.0, + "step": 1743 + }, + { + "epoch": 0.19152207335822535, + "grad_norm": 2.3228297233581543, + "learning_rate": 9.566410537870472e-07, + "loss": 1.078, + "mean_token_accuracy": 0.6752594709396362, + "num_tokens": 44745590.0, + "step": 1744 + }, + { + "epoch": 0.191631891060839, + "grad_norm": 2.1206910610198975, + "learning_rate": 9.571899012074642e-07, + "loss": 1.0853, + "mean_token_accuracy": 0.6735995411872864, + "num_tokens": 44776677.0, + "step": 1745 + }, + { + "epoch": 0.19174170876345267, + "grad_norm": 2.3018624782562256, + "learning_rate": 9.577387486278815e-07, + "loss": 1.0118, + "mean_token_accuracy": 0.6934711337089539, + "num_tokens": 44800941.0, + "step": 1746 + }, + { + "epoch": 0.19185152646606632, + "grad_norm": 2.378575563430786, + "learning_rate": 9.582875960482986e-07, + "loss": 0.911, + "mean_token_accuracy": 0.7179555892944336, + "num_tokens": 44823551.0, + "step": 1747 + }, + { + "epoch": 0.19196134416868, + "grad_norm": 2.1777589321136475, + "learning_rate": 9.588364434687156e-07, + "loss": 1.0936, + "mean_token_accuracy": 0.6831318736076355, + "num_tokens": 44851576.0, + "step": 1748 + }, + { + "epoch": 0.19207116187129367, + "grad_norm": 2.5438005924224854, + "learning_rate": 9.593852908891327e-07, + "loss": 0.9091, + "mean_token_accuracy": 0.7166467905044556, + "num_tokens": 44871754.0, + "step": 1749 + }, + { + "epoch": 0.1921809795739073, + "grad_norm": 2.265469789505005, + "learning_rate": 9.5993413830955e-07, + "loss": 0.9871, + "mean_token_accuracy": 0.6986215114593506, + "num_tokens": 44895662.0, + "step": 1750 + }, + { + "epoch": 0.19229079727652099, + "grad_norm": 2.440199136734009, + "learning_rate": 9.60482985729967e-07, + "loss": 1.0585, + "mean_token_accuracy": 0.6841825246810913, + "num_tokens": 44918127.0, + "step": 1751 + }, + { + "epoch": 0.19240061497913463, + "grad_norm": 2.2527248859405518, + "learning_rate": 9.61031833150384e-07, + "loss": 1.0042, + "mean_token_accuracy": 0.6969274282455444, + "num_tokens": 44944316.0, + "step": 1752 + }, + { + "epoch": 0.1925104326817483, + "grad_norm": 2.5473930835723877, + "learning_rate": 9.615806805708014e-07, + "loss": 1.0283, + "mean_token_accuracy": 0.695488691329956, + "num_tokens": 44967110.0, + "step": 1753 + }, + { + "epoch": 0.19262025038436195, + "grad_norm": 2.0690412521362305, + "learning_rate": 9.621295279912184e-07, + "loss": 1.0842, + "mean_token_accuracy": 0.6768261194229126, + "num_tokens": 44996148.0, + "step": 1754 + }, + { + "epoch": 0.19273006808697563, + "grad_norm": 2.3098840713500977, + "learning_rate": 9.626783754116355e-07, + "loss": 0.9983, + "mean_token_accuracy": 0.7042895555496216, + "num_tokens": 45026046.0, + "step": 1755 + }, + { + "epoch": 0.19283988578958927, + "grad_norm": 2.6050148010253906, + "learning_rate": 9.632272228320525e-07, + "loss": 0.9613, + "mean_token_accuracy": 0.7061463594436646, + "num_tokens": 45045757.0, + "step": 1756 + }, + { + "epoch": 0.19294970349220295, + "grad_norm": 2.097996950149536, + "learning_rate": 9.637760702524698e-07, + "loss": 1.0626, + "mean_token_accuracy": 0.683322012424469, + "num_tokens": 45074844.0, + "step": 1757 + }, + { + "epoch": 0.1930595211948166, + "grad_norm": 2.380392551422119, + "learning_rate": 9.643249176728869e-07, + "loss": 1.034, + "mean_token_accuracy": 0.6884952783584595, + "num_tokens": 45100253.0, + "step": 1758 + }, + { + "epoch": 0.19316933889743026, + "grad_norm": 2.3962478637695312, + "learning_rate": 9.64873765093304e-07, + "loss": 1.0426, + "mean_token_accuracy": 0.6862583160400391, + "num_tokens": 45127267.0, + "step": 1759 + }, + { + "epoch": 0.19327915660004394, + "grad_norm": 2.194477081298828, + "learning_rate": 9.654226125137212e-07, + "loss": 1.0404, + "mean_token_accuracy": 0.6883280277252197, + "num_tokens": 45154703.0, + "step": 1760 + }, + { + "epoch": 0.19338897430265758, + "grad_norm": 2.2523064613342285, + "learning_rate": 9.659714599341383e-07, + "loss": 1.1134, + "mean_token_accuracy": 0.6801418662071228, + "num_tokens": 45183860.0, + "step": 1761 + }, + { + "epoch": 0.19349879200527126, + "grad_norm": 2.0044004917144775, + "learning_rate": 9.665203073545553e-07, + "loss": 1.1062, + "mean_token_accuracy": 0.6752561330795288, + "num_tokens": 45216083.0, + "step": 1762 + }, + { + "epoch": 0.1936086097078849, + "grad_norm": 2.441490888595581, + "learning_rate": 9.670691547749726e-07, + "loss": 1.0612, + "mean_token_accuracy": 0.6768019199371338, + "num_tokens": 45241691.0, + "step": 1763 + }, + { + "epoch": 0.19371842741049858, + "grad_norm": 2.2885215282440186, + "learning_rate": 9.676180021953897e-07, + "loss": 1.0219, + "mean_token_accuracy": 0.6932134032249451, + "num_tokens": 45269277.0, + "step": 1764 + }, + { + "epoch": 0.19382824511311222, + "grad_norm": 2.3087730407714844, + "learning_rate": 9.681668496158067e-07, + "loss": 0.9549, + "mean_token_accuracy": 0.7131921648979187, + "num_tokens": 45293249.0, + "step": 1765 + }, + { + "epoch": 0.1939380628157259, + "grad_norm": 2.6486082077026367, + "learning_rate": 9.68715697036224e-07, + "loss": 1.022, + "mean_token_accuracy": 0.6913659572601318, + "num_tokens": 45313002.0, + "step": 1766 + }, + { + "epoch": 0.19404788051833954, + "grad_norm": 2.777827501296997, + "learning_rate": 9.692645444566409e-07, + "loss": 1.04, + "mean_token_accuracy": 0.695953369140625, + "num_tokens": 45332061.0, + "step": 1767 + }, + { + "epoch": 0.19415769822095322, + "grad_norm": 2.3547134399414062, + "learning_rate": 9.698133918770581e-07, + "loss": 1.0299, + "mean_token_accuracy": 0.6928379535675049, + "num_tokens": 45355387.0, + "step": 1768 + }, + { + "epoch": 0.1942675159235669, + "grad_norm": 2.3649840354919434, + "learning_rate": 9.703622392974752e-07, + "loss": 1.0473, + "mean_token_accuracy": 0.6896886229515076, + "num_tokens": 45378421.0, + "step": 1769 + }, + { + "epoch": 0.19437733362618054, + "grad_norm": 2.289546012878418, + "learning_rate": 9.709110867178923e-07, + "loss": 0.9666, + "mean_token_accuracy": 0.706663966178894, + "num_tokens": 45401686.0, + "step": 1770 + }, + { + "epoch": 0.1944871513287942, + "grad_norm": 2.4010558128356934, + "learning_rate": 9.714599341383095e-07, + "loss": 1.0814, + "mean_token_accuracy": 0.6815705895423889, + "num_tokens": 45425922.0, + "step": 1771 + }, + { + "epoch": 0.19459696903140786, + "grad_norm": 2.0392868518829346, + "learning_rate": 9.720087815587266e-07, + "loss": 0.991, + "mean_token_accuracy": 0.7036350965499878, + "num_tokens": 45455891.0, + "step": 1772 + }, + { + "epoch": 0.19470678673402153, + "grad_norm": 2.6401987075805664, + "learning_rate": 9.725576289791437e-07, + "loss": 1.056, + "mean_token_accuracy": 0.6802186965942383, + "num_tokens": 45476931.0, + "step": 1773 + }, + { + "epoch": 0.19481660443663518, + "grad_norm": 2.2323012351989746, + "learning_rate": 9.73106476399561e-07, + "loss": 1.0105, + "mean_token_accuracy": 0.6906909942626953, + "num_tokens": 45503798.0, + "step": 1774 + }, + { + "epoch": 0.19492642213924885, + "grad_norm": 2.333603858947754, + "learning_rate": 9.73655323819978e-07, + "loss": 1.0129, + "mean_token_accuracy": 0.6920074224472046, + "num_tokens": 45530136.0, + "step": 1775 + }, + { + "epoch": 0.1950362398418625, + "grad_norm": 2.499377965927124, + "learning_rate": 9.74204171240395e-07, + "loss": 0.9828, + "mean_token_accuracy": 0.7033802270889282, + "num_tokens": 45551947.0, + "step": 1776 + }, + { + "epoch": 0.19514605754447617, + "grad_norm": 1.9529868364334106, + "learning_rate": 9.747530186608123e-07, + "loss": 1.1144, + "mean_token_accuracy": 0.6660401821136475, + "num_tokens": 45588034.0, + "step": 1777 + }, + { + "epoch": 0.19525587524708984, + "grad_norm": 2.1107239723205566, + "learning_rate": 9.753018660812294e-07, + "loss": 1.1067, + "mean_token_accuracy": 0.667898416519165, + "num_tokens": 45617014.0, + "step": 1778 + }, + { + "epoch": 0.1953656929497035, + "grad_norm": 2.303509473800659, + "learning_rate": 9.758507135016465e-07, + "loss": 1.0171, + "mean_token_accuracy": 0.6913744211196899, + "num_tokens": 45641662.0, + "step": 1779 + }, + { + "epoch": 0.19547551065231716, + "grad_norm": 2.2457213401794434, + "learning_rate": 9.763995609220637e-07, + "loss": 1.058, + "mean_token_accuracy": 0.6804040670394897, + "num_tokens": 45667699.0, + "step": 1780 + }, + { + "epoch": 0.1955853283549308, + "grad_norm": 2.08526349067688, + "learning_rate": 9.769484083424808e-07, + "loss": 1.0422, + "mean_token_accuracy": 0.6892538070678711, + "num_tokens": 45698090.0, + "step": 1781 + }, + { + "epoch": 0.19569514605754448, + "grad_norm": 2.3182337284088135, + "learning_rate": 9.774972557628979e-07, + "loss": 1.1216, + "mean_token_accuracy": 0.6698352098464966, + "num_tokens": 45723295.0, + "step": 1782 + }, + { + "epoch": 0.19580496376015813, + "grad_norm": 2.1907448768615723, + "learning_rate": 9.780461031833151e-07, + "loss": 1.0157, + "mean_token_accuracy": 0.6951195001602173, + "num_tokens": 45750635.0, + "step": 1783 + }, + { + "epoch": 0.1959147814627718, + "grad_norm": 2.641176223754883, + "learning_rate": 9.785949506037322e-07, + "loss": 0.9877, + "mean_token_accuracy": 0.6969117522239685, + "num_tokens": 45770945.0, + "step": 1784 + }, + { + "epoch": 0.19602459916538545, + "grad_norm": 2.0643293857574463, + "learning_rate": 9.791437980241493e-07, + "loss": 0.9901, + "mean_token_accuracy": 0.7042144536972046, + "num_tokens": 45799768.0, + "step": 1785 + }, + { + "epoch": 0.19613441686799912, + "grad_norm": 2.516240119934082, + "learning_rate": 9.796926454445663e-07, + "loss": 1.0261, + "mean_token_accuracy": 0.6888929605484009, + "num_tokens": 45821868.0, + "step": 1786 + }, + { + "epoch": 0.1962442345706128, + "grad_norm": 2.5245718955993652, + "learning_rate": 9.802414928649834e-07, + "loss": 1.0285, + "mean_token_accuracy": 0.6887670755386353, + "num_tokens": 45844280.0, + "step": 1787 + }, + { + "epoch": 0.19635405227322644, + "grad_norm": 2.252725839614868, + "learning_rate": 9.807903402854007e-07, + "loss": 0.9505, + "mean_token_accuracy": 0.7191003561019897, + "num_tokens": 45870713.0, + "step": 1788 + }, + { + "epoch": 0.19646386997584012, + "grad_norm": 2.075179100036621, + "learning_rate": 9.813391877058177e-07, + "loss": 1.0577, + "mean_token_accuracy": 0.6855546236038208, + "num_tokens": 45902631.0, + "step": 1789 + }, + { + "epoch": 0.19657368767845376, + "grad_norm": 2.3974220752716064, + "learning_rate": 9.818880351262348e-07, + "loss": 0.94, + "mean_token_accuracy": 0.7110073566436768, + "num_tokens": 45924409.0, + "step": 1790 + }, + { + "epoch": 0.19668350538106744, + "grad_norm": 2.3537604808807373, + "learning_rate": 9.82436882546652e-07, + "loss": 1.0729, + "mean_token_accuracy": 0.6773549914360046, + "num_tokens": 45949665.0, + "step": 1791 + }, + { + "epoch": 0.19679332308368108, + "grad_norm": 2.2280750274658203, + "learning_rate": 9.829857299670691e-07, + "loss": 1.0321, + "mean_token_accuracy": 0.6947826147079468, + "num_tokens": 45976721.0, + "step": 1792 + }, + { + "epoch": 0.19690314078629476, + "grad_norm": 2.331806182861328, + "learning_rate": 9.835345773874862e-07, + "loss": 0.9727, + "mean_token_accuracy": 0.7045990228652954, + "num_tokens": 46001128.0, + "step": 1793 + }, + { + "epoch": 0.1970129584889084, + "grad_norm": 2.1534297466278076, + "learning_rate": 9.840834248079035e-07, + "loss": 0.9911, + "mean_token_accuracy": 0.7101663947105408, + "num_tokens": 46031487.0, + "step": 1794 + }, + { + "epoch": 0.19712277619152208, + "grad_norm": 2.54717755317688, + "learning_rate": 9.846322722283205e-07, + "loss": 0.9316, + "mean_token_accuracy": 0.7175167798995972, + "num_tokens": 46052618.0, + "step": 1795 + }, + { + "epoch": 0.19723259389413572, + "grad_norm": 2.325279474258423, + "learning_rate": 9.851811196487376e-07, + "loss": 1.0563, + "mean_token_accuracy": 0.6829883456230164, + "num_tokens": 46076690.0, + "step": 1796 + }, + { + "epoch": 0.1973424115967494, + "grad_norm": 2.5254337787628174, + "learning_rate": 9.857299670691546e-07, + "loss": 1.0374, + "mean_token_accuracy": 0.6848822832107544, + "num_tokens": 46097322.0, + "step": 1797 + }, + { + "epoch": 0.19745222929936307, + "grad_norm": 2.159651041030884, + "learning_rate": 9.86278814489572e-07, + "loss": 0.9785, + "mean_token_accuracy": 0.7053061723709106, + "num_tokens": 46123132.0, + "step": 1798 + }, + { + "epoch": 0.19756204700197671, + "grad_norm": 2.0261547565460205, + "learning_rate": 9.86827661909989e-07, + "loss": 1.0817, + "mean_token_accuracy": 0.6779940128326416, + "num_tokens": 46154196.0, + "step": 1799 + }, + { + "epoch": 0.1976718647045904, + "grad_norm": 2.24766206741333, + "learning_rate": 9.87376509330406e-07, + "loss": 1.0039, + "mean_token_accuracy": 0.6956757307052612, + "num_tokens": 46181704.0, + "step": 1800 + }, + { + "epoch": 0.19778168240720403, + "grad_norm": 2.407432794570923, + "learning_rate": 9.879253567508233e-07, + "loss": 1.0129, + "mean_token_accuracy": 0.6943512558937073, + "num_tokens": 46205764.0, + "step": 1801 + }, + { + "epoch": 0.1978915001098177, + "grad_norm": 2.471642255783081, + "learning_rate": 9.884742041712404e-07, + "loss": 0.9661, + "mean_token_accuracy": 0.7089533805847168, + "num_tokens": 46225775.0, + "step": 1802 + }, + { + "epoch": 0.19800131781243135, + "grad_norm": 2.277858018875122, + "learning_rate": 9.890230515916574e-07, + "loss": 1.012, + "mean_token_accuracy": 0.6956068277359009, + "num_tokens": 46251344.0, + "step": 1803 + }, + { + "epoch": 0.19811113551504503, + "grad_norm": 2.31412935256958, + "learning_rate": 9.895718990120747e-07, + "loss": 0.9686, + "mean_token_accuracy": 0.7055704593658447, + "num_tokens": 46277666.0, + "step": 1804 + }, + { + "epoch": 0.19822095321765867, + "grad_norm": 2.2847888469696045, + "learning_rate": 9.901207464324918e-07, + "loss": 0.9791, + "mean_token_accuracy": 0.708509087562561, + "num_tokens": 46302412.0, + "step": 1805 + }, + { + "epoch": 0.19833077092027235, + "grad_norm": 2.2795629501342773, + "learning_rate": 9.906695938529088e-07, + "loss": 0.9576, + "mean_token_accuracy": 0.7165923714637756, + "num_tokens": 46327275.0, + "step": 1806 + }, + { + "epoch": 0.19844058862288602, + "grad_norm": 2.090034246444702, + "learning_rate": 9.91218441273326e-07, + "loss": 0.9963, + "mean_token_accuracy": 0.6977969408035278, + "num_tokens": 46357031.0, + "step": 1807 + }, + { + "epoch": 0.19855040632549967, + "grad_norm": 2.3570780754089355, + "learning_rate": 9.91767288693743e-07, + "loss": 1.1051, + "mean_token_accuracy": 0.6915499567985535, + "num_tokens": 46383721.0, + "step": 1808 + }, + { + "epoch": 0.19866022402811334, + "grad_norm": 2.565809965133667, + "learning_rate": 9.923161361141602e-07, + "loss": 1.0398, + "mean_token_accuracy": 0.6867607831954956, + "num_tokens": 46405506.0, + "step": 1809 + }, + { + "epoch": 0.198770041730727, + "grad_norm": 2.340137243270874, + "learning_rate": 9.928649835345773e-07, + "loss": 1.0323, + "mean_token_accuracy": 0.691371500492096, + "num_tokens": 46432307.0, + "step": 1810 + }, + { + "epoch": 0.19887985943334066, + "grad_norm": 2.3450965881347656, + "learning_rate": 9.934138309549944e-07, + "loss": 1.0391, + "mean_token_accuracy": 0.6843547821044922, + "num_tokens": 46458511.0, + "step": 1811 + }, + { + "epoch": 0.1989896771359543, + "grad_norm": 2.523714065551758, + "learning_rate": 9.939626783754116e-07, + "loss": 0.964, + "mean_token_accuracy": 0.709763765335083, + "num_tokens": 46480290.0, + "step": 1812 + }, + { + "epoch": 0.19909949483856798, + "grad_norm": 2.275003433227539, + "learning_rate": 9.945115257958287e-07, + "loss": 1.071, + "mean_token_accuracy": 0.6842846870422363, + "num_tokens": 46505961.0, + "step": 1813 + }, + { + "epoch": 0.19920931254118163, + "grad_norm": 2.3137125968933105, + "learning_rate": 9.950603732162458e-07, + "loss": 0.9816, + "mean_token_accuracy": 0.7074854373931885, + "num_tokens": 46533032.0, + "step": 1814 + }, + { + "epoch": 0.1993191302437953, + "grad_norm": 2.2345545291900635, + "learning_rate": 9.95609220636663e-07, + "loss": 1.1537, + "mean_token_accuracy": 0.6608373522758484, + "num_tokens": 46559697.0, + "step": 1815 + }, + { + "epoch": 0.19942894794640897, + "grad_norm": 2.746727228164673, + "learning_rate": 9.9615806805708e-07, + "loss": 1.0359, + "mean_token_accuracy": 0.691417396068573, + "num_tokens": 46583810.0, + "step": 1816 + }, + { + "epoch": 0.19953876564902262, + "grad_norm": 2.9034218788146973, + "learning_rate": 9.967069154774972e-07, + "loss": 0.9972, + "mean_token_accuracy": 0.7044723033905029, + "num_tokens": 46600475.0, + "step": 1817 + }, + { + "epoch": 0.1996485833516363, + "grad_norm": 2.221031904220581, + "learning_rate": 9.972557628979144e-07, + "loss": 1.0918, + "mean_token_accuracy": 0.6791951656341553, + "num_tokens": 46632831.0, + "step": 1818 + }, + { + "epoch": 0.19975840105424994, + "grad_norm": 2.622307777404785, + "learning_rate": 9.978046103183315e-07, + "loss": 0.9267, + "mean_token_accuracy": 0.7118135094642639, + "num_tokens": 46651755.0, + "step": 1819 + }, + { + "epoch": 0.1998682187568636, + "grad_norm": 2.1758620738983154, + "learning_rate": 9.983534577387486e-07, + "loss": 1.0399, + "mean_token_accuracy": 0.6866313219070435, + "num_tokens": 46681147.0, + "step": 1820 + }, + { + "epoch": 0.19997803645947726, + "grad_norm": 2.2621264457702637, + "learning_rate": 9.989023051591658e-07, + "loss": 1.0336, + "mean_token_accuracy": 0.6927070617675781, + "num_tokens": 46706418.0, + "step": 1821 + }, + { + "epoch": 0.20008785416209093, + "grad_norm": 1.9662445783615112, + "learning_rate": 9.994511525795829e-07, + "loss": 1.093, + "mean_token_accuracy": 0.6710301637649536, + "num_tokens": 46738512.0, + "step": 1822 + }, + { + "epoch": 0.20019767186470458, + "grad_norm": 2.3120687007904053, + "learning_rate": 1e-06, + "loss": 1.1365, + "mean_token_accuracy": 0.6644891500473022, + "num_tokens": 46763726.0, + "step": 1823 + }, + { + "epoch": 0.20030748956731825, + "grad_norm": 2.485074758529663, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6991680860519409, + "num_tokens": 46785439.0, + "step": 1824 + }, + { + "epoch": 0.2004173072699319, + "grad_norm": 2.6051957607269287, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7166796922683716, + "num_tokens": 46804312.0, + "step": 1825 + }, + { + "epoch": 0.20052712497254557, + "grad_norm": 2.3170166015625, + "learning_rate": 1e-06, + "loss": 1.0855, + "mean_token_accuracy": 0.6792519092559814, + "num_tokens": 46831537.0, + "step": 1826 + }, + { + "epoch": 0.20063694267515925, + "grad_norm": 2.1554791927337646, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6968163251876831, + "num_tokens": 46860126.0, + "step": 1827 + }, + { + "epoch": 0.2007467603777729, + "grad_norm": 2.3295769691467285, + "learning_rate": 1e-06, + "loss": 1.0244, + "mean_token_accuracy": 0.6941146850585938, + "num_tokens": 46884250.0, + "step": 1828 + }, + { + "epoch": 0.20085657808038657, + "grad_norm": 2.1394057273864746, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7116543054580688, + "num_tokens": 46910895.0, + "step": 1829 + }, + { + "epoch": 0.2009663957830002, + "grad_norm": 2.5472898483276367, + "learning_rate": 1e-06, + "loss": 1.0946, + "mean_token_accuracy": 0.6763050556182861, + "num_tokens": 46933984.0, + "step": 1830 + }, + { + "epoch": 0.20107621348561389, + "grad_norm": 1.957506537437439, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7070109844207764, + "num_tokens": 46964599.0, + "step": 1831 + }, + { + "epoch": 0.20118603118822753, + "grad_norm": 2.4778079986572266, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7180216908454895, + "num_tokens": 46985436.0, + "step": 1832 + }, + { + "epoch": 0.2012958488908412, + "grad_norm": 2.2751920223236084, + "learning_rate": 1e-06, + "loss": 1.0473, + "mean_token_accuracy": 0.6890791058540344, + "num_tokens": 47012085.0, + "step": 1833 + }, + { + "epoch": 0.20140566659345485, + "grad_norm": 2.7413649559020996, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.7006075382232666, + "num_tokens": 47031178.0, + "step": 1834 + }, + { + "epoch": 0.20151548429606853, + "grad_norm": 2.2443385124206543, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.7007827758789062, + "num_tokens": 47060143.0, + "step": 1835 + }, + { + "epoch": 0.2016253019986822, + "grad_norm": 2.633084774017334, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.7000386118888855, + "num_tokens": 47080094.0, + "step": 1836 + }, + { + "epoch": 0.20173511970129585, + "grad_norm": 2.134051561355591, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.7022310495376587, + "num_tokens": 47108330.0, + "step": 1837 + }, + { + "epoch": 0.20184493740390952, + "grad_norm": 2.164771556854248, + "learning_rate": 1e-06, + "loss": 1.0608, + "mean_token_accuracy": 0.7054588794708252, + "num_tokens": 47136491.0, + "step": 1838 + }, + { + "epoch": 0.20195475510652316, + "grad_norm": 2.4717442989349365, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7309930324554443, + "num_tokens": 47156766.0, + "step": 1839 + }, + { + "epoch": 0.20206457280913684, + "grad_norm": 2.686580181121826, + "learning_rate": 1e-06, + "loss": 1.054, + "mean_token_accuracy": 0.6873010993003845, + "num_tokens": 47177986.0, + "step": 1840 + }, + { + "epoch": 0.20217439051175048, + "grad_norm": 2.4842336177825928, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.6919023990631104, + "num_tokens": 47199426.0, + "step": 1841 + }, + { + "epoch": 0.20228420821436416, + "grad_norm": 2.336819887161255, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7096408605575562, + "num_tokens": 47222975.0, + "step": 1842 + }, + { + "epoch": 0.2023940259169778, + "grad_norm": 2.3373641967773438, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.7077624797821045, + "num_tokens": 47247147.0, + "step": 1843 + }, + { + "epoch": 0.20250384361959148, + "grad_norm": 2.2861618995666504, + "learning_rate": 1e-06, + "loss": 1.0995, + "mean_token_accuracy": 0.6749222278594971, + "num_tokens": 47273039.0, + "step": 1844 + }, + { + "epoch": 0.20261366132220515, + "grad_norm": 2.4514429569244385, + "learning_rate": 1e-06, + "loss": 1.0417, + "mean_token_accuracy": 0.6860672235488892, + "num_tokens": 47294710.0, + "step": 1845 + }, + { + "epoch": 0.2027234790248188, + "grad_norm": 2.0725581645965576, + "learning_rate": 1e-06, + "loss": 1.0749, + "mean_token_accuracy": 0.6737958192825317, + "num_tokens": 47324302.0, + "step": 1846 + }, + { + "epoch": 0.20283329672743247, + "grad_norm": 2.088717460632324, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7153208255767822, + "num_tokens": 47350257.0, + "step": 1847 + }, + { + "epoch": 0.20294311443004612, + "grad_norm": 2.524158239364624, + "learning_rate": 1e-06, + "loss": 1.0451, + "mean_token_accuracy": 0.690074622631073, + "num_tokens": 47371901.0, + "step": 1848 + }, + { + "epoch": 0.2030529321326598, + "grad_norm": 2.4384119510650635, + "learning_rate": 1e-06, + "loss": 1.111, + "mean_token_accuracy": 0.6702181696891785, + "num_tokens": 47396337.0, + "step": 1849 + }, + { + "epoch": 0.20316274983527344, + "grad_norm": 2.5067150592803955, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6919283270835876, + "num_tokens": 47417519.0, + "step": 1850 + }, + { + "epoch": 0.2032725675378871, + "grad_norm": 2.492353677749634, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7190941572189331, + "num_tokens": 47438494.0, + "step": 1851 + }, + { + "epoch": 0.20338238524050076, + "grad_norm": 2.0932016372680664, + "learning_rate": 1e-06, + "loss": 1.065, + "mean_token_accuracy": 0.680422842502594, + "num_tokens": 47465230.0, + "step": 1852 + }, + { + "epoch": 0.20349220294311443, + "grad_norm": 2.165043592453003, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.697799801826477, + "num_tokens": 47493708.0, + "step": 1853 + }, + { + "epoch": 0.2036020206457281, + "grad_norm": 2.2407124042510986, + "learning_rate": 1e-06, + "loss": 1.0681, + "mean_token_accuracy": 0.6798239946365356, + "num_tokens": 47521237.0, + "step": 1854 + }, + { + "epoch": 0.20371183834834175, + "grad_norm": 2.2121329307556152, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6979211568832397, + "num_tokens": 47548746.0, + "step": 1855 + }, + { + "epoch": 0.20382165605095542, + "grad_norm": 2.10239315032959, + "learning_rate": 1e-06, + "loss": 1.0971, + "mean_token_accuracy": 0.6739606261253357, + "num_tokens": 47577919.0, + "step": 1856 + }, + { + "epoch": 0.20393147375356907, + "grad_norm": 2.0552902221679688, + "learning_rate": 1e-06, + "loss": 1.0922, + "mean_token_accuracy": 0.6857494115829468, + "num_tokens": 47610047.0, + "step": 1857 + }, + { + "epoch": 0.20404129145618274, + "grad_norm": 2.3654088973999023, + "learning_rate": 1e-06, + "loss": 1.0351, + "mean_token_accuracy": 0.6947007775306702, + "num_tokens": 47633775.0, + "step": 1858 + }, + { + "epoch": 0.2041511091587964, + "grad_norm": 2.313352346420288, + "learning_rate": 1e-06, + "loss": 1.0597, + "mean_token_accuracy": 0.6816527843475342, + "num_tokens": 47660728.0, + "step": 1859 + }, + { + "epoch": 0.20426092686141006, + "grad_norm": 2.418107032775879, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7142105102539062, + "num_tokens": 47683553.0, + "step": 1860 + }, + { + "epoch": 0.2043707445640237, + "grad_norm": 2.420201063156128, + "learning_rate": 1e-06, + "loss": 1.0458, + "mean_token_accuracy": 0.6899716258049011, + "num_tokens": 47706283.0, + "step": 1861 + }, + { + "epoch": 0.20448056226663738, + "grad_norm": 2.4258909225463867, + "learning_rate": 1e-06, + "loss": 1.0782, + "mean_token_accuracy": 0.6784303784370422, + "num_tokens": 47730389.0, + "step": 1862 + }, + { + "epoch": 0.20459037996925103, + "grad_norm": 2.3300352096557617, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.7071202993392944, + "num_tokens": 47754267.0, + "step": 1863 + }, + { + "epoch": 0.2047001976718647, + "grad_norm": 2.108994245529175, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6974079608917236, + "num_tokens": 47784817.0, + "step": 1864 + }, + { + "epoch": 0.20481001537447838, + "grad_norm": 2.055316209793091, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7117025256156921, + "num_tokens": 47812901.0, + "step": 1865 + }, + { + "epoch": 0.20491983307709202, + "grad_norm": 2.3718626499176025, + "learning_rate": 1e-06, + "loss": 1.0876, + "mean_token_accuracy": 0.674240231513977, + "num_tokens": 47836692.0, + "step": 1866 + }, + { + "epoch": 0.2050296507797057, + "grad_norm": 2.7503387928009033, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7040201425552368, + "num_tokens": 47854892.0, + "step": 1867 + }, + { + "epoch": 0.20513946848231934, + "grad_norm": 2.263584613800049, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.6984476447105408, + "num_tokens": 47881575.0, + "step": 1868 + }, + { + "epoch": 0.20524928618493302, + "grad_norm": 2.2438251972198486, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.6999492049217224, + "num_tokens": 47905553.0, + "step": 1869 + }, + { + "epoch": 0.20535910388754666, + "grad_norm": 2.1765708923339844, + "learning_rate": 1e-06, + "loss": 1.0518, + "mean_token_accuracy": 0.687527060508728, + "num_tokens": 47936013.0, + "step": 1870 + }, + { + "epoch": 0.20546892159016034, + "grad_norm": 2.3440322875976562, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6964861154556274, + "num_tokens": 47962377.0, + "step": 1871 + }, + { + "epoch": 0.20557873929277398, + "grad_norm": 2.4975180625915527, + "learning_rate": 1e-06, + "loss": 1.0351, + "mean_token_accuracy": 0.6865421533584595, + "num_tokens": 47986268.0, + "step": 1872 + }, + { + "epoch": 0.20568855699538766, + "grad_norm": 2.07589054107666, + "learning_rate": 1e-06, + "loss": 1.0734, + "mean_token_accuracy": 0.6769613027572632, + "num_tokens": 48018586.0, + "step": 1873 + }, + { + "epoch": 0.20579837469800133, + "grad_norm": 1.9979041814804077, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.6990939378738403, + "num_tokens": 48051557.0, + "step": 1874 + }, + { + "epoch": 0.20590819240061498, + "grad_norm": 2.1683309078216553, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.711380124092102, + "num_tokens": 48079458.0, + "step": 1875 + }, + { + "epoch": 0.20601801010322865, + "grad_norm": 2.1794776916503906, + "learning_rate": 1e-06, + "loss": 1.0408, + "mean_token_accuracy": 0.6852594017982483, + "num_tokens": 48107917.0, + "step": 1876 + }, + { + "epoch": 0.2061278278058423, + "grad_norm": 2.5314788818359375, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7158206701278687, + "num_tokens": 48128093.0, + "step": 1877 + }, + { + "epoch": 0.20623764550845597, + "grad_norm": 2.0155293941497803, + "learning_rate": 1e-06, + "loss": 1.0993, + "mean_token_accuracy": 0.6713372468948364, + "num_tokens": 48159506.0, + "step": 1878 + }, + { + "epoch": 0.20634746321106961, + "grad_norm": 2.639070510864258, + "learning_rate": 1e-06, + "loss": 1.0165, + "mean_token_accuracy": 0.6920919418334961, + "num_tokens": 48183329.0, + "step": 1879 + }, + { + "epoch": 0.2064572809136833, + "grad_norm": 2.272857666015625, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.701287031173706, + "num_tokens": 48208203.0, + "step": 1880 + }, + { + "epoch": 0.20656709861629693, + "grad_norm": 2.1681697368621826, + "learning_rate": 1e-06, + "loss": 1.1012, + "mean_token_accuracy": 0.6706624627113342, + "num_tokens": 48237037.0, + "step": 1881 + }, + { + "epoch": 0.2066769163189106, + "grad_norm": 2.29114031791687, + "learning_rate": 1e-06, + "loss": 1.0457, + "mean_token_accuracy": 0.6894196271896362, + "num_tokens": 48263051.0, + "step": 1882 + }, + { + "epoch": 0.20678673402152428, + "grad_norm": 2.2598373889923096, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6983518600463867, + "num_tokens": 48290619.0, + "step": 1883 + }, + { + "epoch": 0.20689655172413793, + "grad_norm": 2.3902323246002197, + "learning_rate": 1e-06, + "loss": 1.0774, + "mean_token_accuracy": 0.6802657842636108, + "num_tokens": 48313170.0, + "step": 1884 + }, + { + "epoch": 0.2070063694267516, + "grad_norm": 2.1894571781158447, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7228959202766418, + "num_tokens": 48338839.0, + "step": 1885 + }, + { + "epoch": 0.20711618712936525, + "grad_norm": 2.1666715145111084, + "learning_rate": 1e-06, + "loss": 1.0285, + "mean_token_accuracy": 0.6877317428588867, + "num_tokens": 48368370.0, + "step": 1886 + }, + { + "epoch": 0.20722600483197892, + "grad_norm": 2.3753814697265625, + "learning_rate": 1e-06, + "loss": 1.0541, + "mean_token_accuracy": 0.6850467920303345, + "num_tokens": 48390243.0, + "step": 1887 + }, + { + "epoch": 0.20733582253459257, + "grad_norm": 2.282839298248291, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6867498159408569, + "num_tokens": 48416432.0, + "step": 1888 + }, + { + "epoch": 0.20744564023720624, + "grad_norm": 2.6675565242767334, + "learning_rate": 1e-06, + "loss": 1.0313, + "mean_token_accuracy": 0.6962178349494934, + "num_tokens": 48438414.0, + "step": 1889 + }, + { + "epoch": 0.2075554579398199, + "grad_norm": 2.181581497192383, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.6903871893882751, + "num_tokens": 48466299.0, + "step": 1890 + }, + { + "epoch": 0.20766527564243356, + "grad_norm": 2.096370220184326, + "learning_rate": 1e-06, + "loss": 1.1361, + "mean_token_accuracy": 0.6659756898880005, + "num_tokens": 48499037.0, + "step": 1891 + }, + { + "epoch": 0.20777509334504723, + "grad_norm": 2.4085628986358643, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7093230485916138, + "num_tokens": 48521776.0, + "step": 1892 + }, + { + "epoch": 0.20788491104766088, + "grad_norm": 2.2932026386260986, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6924217939376831, + "num_tokens": 48548147.0, + "step": 1893 + }, + { + "epoch": 0.20799472875027455, + "grad_norm": 2.149141311645508, + "learning_rate": 1e-06, + "loss": 1.0582, + "mean_token_accuracy": 0.6767287254333496, + "num_tokens": 48575310.0, + "step": 1894 + }, + { + "epoch": 0.2081045464528882, + "grad_norm": 2.2868118286132812, + "learning_rate": 1e-06, + "loss": 1.0811, + "mean_token_accuracy": 0.6772395968437195, + "num_tokens": 48600434.0, + "step": 1895 + }, + { + "epoch": 0.20821436415550187, + "grad_norm": 2.278327703475952, + "learning_rate": 1e-06, + "loss": 1.0335, + "mean_token_accuracy": 0.6839046478271484, + "num_tokens": 48625449.0, + "step": 1896 + }, + { + "epoch": 0.20832418185811552, + "grad_norm": 2.7941055297851562, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7208555936813354, + "num_tokens": 48642657.0, + "step": 1897 + }, + { + "epoch": 0.2084339995607292, + "grad_norm": 2.7385034561157227, + "learning_rate": 1e-06, + "loss": 1.0464, + "mean_token_accuracy": 0.6873526573181152, + "num_tokens": 48662286.0, + "step": 1898 + }, + { + "epoch": 0.20854381726334284, + "grad_norm": 2.438131332397461, + "learning_rate": 1e-06, + "loss": 1.0784, + "mean_token_accuracy": 0.681771993637085, + "num_tokens": 48686286.0, + "step": 1899 + }, + { + "epoch": 0.2086536349659565, + "grad_norm": 2.498422861099243, + "learning_rate": 1e-06, + "loss": 1.1009, + "mean_token_accuracy": 0.6844297051429749, + "num_tokens": 48708667.0, + "step": 1900 + }, + { + "epoch": 0.20876345266857016, + "grad_norm": 1.973521113395691, + "learning_rate": 1e-06, + "loss": 1.0542, + "mean_token_accuracy": 0.6849799752235413, + "num_tokens": 48739425.0, + "step": 1901 + }, + { + "epoch": 0.20887327037118383, + "grad_norm": 2.045576572418213, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7100231051445007, + "num_tokens": 48769542.0, + "step": 1902 + }, + { + "epoch": 0.2089830880737975, + "grad_norm": 2.5055556297302246, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.6930983662605286, + "num_tokens": 48789812.0, + "step": 1903 + }, + { + "epoch": 0.20909290577641115, + "grad_norm": 2.0937392711639404, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.695387601852417, + "num_tokens": 48816682.0, + "step": 1904 + }, + { + "epoch": 0.20920272347902483, + "grad_norm": 2.2035787105560303, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7148802876472473, + "num_tokens": 48842022.0, + "step": 1905 + }, + { + "epoch": 0.20931254118163847, + "grad_norm": 2.3395543098449707, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.6994419097900391, + "num_tokens": 48864233.0, + "step": 1906 + }, + { + "epoch": 0.20942235888425215, + "grad_norm": 2.419421911239624, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6948496699333191, + "num_tokens": 48886311.0, + "step": 1907 + }, + { + "epoch": 0.2095321765868658, + "grad_norm": 2.4812145233154297, + "learning_rate": 1e-06, + "loss": 1.0876, + "mean_token_accuracy": 0.6826854348182678, + "num_tokens": 48908060.0, + "step": 1908 + }, + { + "epoch": 0.20964199428947947, + "grad_norm": 2.256007432937622, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.690258264541626, + "num_tokens": 48932582.0, + "step": 1909 + }, + { + "epoch": 0.2097518119920931, + "grad_norm": 2.408792018890381, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.6895116567611694, + "num_tokens": 48957647.0, + "step": 1910 + }, + { + "epoch": 0.20986162969470679, + "grad_norm": 2.0228936672210693, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.718873918056488, + "num_tokens": 48986163.0, + "step": 1911 + }, + { + "epoch": 0.20997144739732046, + "grad_norm": 2.5358259677886963, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7287595272064209, + "num_tokens": 49004661.0, + "step": 1912 + }, + { + "epoch": 0.2100812650999341, + "grad_norm": 2.510622978210449, + "learning_rate": 1e-06, + "loss": 1.0611, + "mean_token_accuracy": 0.6842638254165649, + "num_tokens": 49027192.0, + "step": 1913 + }, + { + "epoch": 0.21019108280254778, + "grad_norm": 2.1546578407287598, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6985030174255371, + "num_tokens": 49056079.0, + "step": 1914 + }, + { + "epoch": 0.21030090050516143, + "grad_norm": 2.367797613143921, + "learning_rate": 1e-06, + "loss": 1.038, + "mean_token_accuracy": 0.6901825666427612, + "num_tokens": 49080358.0, + "step": 1915 + }, + { + "epoch": 0.2104107182077751, + "grad_norm": 2.56548810005188, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7272432446479797, + "num_tokens": 49100179.0, + "step": 1916 + }, + { + "epoch": 0.21052053591038875, + "grad_norm": 2.2843148708343506, + "learning_rate": 1e-06, + "loss": 1.0787, + "mean_token_accuracy": 0.6765473484992981, + "num_tokens": 49124757.0, + "step": 1917 + }, + { + "epoch": 0.21063035361300242, + "grad_norm": 2.508415460586548, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6979464292526245, + "num_tokens": 49146296.0, + "step": 1918 + }, + { + "epoch": 0.21074017131561606, + "grad_norm": 2.4479522705078125, + "learning_rate": 1e-06, + "loss": 1.043, + "mean_token_accuracy": 0.6886276006698608, + "num_tokens": 49168742.0, + "step": 1919 + }, + { + "epoch": 0.21084998901822974, + "grad_norm": 2.1045470237731934, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.689822793006897, + "num_tokens": 49199907.0, + "step": 1920 + }, + { + "epoch": 0.2109598067208434, + "grad_norm": 2.0381977558135986, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.690919816493988, + "num_tokens": 49227537.0, + "step": 1921 + }, + { + "epoch": 0.21106962442345706, + "grad_norm": 2.3334732055664062, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6931883096694946, + "num_tokens": 49254048.0, + "step": 1922 + }, + { + "epoch": 0.21117944212607073, + "grad_norm": 2.0944504737854004, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7158457040786743, + "num_tokens": 49281351.0, + "step": 1923 + }, + { + "epoch": 0.21128925982868438, + "grad_norm": 2.303967237472534, + "learning_rate": 1e-06, + "loss": 1.029, + "mean_token_accuracy": 0.6918970346450806, + "num_tokens": 49304979.0, + "step": 1924 + }, + { + "epoch": 0.21139907753129805, + "grad_norm": 2.307746171951294, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.7074521780014038, + "num_tokens": 49331189.0, + "step": 1925 + }, + { + "epoch": 0.2115088952339117, + "grad_norm": 2.444124221801758, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7119706869125366, + "num_tokens": 49354274.0, + "step": 1926 + }, + { + "epoch": 0.21161871293652537, + "grad_norm": 2.036951780319214, + "learning_rate": 1e-06, + "loss": 1.0644, + "mean_token_accuracy": 0.6806427240371704, + "num_tokens": 49385245.0, + "step": 1927 + }, + { + "epoch": 0.21172853063913902, + "grad_norm": 2.285122871398926, + "learning_rate": 1e-06, + "loss": 1.0375, + "mean_token_accuracy": 0.6880602240562439, + "num_tokens": 49410893.0, + "step": 1928 + }, + { + "epoch": 0.2118383483417527, + "grad_norm": 2.399278163909912, + "learning_rate": 1e-06, + "loss": 1.0684, + "mean_token_accuracy": 0.6804186701774597, + "num_tokens": 49434983.0, + "step": 1929 + }, + { + "epoch": 0.21194816604436637, + "grad_norm": 2.2761828899383545, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7076942920684814, + "num_tokens": 49459549.0, + "step": 1930 + }, + { + "epoch": 0.21205798374698, + "grad_norm": 2.412506341934204, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7336968183517456, + "num_tokens": 49480280.0, + "step": 1931 + }, + { + "epoch": 0.21216780144959368, + "grad_norm": 2.367778778076172, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.6965013742446899, + "num_tokens": 49505611.0, + "step": 1932 + }, + { + "epoch": 0.21227761915220733, + "grad_norm": 2.347043991088867, + "learning_rate": 1e-06, + "loss": 1.0967, + "mean_token_accuracy": 0.6765439510345459, + "num_tokens": 49530105.0, + "step": 1933 + }, + { + "epoch": 0.212387436854821, + "grad_norm": 2.0562925338745117, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6909890174865723, + "num_tokens": 49561784.0, + "step": 1934 + }, + { + "epoch": 0.21249725455743465, + "grad_norm": 2.493947744369507, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7316104173660278, + "num_tokens": 49581031.0, + "step": 1935 + }, + { + "epoch": 0.21260707226004832, + "grad_norm": 2.4617559909820557, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.6984478235244751, + "num_tokens": 49602678.0, + "step": 1936 + }, + { + "epoch": 0.21271688996266197, + "grad_norm": 2.039895534515381, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.696416437625885, + "num_tokens": 49634141.0, + "step": 1937 + }, + { + "epoch": 0.21282670766527564, + "grad_norm": 2.1312055587768555, + "learning_rate": 1e-06, + "loss": 1.109, + "mean_token_accuracy": 0.6730678081512451, + "num_tokens": 49663858.0, + "step": 1938 + }, + { + "epoch": 0.2129365253678893, + "grad_norm": 1.9395467042922974, + "learning_rate": 1e-06, + "loss": 1.0498, + "mean_token_accuracy": 0.682132363319397, + "num_tokens": 49700281.0, + "step": 1939 + }, + { + "epoch": 0.21304634307050296, + "grad_norm": 2.4979794025421143, + "learning_rate": 1e-06, + "loss": 1.0533, + "mean_token_accuracy": 0.6855789422988892, + "num_tokens": 49721526.0, + "step": 1940 + }, + { + "epoch": 0.21315616077311664, + "grad_norm": 2.1186630725860596, + "learning_rate": 1e-06, + "loss": 1.0428, + "mean_token_accuracy": 0.6934350728988647, + "num_tokens": 49751348.0, + "step": 1941 + }, + { + "epoch": 0.21326597847573028, + "grad_norm": 2.0267863273620605, + "learning_rate": 1e-06, + "loss": 1.0395, + "mean_token_accuracy": 0.6863871812820435, + "num_tokens": 49784331.0, + "step": 1942 + }, + { + "epoch": 0.21337579617834396, + "grad_norm": 2.338797092437744, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6976770162582397, + "num_tokens": 49808605.0, + "step": 1943 + }, + { + "epoch": 0.2134856138809576, + "grad_norm": 2.0984208583831787, + "learning_rate": 1e-06, + "loss": 1.183, + "mean_token_accuracy": 0.655809760093689, + "num_tokens": 49838794.0, + "step": 1944 + }, + { + "epoch": 0.21359543158357128, + "grad_norm": 2.178368330001831, + "learning_rate": 1e-06, + "loss": 1.0784, + "mean_token_accuracy": 0.6808155179023743, + "num_tokens": 49869274.0, + "step": 1945 + }, + { + "epoch": 0.21370524928618492, + "grad_norm": 2.5313823223114014, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.7085675597190857, + "num_tokens": 49891335.0, + "step": 1946 + }, + { + "epoch": 0.2138150669887986, + "grad_norm": 2.0980656147003174, + "learning_rate": 1e-06, + "loss": 1.1273, + "mean_token_accuracy": 0.6628053188323975, + "num_tokens": 49924017.0, + "step": 1947 + }, + { + "epoch": 0.21392488469141224, + "grad_norm": 2.0679774284362793, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7141909599304199, + "num_tokens": 49953198.0, + "step": 1948 + }, + { + "epoch": 0.21403470239402592, + "grad_norm": 2.3584938049316406, + "learning_rate": 1e-06, + "loss": 1.1192, + "mean_token_accuracy": 0.6662644147872925, + "num_tokens": 49976231.0, + "step": 1949 + }, + { + "epoch": 0.2141445200966396, + "grad_norm": 2.087722063064575, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.7041572332382202, + "num_tokens": 50003244.0, + "step": 1950 + }, + { + "epoch": 0.21425433779925324, + "grad_norm": 2.2064642906188965, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.6897525191307068, + "num_tokens": 50029778.0, + "step": 1951 + }, + { + "epoch": 0.2143641555018669, + "grad_norm": 2.3799588680267334, + "learning_rate": 1e-06, + "loss": 1.0809, + "mean_token_accuracy": 0.6762977838516235, + "num_tokens": 50054023.0, + "step": 1952 + }, + { + "epoch": 0.21447397320448056, + "grad_norm": 2.131908416748047, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.7129582166671753, + "num_tokens": 50080771.0, + "step": 1953 + }, + { + "epoch": 0.21458379090709423, + "grad_norm": 2.6142663955688477, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7013675570487976, + "num_tokens": 50099440.0, + "step": 1954 + }, + { + "epoch": 0.21469360860970788, + "grad_norm": 2.6950197219848633, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.7140393853187561, + "num_tokens": 50118978.0, + "step": 1955 + }, + { + "epoch": 0.21480342631232155, + "grad_norm": 2.29260516166687, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7212772369384766, + "num_tokens": 50144313.0, + "step": 1956 + }, + { + "epoch": 0.2149132440149352, + "grad_norm": 2.1741623878479004, + "learning_rate": 1e-06, + "loss": 1.075, + "mean_token_accuracy": 0.6778122186660767, + "num_tokens": 50173062.0, + "step": 1957 + }, + { + "epoch": 0.21502306171754887, + "grad_norm": 2.3823909759521484, + "learning_rate": 1e-06, + "loss": 1.1205, + "mean_token_accuracy": 0.6645923852920532, + "num_tokens": 50197175.0, + "step": 1958 + }, + { + "epoch": 0.21513287942016254, + "grad_norm": 2.4494688510894775, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.7026703357696533, + "num_tokens": 50219849.0, + "step": 1959 + }, + { + "epoch": 0.2152426971227762, + "grad_norm": 2.089789867401123, + "learning_rate": 1e-06, + "loss": 1.0256, + "mean_token_accuracy": 0.6870644092559814, + "num_tokens": 50248519.0, + "step": 1960 + }, + { + "epoch": 0.21535251482538986, + "grad_norm": 2.150209665298462, + "learning_rate": 1e-06, + "loss": 1.0713, + "mean_token_accuracy": 0.6746655702590942, + "num_tokens": 50277701.0, + "step": 1961 + }, + { + "epoch": 0.2154623325280035, + "grad_norm": 2.1299729347229004, + "learning_rate": 1e-06, + "loss": 1.0757, + "mean_token_accuracy": 0.6795121431350708, + "num_tokens": 50306563.0, + "step": 1962 + }, + { + "epoch": 0.21557215023061718, + "grad_norm": 2.547358751296997, + "learning_rate": 1e-06, + "loss": 1.0573, + "mean_token_accuracy": 0.6834840774536133, + "num_tokens": 50329625.0, + "step": 1963 + }, + { + "epoch": 0.21568196793323083, + "grad_norm": 2.027834177017212, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.70867919921875, + "num_tokens": 50357273.0, + "step": 1964 + }, + { + "epoch": 0.2157917856358445, + "grad_norm": 2.0638601779937744, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6939813494682312, + "num_tokens": 50385587.0, + "step": 1965 + }, + { + "epoch": 0.21590160333845815, + "grad_norm": 2.0930135250091553, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.69451904296875, + "num_tokens": 50413245.0, + "step": 1966 + }, + { + "epoch": 0.21601142104107182, + "grad_norm": 2.3063626289367676, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7028416395187378, + "num_tokens": 50436511.0, + "step": 1967 + }, + { + "epoch": 0.2161212387436855, + "grad_norm": 2.4811525344848633, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7041698098182678, + "num_tokens": 50457754.0, + "step": 1968 + }, + { + "epoch": 0.21623105644629914, + "grad_norm": 2.2358486652374268, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6973285675048828, + "num_tokens": 50484205.0, + "step": 1969 + }, + { + "epoch": 0.21634087414891282, + "grad_norm": 2.372633457183838, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.6853641867637634, + "num_tokens": 50507862.0, + "step": 1970 + }, + { + "epoch": 0.21645069185152646, + "grad_norm": 2.4135138988494873, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6954790949821472, + "num_tokens": 50530416.0, + "step": 1971 + }, + { + "epoch": 0.21656050955414013, + "grad_norm": 2.2403500080108643, + "learning_rate": 1e-06, + "loss": 1.0345, + "mean_token_accuracy": 0.6895432472229004, + "num_tokens": 50556135.0, + "step": 1972 + }, + { + "epoch": 0.21667032725675378, + "grad_norm": 2.529081344604492, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.6995288133621216, + "num_tokens": 50577138.0, + "step": 1973 + }, + { + "epoch": 0.21678014495936745, + "grad_norm": 2.192265748977661, + "learning_rate": 1e-06, + "loss": 1.069, + "mean_token_accuracy": 0.6746136546134949, + "num_tokens": 50603872.0, + "step": 1974 + }, + { + "epoch": 0.2168899626619811, + "grad_norm": 2.2169687747955322, + "learning_rate": 1e-06, + "loss": 1.0527, + "mean_token_accuracy": 0.6878988146781921, + "num_tokens": 50629269.0, + "step": 1975 + }, + { + "epoch": 0.21699978036459477, + "grad_norm": 2.1035377979278564, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7143137454986572, + "num_tokens": 50654795.0, + "step": 1976 + }, + { + "epoch": 0.21710959806720842, + "grad_norm": 1.976952314376831, + "learning_rate": 1e-06, + "loss": 1.0729, + "mean_token_accuracy": 0.6831836700439453, + "num_tokens": 50689243.0, + "step": 1977 + }, + { + "epoch": 0.2172194157698221, + "grad_norm": 2.083059787750244, + "learning_rate": 1e-06, + "loss": 1.036, + "mean_token_accuracy": 0.6874609589576721, + "num_tokens": 50717236.0, + "step": 1978 + }, + { + "epoch": 0.21732923347243577, + "grad_norm": 2.155139923095703, + "learning_rate": 1e-06, + "loss": 1.0705, + "mean_token_accuracy": 0.678331732749939, + "num_tokens": 50747113.0, + "step": 1979 + }, + { + "epoch": 0.2174390511750494, + "grad_norm": 2.24519419670105, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.6996119022369385, + "num_tokens": 50771479.0, + "step": 1980 + }, + { + "epoch": 0.2175488688776631, + "grad_norm": 2.238784074783325, + "learning_rate": 1e-06, + "loss": 1.0555, + "mean_token_accuracy": 0.6876288652420044, + "num_tokens": 50798915.0, + "step": 1981 + }, + { + "epoch": 0.21765868658027673, + "grad_norm": 2.3035125732421875, + "learning_rate": 1e-06, + "loss": 1.0817, + "mean_token_accuracy": 0.6743977069854736, + "num_tokens": 50824844.0, + "step": 1982 + }, + { + "epoch": 0.2177685042828904, + "grad_norm": 2.099879503250122, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7118463516235352, + "num_tokens": 50856417.0, + "step": 1983 + }, + { + "epoch": 0.21787832198550405, + "grad_norm": 2.478435754776001, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.6983525156974792, + "num_tokens": 50879217.0, + "step": 1984 + }, + { + "epoch": 0.21798813968811773, + "grad_norm": 1.93648099899292, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.6914210915565491, + "num_tokens": 50912489.0, + "step": 1985 + }, + { + "epoch": 0.21809795739073137, + "grad_norm": 2.7031359672546387, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7161263823509216, + "num_tokens": 50929988.0, + "step": 1986 + }, + { + "epoch": 0.21820777509334505, + "grad_norm": 2.3200247287750244, + "learning_rate": 1e-06, + "loss": 1.0535, + "mean_token_accuracy": 0.6805316805839539, + "num_tokens": 50954802.0, + "step": 1987 + }, + { + "epoch": 0.21831759279595872, + "grad_norm": 2.5067410469055176, + "learning_rate": 1e-06, + "loss": 1.0295, + "mean_token_accuracy": 0.6930273771286011, + "num_tokens": 50978173.0, + "step": 1988 + }, + { + "epoch": 0.21842741049857237, + "grad_norm": 2.348111391067505, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.6951878070831299, + "num_tokens": 51005018.0, + "step": 1989 + }, + { + "epoch": 0.21853722820118604, + "grad_norm": 2.7345566749572754, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7048032283782959, + "num_tokens": 51024230.0, + "step": 1990 + }, + { + "epoch": 0.21864704590379969, + "grad_norm": 2.0271732807159424, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6931146383285522, + "num_tokens": 51055554.0, + "step": 1991 + }, + { + "epoch": 0.21875686360641336, + "grad_norm": 2.483203649520874, + "learning_rate": 1e-06, + "loss": 1.0639, + "mean_token_accuracy": 0.6840665936470032, + "num_tokens": 51079065.0, + "step": 1992 + }, + { + "epoch": 0.218866681309027, + "grad_norm": 2.8137388229370117, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7150906920433044, + "num_tokens": 51096633.0, + "step": 1993 + }, + { + "epoch": 0.21897649901164068, + "grad_norm": 2.2118496894836426, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7070980072021484, + "num_tokens": 51121775.0, + "step": 1994 + }, + { + "epoch": 0.21908631671425433, + "grad_norm": 2.0126612186431885, + "learning_rate": 1e-06, + "loss": 1.0564, + "mean_token_accuracy": 0.6995452046394348, + "num_tokens": 51153060.0, + "step": 1995 + }, + { + "epoch": 0.219196134416868, + "grad_norm": 2.243307590484619, + "learning_rate": 1e-06, + "loss": 1.0519, + "mean_token_accuracy": 0.6820775270462036, + "num_tokens": 51179873.0, + "step": 1996 + }, + { + "epoch": 0.21930595211948167, + "grad_norm": 2.2661523818969727, + "learning_rate": 1e-06, + "loss": 1.0425, + "mean_token_accuracy": 0.6884940266609192, + "num_tokens": 51206092.0, + "step": 1997 + }, + { + "epoch": 0.21941576982209532, + "grad_norm": 2.259751319885254, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7170263528823853, + "num_tokens": 51231918.0, + "step": 1998 + }, + { + "epoch": 0.219525587524709, + "grad_norm": 2.4939751625061035, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7032270431518555, + "num_tokens": 51253839.0, + "step": 1999 + }, + { + "epoch": 0.21963540522732264, + "grad_norm": 1.9220635890960693, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7275457382202148, + "num_tokens": 51284731.0, + "step": 2000 + }, + { + "epoch": 0.2197452229299363, + "grad_norm": 2.4960076808929443, + "learning_rate": 1e-06, + "loss": 1.0356, + "mean_token_accuracy": 0.6877401471138, + "num_tokens": 51307152.0, + "step": 2001 + }, + { + "epoch": 0.21985504063254996, + "grad_norm": 2.420400381088257, + "learning_rate": 1e-06, + "loss": 1.0884, + "mean_token_accuracy": 0.6838093996047974, + "num_tokens": 51332386.0, + "step": 2002 + }, + { + "epoch": 0.21996485833516363, + "grad_norm": 2.188988447189331, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7121546268463135, + "num_tokens": 51358766.0, + "step": 2003 + }, + { + "epoch": 0.22007467603777728, + "grad_norm": 2.3291501998901367, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7075917720794678, + "num_tokens": 51381932.0, + "step": 2004 + }, + { + "epoch": 0.22018449374039095, + "grad_norm": 2.0861101150512695, + "learning_rate": 1e-06, + "loss": 1.1317, + "mean_token_accuracy": 0.6708868145942688, + "num_tokens": 51414865.0, + "step": 2005 + }, + { + "epoch": 0.22029431144300463, + "grad_norm": 2.633754014968872, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7158298492431641, + "num_tokens": 51435899.0, + "step": 2006 + }, + { + "epoch": 0.22040412914561827, + "grad_norm": 2.634037971496582, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.704791247844696, + "num_tokens": 51455899.0, + "step": 2007 + }, + { + "epoch": 0.22051394684823195, + "grad_norm": 2.3267123699188232, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.7025011777877808, + "num_tokens": 51478816.0, + "step": 2008 + }, + { + "epoch": 0.2206237645508456, + "grad_norm": 2.3575849533081055, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7147809267044067, + "num_tokens": 51502537.0, + "step": 2009 + }, + { + "epoch": 0.22073358225345927, + "grad_norm": 2.4974663257598877, + "learning_rate": 1e-06, + "loss": 1.0695, + "mean_token_accuracy": 0.6861217021942139, + "num_tokens": 51524236.0, + "step": 2010 + }, + { + "epoch": 0.2208433999560729, + "grad_norm": 2.3808822631835938, + "learning_rate": 1e-06, + "loss": 1.0749, + "mean_token_accuracy": 0.6747477650642395, + "num_tokens": 51549227.0, + "step": 2011 + }, + { + "epoch": 0.22095321765868658, + "grad_norm": 2.3090598583221436, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7100456953048706, + "num_tokens": 51576011.0, + "step": 2012 + }, + { + "epoch": 0.22106303536130023, + "grad_norm": 2.4286420345306396, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7022553086280823, + "num_tokens": 51599345.0, + "step": 2013 + }, + { + "epoch": 0.2211728530639139, + "grad_norm": 1.9818366765975952, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.7032405138015747, + "num_tokens": 51629187.0, + "step": 2014 + }, + { + "epoch": 0.22128267076652755, + "grad_norm": 1.983656644821167, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6935248970985413, + "num_tokens": 51663092.0, + "step": 2015 + }, + { + "epoch": 0.22139248846914122, + "grad_norm": 2.194605827331543, + "learning_rate": 1e-06, + "loss": 1.0594, + "mean_token_accuracy": 0.6895576119422913, + "num_tokens": 51692279.0, + "step": 2016 + }, + { + "epoch": 0.2215023061717549, + "grad_norm": 2.2877390384674072, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7152970433235168, + "num_tokens": 51717600.0, + "step": 2017 + }, + { + "epoch": 0.22161212387436854, + "grad_norm": 2.509392023086548, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6898127794265747, + "num_tokens": 51740471.0, + "step": 2018 + }, + { + "epoch": 0.22172194157698222, + "grad_norm": 2.2533206939697266, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.7086129188537598, + "num_tokens": 51766497.0, + "step": 2019 + }, + { + "epoch": 0.22183175927959586, + "grad_norm": 2.1831815242767334, + "learning_rate": 1e-06, + "loss": 1.0118, + "mean_token_accuracy": 0.6966153383255005, + "num_tokens": 51790802.0, + "step": 2020 + }, + { + "epoch": 0.22194157698220954, + "grad_norm": 2.3699419498443604, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.6902765035629272, + "num_tokens": 51814984.0, + "step": 2021 + }, + { + "epoch": 0.22205139468482318, + "grad_norm": 2.514458417892456, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.701566219329834, + "num_tokens": 51837647.0, + "step": 2022 + }, + { + "epoch": 0.22216121238743686, + "grad_norm": 2.3253016471862793, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7171939015388489, + "num_tokens": 51862653.0, + "step": 2023 + }, + { + "epoch": 0.2222710300900505, + "grad_norm": 2.5300517082214355, + "learning_rate": 1e-06, + "loss": 1.0726, + "mean_token_accuracy": 0.68472820520401, + "num_tokens": 51884710.0, + "step": 2024 + }, + { + "epoch": 0.22238084779266418, + "grad_norm": 2.4499735832214355, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.6918603777885437, + "num_tokens": 51909462.0, + "step": 2025 + }, + { + "epoch": 0.22249066549527785, + "grad_norm": 2.2449262142181396, + "learning_rate": 1e-06, + "loss": 1.0908, + "mean_token_accuracy": 0.6839259266853333, + "num_tokens": 51935845.0, + "step": 2026 + }, + { + "epoch": 0.2226004831978915, + "grad_norm": 2.4194846153259277, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.7024824023246765, + "num_tokens": 51958872.0, + "step": 2027 + }, + { + "epoch": 0.22271030090050517, + "grad_norm": 2.379295825958252, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.6896654367446899, + "num_tokens": 51983814.0, + "step": 2028 + }, + { + "epoch": 0.22282011860311882, + "grad_norm": 2.3833746910095215, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7082991003990173, + "num_tokens": 52005984.0, + "step": 2029 + }, + { + "epoch": 0.2229299363057325, + "grad_norm": 2.5558764934539795, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7214143872261047, + "num_tokens": 52026130.0, + "step": 2030 + }, + { + "epoch": 0.22303975400834614, + "grad_norm": 2.1771626472473145, + "learning_rate": 1e-06, + "loss": 1.182, + "mean_token_accuracy": 0.6589164733886719, + "num_tokens": 52056679.0, + "step": 2031 + }, + { + "epoch": 0.2231495717109598, + "grad_norm": 2.2122271060943604, + "learning_rate": 1e-06, + "loss": 1.0893, + "mean_token_accuracy": 0.6713712215423584, + "num_tokens": 52085952.0, + "step": 2032 + }, + { + "epoch": 0.22325938941357346, + "grad_norm": 2.256757974624634, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6925718188285828, + "num_tokens": 52112006.0, + "step": 2033 + }, + { + "epoch": 0.22336920711618713, + "grad_norm": 2.183907985687256, + "learning_rate": 1e-06, + "loss": 1.0327, + "mean_token_accuracy": 0.6976318359375, + "num_tokens": 52140488.0, + "step": 2034 + }, + { + "epoch": 0.2234790248188008, + "grad_norm": 2.62353253364563, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7071214914321899, + "num_tokens": 52171693.0, + "step": 2035 + }, + { + "epoch": 0.22358884252141445, + "grad_norm": 2.078037977218628, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.7024267315864563, + "num_tokens": 52200102.0, + "step": 2036 + }, + { + "epoch": 0.22369866022402812, + "grad_norm": 2.1509416103363037, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7053866386413574, + "num_tokens": 52227371.0, + "step": 2037 + }, + { + "epoch": 0.22380847792664177, + "grad_norm": 2.134850263595581, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6951887011528015, + "num_tokens": 52255549.0, + "step": 2038 + }, + { + "epoch": 0.22391829562925544, + "grad_norm": 2.062913417816162, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7154066562652588, + "num_tokens": 52283627.0, + "step": 2039 + }, + { + "epoch": 0.2240281133318691, + "grad_norm": 2.277320384979248, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7206809520721436, + "num_tokens": 52308197.0, + "step": 2040 + }, + { + "epoch": 0.22413793103448276, + "grad_norm": 2.3037772178649902, + "learning_rate": 1e-06, + "loss": 1.0725, + "mean_token_accuracy": 0.6776714324951172, + "num_tokens": 52331914.0, + "step": 2041 + }, + { + "epoch": 0.2242477487370964, + "grad_norm": 2.103203535079956, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.699234127998352, + "num_tokens": 52359977.0, + "step": 2042 + }, + { + "epoch": 0.22435756643971008, + "grad_norm": 2.498711347579956, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7299830913543701, + "num_tokens": 52382138.0, + "step": 2043 + }, + { + "epoch": 0.22446738414232376, + "grad_norm": 2.4822750091552734, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.7102667689323425, + "num_tokens": 52404958.0, + "step": 2044 + }, + { + "epoch": 0.2245772018449374, + "grad_norm": 2.1358044147491455, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7046038508415222, + "num_tokens": 52431905.0, + "step": 2045 + }, + { + "epoch": 0.22468701954755108, + "grad_norm": 2.5024971961975098, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6973339319229126, + "num_tokens": 52454059.0, + "step": 2046 + }, + { + "epoch": 0.22479683725016472, + "grad_norm": 2.2471206188201904, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.7009268999099731, + "num_tokens": 52479026.0, + "step": 2047 + }, + { + "epoch": 0.2249066549527784, + "grad_norm": 2.324838161468506, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7154345512390137, + "num_tokens": 52502695.0, + "step": 2048 + }, + { + "epoch": 0.22501647265539204, + "grad_norm": 2.7332606315612793, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7108231782913208, + "num_tokens": 52521135.0, + "step": 2049 + }, + { + "epoch": 0.22512629035800572, + "grad_norm": 2.087327003479004, + "learning_rate": 1e-06, + "loss": 1.0853, + "mean_token_accuracy": 0.6784700155258179, + "num_tokens": 52552736.0, + "step": 2050 + }, + { + "epoch": 0.22523610806061936, + "grad_norm": 2.3089942932128906, + "learning_rate": 1e-06, + "loss": 1.0325, + "mean_token_accuracy": 0.6919547319412231, + "num_tokens": 52575948.0, + "step": 2051 + }, + { + "epoch": 0.22534592576323303, + "grad_norm": 2.080859661102295, + "learning_rate": 1e-06, + "loss": 1.0639, + "mean_token_accuracy": 0.6851884722709656, + "num_tokens": 52604073.0, + "step": 2052 + }, + { + "epoch": 0.22545574346584668, + "grad_norm": 2.6088924407958984, + "learning_rate": 1e-06, + "loss": 1.0451, + "mean_token_accuracy": 0.6833223104476929, + "num_tokens": 52624742.0, + "step": 2053 + }, + { + "epoch": 0.22556556116846035, + "grad_norm": 2.5886800289154053, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.7044286131858826, + "num_tokens": 52645275.0, + "step": 2054 + }, + { + "epoch": 0.22567537887107403, + "grad_norm": 1.9939746856689453, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.724173903465271, + "num_tokens": 52674431.0, + "step": 2055 + }, + { + "epoch": 0.22578519657368767, + "grad_norm": 3.828526258468628, + "learning_rate": 1e-06, + "loss": 1.0249, + "mean_token_accuracy": 0.6808788180351257, + "num_tokens": 52700546.0, + "step": 2056 + }, + { + "epoch": 0.22589501427630135, + "grad_norm": 2.4194343090057373, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.699842095375061, + "num_tokens": 52721948.0, + "step": 2057 + }, + { + "epoch": 0.226004831978915, + "grad_norm": 2.571911096572876, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.6963680386543274, + "num_tokens": 52744010.0, + "step": 2058 + }, + { + "epoch": 0.22611464968152867, + "grad_norm": 2.0693016052246094, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6937236189842224, + "num_tokens": 52774979.0, + "step": 2059 + }, + { + "epoch": 0.2262244673841423, + "grad_norm": 2.7002670764923096, + "learning_rate": 1e-06, + "loss": 1.046, + "mean_token_accuracy": 0.6886487007141113, + "num_tokens": 52796137.0, + "step": 2060 + }, + { + "epoch": 0.226334285086756, + "grad_norm": 2.491013288497925, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.715528130531311, + "num_tokens": 52817501.0, + "step": 2061 + }, + { + "epoch": 0.22644410278936963, + "grad_norm": 2.148940324783325, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.6899363994598389, + "num_tokens": 52847606.0, + "step": 2062 + }, + { + "epoch": 0.2265539204919833, + "grad_norm": 2.349799394607544, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.6968624591827393, + "num_tokens": 52872790.0, + "step": 2063 + }, + { + "epoch": 0.22666373819459698, + "grad_norm": 2.122924327850342, + "learning_rate": 1e-06, + "loss": 1.0303, + "mean_token_accuracy": 0.6890870332717896, + "num_tokens": 52900975.0, + "step": 2064 + }, + { + "epoch": 0.22677355589721063, + "grad_norm": 2.3392956256866455, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.6886301636695862, + "num_tokens": 52926048.0, + "step": 2065 + }, + { + "epoch": 0.2268833735998243, + "grad_norm": 2.1817922592163086, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7109887599945068, + "num_tokens": 52953340.0, + "step": 2066 + }, + { + "epoch": 0.22699319130243795, + "grad_norm": 2.54369854927063, + "learning_rate": 1e-06, + "loss": 1.0616, + "mean_token_accuracy": 0.6834089159965515, + "num_tokens": 52976861.0, + "step": 2067 + }, + { + "epoch": 0.22710300900505162, + "grad_norm": 2.4816508293151855, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6964588165283203, + "num_tokens": 53000837.0, + "step": 2068 + }, + { + "epoch": 0.22721282670766527, + "grad_norm": 2.0834012031555176, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6885300874710083, + "num_tokens": 53029346.0, + "step": 2069 + }, + { + "epoch": 0.22732264441027894, + "grad_norm": 2.2732064723968506, + "learning_rate": 1e-06, + "loss": 1.0337, + "mean_token_accuracy": 0.6910388469696045, + "num_tokens": 53055270.0, + "step": 2070 + }, + { + "epoch": 0.22743246211289259, + "grad_norm": 2.4584033489227295, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.7024927735328674, + "num_tokens": 53074742.0, + "step": 2071 + }, + { + "epoch": 0.22754227981550626, + "grad_norm": 2.2885828018188477, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7019366025924683, + "num_tokens": 53099112.0, + "step": 2072 + }, + { + "epoch": 0.22765209751811993, + "grad_norm": 2.3213088512420654, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7198632955551147, + "num_tokens": 53121819.0, + "step": 2073 + }, + { + "epoch": 0.22776191522073358, + "grad_norm": 2.1784677505493164, + "learning_rate": 1e-06, + "loss": 1.0355, + "mean_token_accuracy": 0.6943522691726685, + "num_tokens": 53148628.0, + "step": 2074 + }, + { + "epoch": 0.22787173292334725, + "grad_norm": 2.099113702774048, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7162972688674927, + "num_tokens": 53179761.0, + "step": 2075 + }, + { + "epoch": 0.2279815506259609, + "grad_norm": 2.054795265197754, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6948217153549194, + "num_tokens": 53210173.0, + "step": 2076 + }, + { + "epoch": 0.22809136832857457, + "grad_norm": 2.2204179763793945, + "learning_rate": 1e-06, + "loss": 1.0578, + "mean_token_accuracy": 0.6829931735992432, + "num_tokens": 53238796.0, + "step": 2077 + }, + { + "epoch": 0.22820118603118822, + "grad_norm": 2.8812215328216553, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.722671389579773, + "num_tokens": 53256719.0, + "step": 2078 + }, + { + "epoch": 0.2283110037338019, + "grad_norm": 2.558037757873535, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.6864591836929321, + "num_tokens": 53277919.0, + "step": 2079 + }, + { + "epoch": 0.22842082143641554, + "grad_norm": 2.4370193481445312, + "learning_rate": 1e-06, + "loss": 1.0393, + "mean_token_accuracy": 0.6860083937644958, + "num_tokens": 53300460.0, + "step": 2080 + }, + { + "epoch": 0.2285306391390292, + "grad_norm": 2.338860273361206, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.698987603187561, + "num_tokens": 53323874.0, + "step": 2081 + }, + { + "epoch": 0.2286404568416429, + "grad_norm": 2.006589889526367, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6986891031265259, + "num_tokens": 53356122.0, + "step": 2082 + }, + { + "epoch": 0.22875027454425653, + "grad_norm": 2.3494973182678223, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.70122230052948, + "num_tokens": 53378541.0, + "step": 2083 + }, + { + "epoch": 0.2288600922468702, + "grad_norm": 2.3898918628692627, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.696648895740509, + "num_tokens": 53400342.0, + "step": 2084 + }, + { + "epoch": 0.22896990994948385, + "grad_norm": 2.4369826316833496, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7234952449798584, + "num_tokens": 53422895.0, + "step": 2085 + }, + { + "epoch": 0.22907972765209753, + "grad_norm": 2.3974153995513916, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7046796679496765, + "num_tokens": 53443656.0, + "step": 2086 + }, + { + "epoch": 0.22918954535471117, + "grad_norm": 2.384460210800171, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7143582105636597, + "num_tokens": 53465824.0, + "step": 2087 + }, + { + "epoch": 0.22929936305732485, + "grad_norm": 2.399027109146118, + "learning_rate": 1e-06, + "loss": 1.0856, + "mean_token_accuracy": 0.6851304173469543, + "num_tokens": 53489783.0, + "step": 2088 + }, + { + "epoch": 0.2294091807599385, + "grad_norm": 2.5300705432891846, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.7040103673934937, + "num_tokens": 53512610.0, + "step": 2089 + }, + { + "epoch": 0.22951899846255217, + "grad_norm": 1.9405778646469116, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.693986713886261, + "num_tokens": 53547254.0, + "step": 2090 + }, + { + "epoch": 0.2296288161651658, + "grad_norm": 2.430952310562134, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.7053592205047607, + "num_tokens": 53570866.0, + "step": 2091 + }, + { + "epoch": 0.22973863386777948, + "grad_norm": 2.1982874870300293, + "learning_rate": 1e-06, + "loss": 1.0488, + "mean_token_accuracy": 0.6787205338478088, + "num_tokens": 53599489.0, + "step": 2092 + }, + { + "epoch": 0.22984845157039316, + "grad_norm": 1.9884198904037476, + "learning_rate": 1e-06, + "loss": 1.0986, + "mean_token_accuracy": 0.6751536726951599, + "num_tokens": 53630625.0, + "step": 2093 + }, + { + "epoch": 0.2299582692730068, + "grad_norm": 2.384939670562744, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7260786294937134, + "num_tokens": 53651913.0, + "step": 2094 + }, + { + "epoch": 0.23006808697562048, + "grad_norm": 2.2200825214385986, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.704770565032959, + "num_tokens": 53678525.0, + "step": 2095 + }, + { + "epoch": 0.23017790467823412, + "grad_norm": 2.7549448013305664, + "learning_rate": 1e-06, + "loss": 1.0294, + "mean_token_accuracy": 0.6924659013748169, + "num_tokens": 53697705.0, + "step": 2096 + }, + { + "epoch": 0.2302877223808478, + "grad_norm": 1.998887062072754, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7355489730834961, + "num_tokens": 53725786.0, + "step": 2097 + }, + { + "epoch": 0.23039754008346144, + "grad_norm": 2.7397258281707764, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7143147587776184, + "num_tokens": 53743698.0, + "step": 2098 + }, + { + "epoch": 0.23050735778607512, + "grad_norm": 2.018298864364624, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7025159597396851, + "num_tokens": 53772392.0, + "step": 2099 + }, + { + "epoch": 0.23061717548868876, + "grad_norm": 2.1417624950408936, + "learning_rate": 1e-06, + "loss": 1.0582, + "mean_token_accuracy": 0.6939424276351929, + "num_tokens": 53802816.0, + "step": 2100 + }, + { + "epoch": 0.23072699319130244, + "grad_norm": 2.434840679168701, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6932074427604675, + "num_tokens": 53828249.0, + "step": 2101 + }, + { + "epoch": 0.2308368108939161, + "grad_norm": 2.1217386722564697, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.694878101348877, + "num_tokens": 53857398.0, + "step": 2102 + }, + { + "epoch": 0.23094662859652976, + "grad_norm": 2.696455717086792, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.693718433380127, + "num_tokens": 53878402.0, + "step": 2103 + }, + { + "epoch": 0.23105644629914343, + "grad_norm": 2.2037456035614014, + "learning_rate": 1e-06, + "loss": 1.0597, + "mean_token_accuracy": 0.6897339820861816, + "num_tokens": 53904728.0, + "step": 2104 + }, + { + "epoch": 0.23116626400175708, + "grad_norm": 2.4781267642974854, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6983295679092407, + "num_tokens": 53925581.0, + "step": 2105 + }, + { + "epoch": 0.23127608170437075, + "grad_norm": 2.095186710357666, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.693818211555481, + "num_tokens": 53952671.0, + "step": 2106 + }, + { + "epoch": 0.2313858994069844, + "grad_norm": 2.251439332962036, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.691470742225647, + "num_tokens": 53980328.0, + "step": 2107 + }, + { + "epoch": 0.23149571710959807, + "grad_norm": 2.1585710048675537, + "learning_rate": 1e-06, + "loss": 1.0672, + "mean_token_accuracy": 0.6780028343200684, + "num_tokens": 54011384.0, + "step": 2108 + }, + { + "epoch": 0.23160553481221172, + "grad_norm": 2.479823112487793, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.689949095249176, + "num_tokens": 54031786.0, + "step": 2109 + }, + { + "epoch": 0.2317153525148254, + "grad_norm": 2.2581422328948975, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7112663984298706, + "num_tokens": 54054857.0, + "step": 2110 + }, + { + "epoch": 0.23182517021743906, + "grad_norm": 2.5319406986236572, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.7026913166046143, + "num_tokens": 54075830.0, + "step": 2111 + }, + { + "epoch": 0.2319349879200527, + "grad_norm": 2.414064884185791, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7161222100257874, + "num_tokens": 54099867.0, + "step": 2112 + }, + { + "epoch": 0.23204480562266638, + "grad_norm": 2.4405786991119385, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7031022906303406, + "num_tokens": 54122569.0, + "step": 2113 + }, + { + "epoch": 0.23215462332528003, + "grad_norm": 2.5381035804748535, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6976908445358276, + "num_tokens": 54144756.0, + "step": 2114 + }, + { + "epoch": 0.2322644410278937, + "grad_norm": 2.328165054321289, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.7001198530197144, + "num_tokens": 54170068.0, + "step": 2115 + }, + { + "epoch": 0.23237425873050735, + "grad_norm": 2.2291195392608643, + "learning_rate": 1e-06, + "loss": 1.0722, + "mean_token_accuracy": 0.685084342956543, + "num_tokens": 54197437.0, + "step": 2116 + }, + { + "epoch": 0.23248407643312102, + "grad_norm": 2.086249828338623, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7084031105041504, + "num_tokens": 54225928.0, + "step": 2117 + }, + { + "epoch": 0.23259389413573467, + "grad_norm": 2.373065948486328, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6925263404846191, + "num_tokens": 54251084.0, + "step": 2118 + }, + { + "epoch": 0.23270371183834834, + "grad_norm": 2.313204050064087, + "learning_rate": 1e-06, + "loss": 1.0685, + "mean_token_accuracy": 0.6821903586387634, + "num_tokens": 54277677.0, + "step": 2119 + }, + { + "epoch": 0.23281352954096202, + "grad_norm": 2.491368532180786, + "learning_rate": 1e-06, + "loss": 1.0721, + "mean_token_accuracy": 0.6782997846603394, + "num_tokens": 54300246.0, + "step": 2120 + }, + { + "epoch": 0.23292334724357566, + "grad_norm": 2.2741055488586426, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7295200824737549, + "num_tokens": 54321850.0, + "step": 2121 + }, + { + "epoch": 0.23303316494618934, + "grad_norm": 2.338794231414795, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6992586851119995, + "num_tokens": 54346037.0, + "step": 2122 + }, + { + "epoch": 0.23314298264880298, + "grad_norm": 2.4812183380126953, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6962606310844421, + "num_tokens": 54370221.0, + "step": 2123 + }, + { + "epoch": 0.23325280035141666, + "grad_norm": 2.608128786087036, + "learning_rate": 1e-06, + "loss": 1.0421, + "mean_token_accuracy": 0.6927133202552795, + "num_tokens": 54389915.0, + "step": 2124 + }, + { + "epoch": 0.2333626180540303, + "grad_norm": 2.3753275871276855, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6955393552780151, + "num_tokens": 54414650.0, + "step": 2125 + }, + { + "epoch": 0.23347243575664398, + "grad_norm": 2.203885555267334, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7001984715461731, + "num_tokens": 54441869.0, + "step": 2126 + }, + { + "epoch": 0.23358225345925762, + "grad_norm": 2.440704822540283, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.6951479911804199, + "num_tokens": 54463515.0, + "step": 2127 + }, + { + "epoch": 0.2336920711618713, + "grad_norm": 2.12083101272583, + "learning_rate": 1e-06, + "loss": 1.0357, + "mean_token_accuracy": 0.6896647214889526, + "num_tokens": 54493597.0, + "step": 2128 + }, + { + "epoch": 0.23380188886448494, + "grad_norm": 2.0364184379577637, + "learning_rate": 1e-06, + "loss": 1.1337, + "mean_token_accuracy": 0.6617499589920044, + "num_tokens": 54527856.0, + "step": 2129 + }, + { + "epoch": 0.23391170656709862, + "grad_norm": 2.0297718048095703, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6999231576919556, + "num_tokens": 54560117.0, + "step": 2130 + }, + { + "epoch": 0.2340215242697123, + "grad_norm": 2.331838607788086, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.707893967628479, + "num_tokens": 54582791.0, + "step": 2131 + }, + { + "epoch": 0.23413134197232593, + "grad_norm": 2.2724735736846924, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6969059705734253, + "num_tokens": 54608012.0, + "step": 2132 + }, + { + "epoch": 0.2342411596749396, + "grad_norm": 2.2393460273742676, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.7039214372634888, + "num_tokens": 54633758.0, + "step": 2133 + }, + { + "epoch": 0.23435097737755325, + "grad_norm": 2.247420072555542, + "learning_rate": 1e-06, + "loss": 1.0305, + "mean_token_accuracy": 0.6878864765167236, + "num_tokens": 54661095.0, + "step": 2134 + }, + { + "epoch": 0.23446079508016693, + "grad_norm": 2.5232465267181396, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6996810436248779, + "num_tokens": 54683697.0, + "step": 2135 + }, + { + "epoch": 0.23457061278278057, + "grad_norm": 2.813570737838745, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.7061429023742676, + "num_tokens": 54700050.0, + "step": 2136 + }, + { + "epoch": 0.23468043048539425, + "grad_norm": 2.2228994369506836, + "learning_rate": 1e-06, + "loss": 1.0305, + "mean_token_accuracy": 0.6869062185287476, + "num_tokens": 54726288.0, + "step": 2137 + }, + { + "epoch": 0.2347902481880079, + "grad_norm": 2.459441900253296, + "learning_rate": 1e-06, + "loss": 1.0757, + "mean_token_accuracy": 0.6783806085586548, + "num_tokens": 54750242.0, + "step": 2138 + }, + { + "epoch": 0.23490006589062157, + "grad_norm": 2.296215295791626, + "learning_rate": 1e-06, + "loss": 1.088, + "mean_token_accuracy": 0.6742459535598755, + "num_tokens": 54777509.0, + "step": 2139 + }, + { + "epoch": 0.23500988359323524, + "grad_norm": 2.4028918743133545, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7222273945808411, + "num_tokens": 54798345.0, + "step": 2140 + }, + { + "epoch": 0.2351197012958489, + "grad_norm": 2.076101541519165, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.6998068690299988, + "num_tokens": 54828657.0, + "step": 2141 + }, + { + "epoch": 0.23522951899846256, + "grad_norm": 2.242539882659912, + "learning_rate": 1e-06, + "loss": 1.07, + "mean_token_accuracy": 0.6810588240623474, + "num_tokens": 54855358.0, + "step": 2142 + }, + { + "epoch": 0.2353393367010762, + "grad_norm": 1.93256676197052, + "learning_rate": 1e-06, + "loss": 1.0999, + "mean_token_accuracy": 0.6784868836402893, + "num_tokens": 54889797.0, + "step": 2143 + }, + { + "epoch": 0.23544915440368988, + "grad_norm": 2.0226874351501465, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7194185853004456, + "num_tokens": 54919530.0, + "step": 2144 + }, + { + "epoch": 0.23555897210630353, + "grad_norm": 2.237058401107788, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.690758466720581, + "num_tokens": 54945218.0, + "step": 2145 + }, + { + "epoch": 0.2356687898089172, + "grad_norm": 2.3377673625946045, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.7024145126342773, + "num_tokens": 54970384.0, + "step": 2146 + }, + { + "epoch": 0.23577860751153085, + "grad_norm": 2.267125368118286, + "learning_rate": 1e-06, + "loss": 1.1254, + "mean_token_accuracy": 0.6673957109451294, + "num_tokens": 54997245.0, + "step": 2147 + }, + { + "epoch": 0.23588842521414452, + "grad_norm": 1.947369933128357, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7096914052963257, + "num_tokens": 55029157.0, + "step": 2148 + }, + { + "epoch": 0.2359982429167582, + "grad_norm": 2.494002103805542, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.720979630947113, + "num_tokens": 55048993.0, + "step": 2149 + }, + { + "epoch": 0.23610806061937184, + "grad_norm": 2.291480541229248, + "learning_rate": 1e-06, + "loss": 1.0569, + "mean_token_accuracy": 0.6947808861732483, + "num_tokens": 55073496.0, + "step": 2150 + }, + { + "epoch": 0.2362178783219855, + "grad_norm": 2.299997568130493, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.7002792954444885, + "num_tokens": 55099269.0, + "step": 2151 + }, + { + "epoch": 0.23632769602459916, + "grad_norm": 2.2151260375976562, + "learning_rate": 1e-06, + "loss": 1.0976, + "mean_token_accuracy": 0.6815970540046692, + "num_tokens": 55129049.0, + "step": 2152 + }, + { + "epoch": 0.23643751372721283, + "grad_norm": 2.190847396850586, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7191056609153748, + "num_tokens": 55155374.0, + "step": 2153 + }, + { + "epoch": 0.23654733142982648, + "grad_norm": 2.3068807125091553, + "learning_rate": 1e-06, + "loss": 1.1015, + "mean_token_accuracy": 0.6745416522026062, + "num_tokens": 55181132.0, + "step": 2154 + }, + { + "epoch": 0.23665714913244015, + "grad_norm": 1.9875067472457886, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7215255498886108, + "num_tokens": 55211596.0, + "step": 2155 + }, + { + "epoch": 0.2367669668350538, + "grad_norm": 2.3780999183654785, + "learning_rate": 1e-06, + "loss": 1.0789, + "mean_token_accuracy": 0.6829890608787537, + "num_tokens": 55237038.0, + "step": 2156 + }, + { + "epoch": 0.23687678453766747, + "grad_norm": 2.5964789390563965, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.71321702003479, + "num_tokens": 55256762.0, + "step": 2157 + }, + { + "epoch": 0.23698660224028115, + "grad_norm": 2.2613298892974854, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6988551020622253, + "num_tokens": 55282040.0, + "step": 2158 + }, + { + "epoch": 0.2370964199428948, + "grad_norm": 2.311168670654297, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6993906497955322, + "num_tokens": 55305876.0, + "step": 2159 + }, + { + "epoch": 0.23720623764550847, + "grad_norm": 2.45100474357605, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7047078609466553, + "num_tokens": 55329672.0, + "step": 2160 + }, + { + "epoch": 0.2373160553481221, + "grad_norm": 2.456895351409912, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.708807647228241, + "num_tokens": 55351220.0, + "step": 2161 + }, + { + "epoch": 0.2374258730507358, + "grad_norm": 2.204005002975464, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.7063558101654053, + "num_tokens": 55378077.0, + "step": 2162 + }, + { + "epoch": 0.23753569075334943, + "grad_norm": 2.4900410175323486, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7048476338386536, + "num_tokens": 55400237.0, + "step": 2163 + }, + { + "epoch": 0.2376455084559631, + "grad_norm": 2.4279818534851074, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6946373581886292, + "num_tokens": 55423245.0, + "step": 2164 + }, + { + "epoch": 0.23775532615857675, + "grad_norm": 2.08248233795166, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7099951505661011, + "num_tokens": 55450654.0, + "step": 2165 + }, + { + "epoch": 0.23786514386119043, + "grad_norm": 2.228203535079956, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6948003768920898, + "num_tokens": 55476414.0, + "step": 2166 + }, + { + "epoch": 0.23797496156380407, + "grad_norm": 2.2238693237304688, + "learning_rate": 1e-06, + "loss": 1.0476, + "mean_token_accuracy": 0.6871346235275269, + "num_tokens": 55504316.0, + "step": 2167 + }, + { + "epoch": 0.23808477926641775, + "grad_norm": 2.307295322418213, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7098913192749023, + "num_tokens": 55529287.0, + "step": 2168 + }, + { + "epoch": 0.23819459696903142, + "grad_norm": 2.203805685043335, + "learning_rate": 1e-06, + "loss": 1.0457, + "mean_token_accuracy": 0.6865732669830322, + "num_tokens": 55555046.0, + "step": 2169 + }, + { + "epoch": 0.23830441467164507, + "grad_norm": 2.2894532680511475, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7132295966148376, + "num_tokens": 55580706.0, + "step": 2170 + }, + { + "epoch": 0.23841423237425874, + "grad_norm": 2.335843801498413, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.6941438913345337, + "num_tokens": 55606534.0, + "step": 2171 + }, + { + "epoch": 0.23852405007687238, + "grad_norm": 2.390393018722534, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.711012065410614, + "num_tokens": 55628569.0, + "step": 2172 + }, + { + "epoch": 0.23863386777948606, + "grad_norm": 2.147515296936035, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6922037601470947, + "num_tokens": 55656293.0, + "step": 2173 + }, + { + "epoch": 0.2387436854820997, + "grad_norm": 2.001889944076538, + "learning_rate": 1e-06, + "loss": 1.0789, + "mean_token_accuracy": 0.6795181035995483, + "num_tokens": 55690015.0, + "step": 2174 + }, + { + "epoch": 0.23885350318471338, + "grad_norm": 2.2084033489227295, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.7000757455825806, + "num_tokens": 55717063.0, + "step": 2175 + }, + { + "epoch": 0.23896332088732702, + "grad_norm": 2.0352914333343506, + "learning_rate": 1e-06, + "loss": 1.051, + "mean_token_accuracy": 0.6799054741859436, + "num_tokens": 55750121.0, + "step": 2176 + }, + { + "epoch": 0.2390731385899407, + "grad_norm": 2.9001030921936035, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7032392024993896, + "num_tokens": 55765805.0, + "step": 2177 + }, + { + "epoch": 0.23918295629255437, + "grad_norm": 2.5176432132720947, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6985211372375488, + "num_tokens": 55787496.0, + "step": 2178 + }, + { + "epoch": 0.23929277399516802, + "grad_norm": 2.042985677719116, + "learning_rate": 1e-06, + "loss": 1.0585, + "mean_token_accuracy": 0.6797275543212891, + "num_tokens": 55818523.0, + "step": 2179 + }, + { + "epoch": 0.2394025916977817, + "grad_norm": 2.3697874546051025, + "learning_rate": 1e-06, + "loss": 1.0331, + "mean_token_accuracy": 0.6920138597488403, + "num_tokens": 55843999.0, + "step": 2180 + }, + { + "epoch": 0.23951240940039534, + "grad_norm": 2.3416788578033447, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6988449692726135, + "num_tokens": 55868285.0, + "step": 2181 + }, + { + "epoch": 0.239622227103009, + "grad_norm": 2.281531810760498, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.7006537914276123, + "num_tokens": 55892885.0, + "step": 2182 + }, + { + "epoch": 0.23973204480562266, + "grad_norm": 2.3053364753723145, + "learning_rate": 1e-06, + "loss": 1.0476, + "mean_token_accuracy": 0.6842601299285889, + "num_tokens": 55919192.0, + "step": 2183 + }, + { + "epoch": 0.23984186250823633, + "grad_norm": 2.1353299617767334, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.6980088949203491, + "num_tokens": 55944399.0, + "step": 2184 + }, + { + "epoch": 0.23995168021084998, + "grad_norm": 2.190425395965576, + "learning_rate": 1e-06, + "loss": 1.0114, + "mean_token_accuracy": 0.704321026802063, + "num_tokens": 55970310.0, + "step": 2185 + }, + { + "epoch": 0.24006149791346365, + "grad_norm": 2.403345823287964, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7125059366226196, + "num_tokens": 55993798.0, + "step": 2186 + }, + { + "epoch": 0.24017131561607732, + "grad_norm": 2.4211604595184326, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.6939224004745483, + "num_tokens": 56015843.0, + "step": 2187 + }, + { + "epoch": 0.24028113331869097, + "grad_norm": 1.8937225341796875, + "learning_rate": 1e-06, + "loss": 1.0216, + "mean_token_accuracy": 0.6892667412757874, + "num_tokens": 56050939.0, + "step": 2188 + }, + { + "epoch": 0.24039095102130464, + "grad_norm": 2.2168352603912354, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.7176489233970642, + "num_tokens": 56077087.0, + "step": 2189 + }, + { + "epoch": 0.2405007687239183, + "grad_norm": 2.2760486602783203, + "learning_rate": 1e-06, + "loss": 1.0525, + "mean_token_accuracy": 0.6897920370101929, + "num_tokens": 56102504.0, + "step": 2190 + }, + { + "epoch": 0.24061058642653196, + "grad_norm": 2.1891467571258545, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7056910991668701, + "num_tokens": 56130163.0, + "step": 2191 + }, + { + "epoch": 0.2407204041291456, + "grad_norm": 2.4404385089874268, + "learning_rate": 1e-06, + "loss": 1.0762, + "mean_token_accuracy": 0.6764756441116333, + "num_tokens": 56152999.0, + "step": 2192 + }, + { + "epoch": 0.24083022183175928, + "grad_norm": 2.268880605697632, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7126367092132568, + "num_tokens": 56177255.0, + "step": 2193 + }, + { + "epoch": 0.24094003953437293, + "grad_norm": 2.4042437076568604, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.7098976373672485, + "num_tokens": 56202398.0, + "step": 2194 + }, + { + "epoch": 0.2410498572369866, + "grad_norm": 2.4157474040985107, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7065852880477905, + "num_tokens": 56225386.0, + "step": 2195 + }, + { + "epoch": 0.24115967493960028, + "grad_norm": 2.2002973556518555, + "learning_rate": 1e-06, + "loss": 1.077, + "mean_token_accuracy": 0.6785928606987, + "num_tokens": 56254156.0, + "step": 2196 + }, + { + "epoch": 0.24126949264221392, + "grad_norm": 2.02083420753479, + "learning_rate": 1e-06, + "loss": 1.0772, + "mean_token_accuracy": 0.6913036108016968, + "num_tokens": 56285619.0, + "step": 2197 + }, + { + "epoch": 0.2413793103448276, + "grad_norm": 2.1645102500915527, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6913976669311523, + "num_tokens": 56312935.0, + "step": 2198 + }, + { + "epoch": 0.24148912804744124, + "grad_norm": 2.228790521621704, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.7009392976760864, + "num_tokens": 56338134.0, + "step": 2199 + }, + { + "epoch": 0.24159894575005492, + "grad_norm": 2.572611093521118, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.6942082643508911, + "num_tokens": 56360919.0, + "step": 2200 + }, + { + "epoch": 0.24170876345266856, + "grad_norm": 2.1882596015930176, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7270408868789673, + "num_tokens": 56385625.0, + "step": 2201 + }, + { + "epoch": 0.24181858115528224, + "grad_norm": 2.0399584770202637, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.6847898960113525, + "num_tokens": 56414777.0, + "step": 2202 + }, + { + "epoch": 0.24192839885789588, + "grad_norm": 2.076021909713745, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.6852731704711914, + "num_tokens": 56443228.0, + "step": 2203 + }, + { + "epoch": 0.24203821656050956, + "grad_norm": 2.1083154678344727, + "learning_rate": 1e-06, + "loss": 1.0613, + "mean_token_accuracy": 0.677230954170227, + "num_tokens": 56472585.0, + "step": 2204 + }, + { + "epoch": 0.2421480342631232, + "grad_norm": 2.281592845916748, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6939834356307983, + "num_tokens": 56498169.0, + "step": 2205 + }, + { + "epoch": 0.24225785196573688, + "grad_norm": 2.6044118404388428, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.6934263706207275, + "num_tokens": 56519650.0, + "step": 2206 + }, + { + "epoch": 0.24236766966835055, + "grad_norm": 2.3037400245666504, + "learning_rate": 1e-06, + "loss": 1.0874, + "mean_token_accuracy": 0.6720625162124634, + "num_tokens": 56544886.0, + "step": 2207 + }, + { + "epoch": 0.2424774873709642, + "grad_norm": 2.102776050567627, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7156745195388794, + "num_tokens": 56573612.0, + "step": 2208 + }, + { + "epoch": 0.24258730507357787, + "grad_norm": 2.0480215549468994, + "learning_rate": 1e-06, + "loss": 1.0871, + "mean_token_accuracy": 0.68038010597229, + "num_tokens": 56606160.0, + "step": 2209 + }, + { + "epoch": 0.24269712277619152, + "grad_norm": 2.509629726409912, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6932193636894226, + "num_tokens": 56628162.0, + "step": 2210 + }, + { + "epoch": 0.2428069404788052, + "grad_norm": 2.218379497528076, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.705539882183075, + "num_tokens": 56654380.0, + "step": 2211 + }, + { + "epoch": 0.24291675818141883, + "grad_norm": 2.2313432693481445, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7063748240470886, + "num_tokens": 56680240.0, + "step": 2212 + }, + { + "epoch": 0.2430265758840325, + "grad_norm": 2.2897963523864746, + "learning_rate": 1e-06, + "loss": 1.0477, + "mean_token_accuracy": 0.6867613196372986, + "num_tokens": 56705057.0, + "step": 2213 + }, + { + "epoch": 0.24313639358664615, + "grad_norm": 2.069488525390625, + "learning_rate": 1e-06, + "loss": 1.1057, + "mean_token_accuracy": 0.6701440811157227, + "num_tokens": 56735080.0, + "step": 2214 + }, + { + "epoch": 0.24324621128925983, + "grad_norm": 2.1543591022491455, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7027782201766968, + "num_tokens": 56763506.0, + "step": 2215 + }, + { + "epoch": 0.2433560289918735, + "grad_norm": 2.137519598007202, + "learning_rate": 1e-06, + "loss": 1.103, + "mean_token_accuracy": 0.6830763220787048, + "num_tokens": 56793674.0, + "step": 2216 + }, + { + "epoch": 0.24346584669448715, + "grad_norm": 2.2019989490509033, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7095332145690918, + "num_tokens": 56820757.0, + "step": 2217 + }, + { + "epoch": 0.24357566439710082, + "grad_norm": 2.35575270652771, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.6941102743148804, + "num_tokens": 56845166.0, + "step": 2218 + }, + { + "epoch": 0.24368548209971447, + "grad_norm": 2.561802625656128, + "learning_rate": 1e-06, + "loss": 1.069, + "mean_token_accuracy": 0.6755983829498291, + "num_tokens": 56867247.0, + "step": 2219 + }, + { + "epoch": 0.24379529980232814, + "grad_norm": 2.1404128074645996, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.689967155456543, + "num_tokens": 56897745.0, + "step": 2220 + }, + { + "epoch": 0.2439051175049418, + "grad_norm": 2.374166488647461, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7130321264266968, + "num_tokens": 56920346.0, + "step": 2221 + }, + { + "epoch": 0.24401493520755546, + "grad_norm": 2.040544271469116, + "learning_rate": 1e-06, + "loss": 1.0401, + "mean_token_accuracy": 0.6865241527557373, + "num_tokens": 56950337.0, + "step": 2222 + }, + { + "epoch": 0.2441247529101691, + "grad_norm": 2.293693780899048, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7097383737564087, + "num_tokens": 56973365.0, + "step": 2223 + }, + { + "epoch": 0.24423457061278278, + "grad_norm": 1.9957969188690186, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6986935138702393, + "num_tokens": 57002597.0, + "step": 2224 + }, + { + "epoch": 0.24434438831539645, + "grad_norm": 2.0642101764678955, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.699615478515625, + "num_tokens": 57031980.0, + "step": 2225 + }, + { + "epoch": 0.2444542060180101, + "grad_norm": 2.322408676147461, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.7083127498626709, + "num_tokens": 57056901.0, + "step": 2226 + }, + { + "epoch": 0.24456402372062377, + "grad_norm": 2.357926368713379, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7024665474891663, + "num_tokens": 57079601.0, + "step": 2227 + }, + { + "epoch": 0.24467384142323742, + "grad_norm": 2.3959572315216064, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7181735038757324, + "num_tokens": 57100511.0, + "step": 2228 + }, + { + "epoch": 0.2447836591258511, + "grad_norm": 2.240016222000122, + "learning_rate": 1e-06, + "loss": 1.0507, + "mean_token_accuracy": 0.6836038827896118, + "num_tokens": 57129460.0, + "step": 2229 + }, + { + "epoch": 0.24489347682846474, + "grad_norm": 2.271306037902832, + "learning_rate": 1e-06, + "loss": 1.0254, + "mean_token_accuracy": 0.6983535289764404, + "num_tokens": 57153194.0, + "step": 2230 + }, + { + "epoch": 0.2450032945310784, + "grad_norm": 2.0635018348693848, + "learning_rate": 1e-06, + "loss": 1.0873, + "mean_token_accuracy": 0.6837972402572632, + "num_tokens": 57184290.0, + "step": 2231 + }, + { + "epoch": 0.24511311223369206, + "grad_norm": 2.4143431186676025, + "learning_rate": 1e-06, + "loss": 1.0487, + "mean_token_accuracy": 0.6841105222702026, + "num_tokens": 57207551.0, + "step": 2232 + }, + { + "epoch": 0.24522292993630573, + "grad_norm": 2.2983076572418213, + "learning_rate": 1e-06, + "loss": 1.0321, + "mean_token_accuracy": 0.6918802857398987, + "num_tokens": 57233146.0, + "step": 2233 + }, + { + "epoch": 0.2453327476389194, + "grad_norm": 2.4916350841522217, + "learning_rate": 1e-06, + "loss": 1.0413, + "mean_token_accuracy": 0.6872308850288391, + "num_tokens": 57256159.0, + "step": 2234 + }, + { + "epoch": 0.24544256534153305, + "grad_norm": 2.6092019081115723, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7114380598068237, + "num_tokens": 57275403.0, + "step": 2235 + }, + { + "epoch": 0.24555238304414673, + "grad_norm": 2.3535149097442627, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7125610709190369, + "num_tokens": 57299073.0, + "step": 2236 + }, + { + "epoch": 0.24566220074676037, + "grad_norm": 2.1704752445220947, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.695494532585144, + "num_tokens": 57326725.0, + "step": 2237 + }, + { + "epoch": 0.24577201844937405, + "grad_norm": 2.2310633659362793, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.6936736702919006, + "num_tokens": 57352376.0, + "step": 2238 + }, + { + "epoch": 0.2458818361519877, + "grad_norm": 2.1976232528686523, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.6988080739974976, + "num_tokens": 57380607.0, + "step": 2239 + }, + { + "epoch": 0.24599165385460137, + "grad_norm": 2.2105698585510254, + "learning_rate": 1e-06, + "loss": 1.0491, + "mean_token_accuracy": 0.6887467503547668, + "num_tokens": 57409890.0, + "step": 2240 + }, + { + "epoch": 0.246101471557215, + "grad_norm": 2.3642899990081787, + "learning_rate": 1e-06, + "loss": 1.0418, + "mean_token_accuracy": 0.6868038773536682, + "num_tokens": 57433148.0, + "step": 2241 + }, + { + "epoch": 0.2462112892598287, + "grad_norm": 2.469254732131958, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7154539823532104, + "num_tokens": 57452251.0, + "step": 2242 + }, + { + "epoch": 0.24632110696244233, + "grad_norm": 2.2336807250976562, + "learning_rate": 1e-06, + "loss": 1.09, + "mean_token_accuracy": 0.6734712719917297, + "num_tokens": 57478644.0, + "step": 2243 + }, + { + "epoch": 0.246430924665056, + "grad_norm": 2.229649305343628, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.7079176902770996, + "num_tokens": 57506937.0, + "step": 2244 + }, + { + "epoch": 0.24654074236766968, + "grad_norm": 2.372525453567505, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.7145735621452332, + "num_tokens": 57530700.0, + "step": 2245 + }, + { + "epoch": 0.24665056007028333, + "grad_norm": 2.361842632293701, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7192844748497009, + "num_tokens": 57554638.0, + "step": 2246 + }, + { + "epoch": 0.246760377772897, + "grad_norm": 1.9245178699493408, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7191330194473267, + "num_tokens": 57587026.0, + "step": 2247 + }, + { + "epoch": 0.24687019547551065, + "grad_norm": 2.5460517406463623, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7212154269218445, + "num_tokens": 57607094.0, + "step": 2248 + }, + { + "epoch": 0.24698001317812432, + "grad_norm": 2.14518666267395, + "learning_rate": 1e-06, + "loss": 1.0397, + "mean_token_accuracy": 0.6849385499954224, + "num_tokens": 57634169.0, + "step": 2249 + }, + { + "epoch": 0.24708983088073797, + "grad_norm": 2.03334641456604, + "learning_rate": 1e-06, + "loss": 1.0665, + "mean_token_accuracy": 0.6765710115432739, + "num_tokens": 57664534.0, + "step": 2250 + }, + { + "epoch": 0.24719964858335164, + "grad_norm": 2.2137084007263184, + "learning_rate": 1e-06, + "loss": 1.0989, + "mean_token_accuracy": 0.6777471899986267, + "num_tokens": 57694255.0, + "step": 2251 + }, + { + "epoch": 0.24730946628596528, + "grad_norm": 2.1735074520111084, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7110522985458374, + "num_tokens": 57721480.0, + "step": 2252 + }, + { + "epoch": 0.24741928398857896, + "grad_norm": 2.3251445293426514, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7186214327812195, + "num_tokens": 57745874.0, + "step": 2253 + }, + { + "epoch": 0.24752910169119263, + "grad_norm": 2.2762322425842285, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7090545892715454, + "num_tokens": 57769643.0, + "step": 2254 + }, + { + "epoch": 0.24763891939380628, + "grad_norm": 2.1135904788970947, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.702125072479248, + "num_tokens": 57797112.0, + "step": 2255 + }, + { + "epoch": 0.24774873709641995, + "grad_norm": 2.2017648220062256, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6962438821792603, + "num_tokens": 57823308.0, + "step": 2256 + }, + { + "epoch": 0.2478585547990336, + "grad_norm": 1.9205305576324463, + "learning_rate": 1e-06, + "loss": 1.0573, + "mean_token_accuracy": 0.6822715401649475, + "num_tokens": 57859295.0, + "step": 2257 + }, + { + "epoch": 0.24796837250164727, + "grad_norm": 2.1254284381866455, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.6963056921958923, + "num_tokens": 57885245.0, + "step": 2258 + }, + { + "epoch": 0.24807819020426092, + "grad_norm": 2.2377965450286865, + "learning_rate": 1e-06, + "loss": 1.0719, + "mean_token_accuracy": 0.674058198928833, + "num_tokens": 57911463.0, + "step": 2259 + }, + { + "epoch": 0.2481880079068746, + "grad_norm": 2.1812069416046143, + "learning_rate": 1e-06, + "loss": 1.0529, + "mean_token_accuracy": 0.6938692927360535, + "num_tokens": 57938871.0, + "step": 2260 + }, + { + "epoch": 0.24829782560948824, + "grad_norm": 2.2732012271881104, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.708410918712616, + "num_tokens": 57962755.0, + "step": 2261 + }, + { + "epoch": 0.2484076433121019, + "grad_norm": 2.2910349369049072, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.698581337928772, + "num_tokens": 57987654.0, + "step": 2262 + }, + { + "epoch": 0.24851746101471558, + "grad_norm": 2.3351011276245117, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7108160257339478, + "num_tokens": 58010845.0, + "step": 2263 + }, + { + "epoch": 0.24862727871732923, + "grad_norm": 2.088179111480713, + "learning_rate": 1e-06, + "loss": 1.0582, + "mean_token_accuracy": 0.6820584535598755, + "num_tokens": 58042658.0, + "step": 2264 + }, + { + "epoch": 0.2487370964199429, + "grad_norm": 2.421515941619873, + "learning_rate": 1e-06, + "loss": 1.0751, + "mean_token_accuracy": 0.6887009143829346, + "num_tokens": 58067237.0, + "step": 2265 + }, + { + "epoch": 0.24884691412255655, + "grad_norm": 2.5396392345428467, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6978252530097961, + "num_tokens": 58087500.0, + "step": 2266 + }, + { + "epoch": 0.24895673182517022, + "grad_norm": 2.490386724472046, + "learning_rate": 1e-06, + "loss": 1.028, + "mean_token_accuracy": 0.6947548389434814, + "num_tokens": 58109312.0, + "step": 2267 + }, + { + "epoch": 0.24906654952778387, + "grad_norm": 2.203324317932129, + "learning_rate": 1e-06, + "loss": 1.1084, + "mean_token_accuracy": 0.6762747764587402, + "num_tokens": 58136967.0, + "step": 2268 + }, + { + "epoch": 0.24917636723039754, + "grad_norm": 2.244551420211792, + "learning_rate": 1e-06, + "loss": 1.0176, + "mean_token_accuracy": 0.7003650665283203, + "num_tokens": 58161602.0, + "step": 2269 + }, + { + "epoch": 0.2492861849330112, + "grad_norm": 2.420647382736206, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7101598978042603, + "num_tokens": 58182925.0, + "step": 2270 + }, + { + "epoch": 0.24939600263562486, + "grad_norm": 2.0880823135375977, + "learning_rate": 1e-06, + "loss": 1.1063, + "mean_token_accuracy": 0.6733367443084717, + "num_tokens": 58214170.0, + "step": 2271 + }, + { + "epoch": 0.24950582033823854, + "grad_norm": 2.1562752723693848, + "learning_rate": 1e-06, + "loss": 1.0544, + "mean_token_accuracy": 0.69393390417099, + "num_tokens": 58241810.0, + "step": 2272 + }, + { + "epoch": 0.24961563804085218, + "grad_norm": 2.1975555419921875, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7169236540794373, + "num_tokens": 58268403.0, + "step": 2273 + }, + { + "epoch": 0.24972545574346586, + "grad_norm": 2.2771384716033936, + "learning_rate": 1e-06, + "loss": 1.1188, + "mean_token_accuracy": 0.6852074861526489, + "num_tokens": 58294276.0, + "step": 2274 + }, + { + "epoch": 0.2498352734460795, + "grad_norm": 2.403968572616577, + "learning_rate": 1e-06, + "loss": 1.068, + "mean_token_accuracy": 0.6780656576156616, + "num_tokens": 58317775.0, + "step": 2275 + }, + { + "epoch": 0.24994509114869318, + "grad_norm": 2.3598897457122803, + "learning_rate": 1e-06, + "loss": 1.0628, + "mean_token_accuracy": 0.6854151487350464, + "num_tokens": 58342475.0, + "step": 2276 + }, + { + "epoch": 0.2500549088513068, + "grad_norm": 2.2059736251831055, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.6929256916046143, + "num_tokens": 58368453.0, + "step": 2277 + }, + { + "epoch": 0.25016472655392047, + "grad_norm": 2.179542303085327, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7243815064430237, + "num_tokens": 58394513.0, + "step": 2278 + }, + { + "epoch": 0.25027454425653417, + "grad_norm": 2.3409993648529053, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7146073579788208, + "num_tokens": 58416294.0, + "step": 2279 + }, + { + "epoch": 0.2503843619591478, + "grad_norm": 2.2361462116241455, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.6934424042701721, + "num_tokens": 58442088.0, + "step": 2280 + }, + { + "epoch": 0.25049417966176146, + "grad_norm": 2.3062806129455566, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.6984596252441406, + "num_tokens": 58467126.0, + "step": 2281 + }, + { + "epoch": 0.25060399736437516, + "grad_norm": 2.384544849395752, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7104742527008057, + "num_tokens": 58490006.0, + "step": 2282 + }, + { + "epoch": 0.2507138150669888, + "grad_norm": 2.4584720134735107, + "learning_rate": 1e-06, + "loss": 1.113, + "mean_token_accuracy": 0.6667225360870361, + "num_tokens": 58513517.0, + "step": 2283 + }, + { + "epoch": 0.25082363276960246, + "grad_norm": 2.1800506114959717, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.704838752746582, + "num_tokens": 58540262.0, + "step": 2284 + }, + { + "epoch": 0.2509334504722161, + "grad_norm": 2.163198709487915, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7187372446060181, + "num_tokens": 58566630.0, + "step": 2285 + }, + { + "epoch": 0.2510432681748298, + "grad_norm": 2.245163917541504, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7048993110656738, + "num_tokens": 58589652.0, + "step": 2286 + }, + { + "epoch": 0.25115308587744345, + "grad_norm": 2.1469879150390625, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.7081020474433899, + "num_tokens": 58615205.0, + "step": 2287 + }, + { + "epoch": 0.2512629035800571, + "grad_norm": 2.1716623306274414, + "learning_rate": 1e-06, + "loss": 1.0307, + "mean_token_accuracy": 0.6901931762695312, + "num_tokens": 58644031.0, + "step": 2288 + }, + { + "epoch": 0.25137272128267074, + "grad_norm": 2.0104012489318848, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.6939870119094849, + "num_tokens": 58675773.0, + "step": 2289 + }, + { + "epoch": 0.25148253898528444, + "grad_norm": 2.217937707901001, + "learning_rate": 1e-06, + "loss": 1.098, + "mean_token_accuracy": 0.6762239933013916, + "num_tokens": 58702808.0, + "step": 2290 + }, + { + "epoch": 0.2515923566878981, + "grad_norm": 2.232072114944458, + "learning_rate": 1e-06, + "loss": 1.0791, + "mean_token_accuracy": 0.6835111379623413, + "num_tokens": 58729044.0, + "step": 2291 + }, + { + "epoch": 0.25170217439051173, + "grad_norm": 2.3524487018585205, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.704831600189209, + "num_tokens": 58751783.0, + "step": 2292 + }, + { + "epoch": 0.25181199209312544, + "grad_norm": 2.440955877304077, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7144544124603271, + "num_tokens": 58772571.0, + "step": 2293 + }, + { + "epoch": 0.2519218097957391, + "grad_norm": 2.479362964630127, + "learning_rate": 1e-06, + "loss": 1.0249, + "mean_token_accuracy": 0.6937781572341919, + "num_tokens": 58794702.0, + "step": 2294 + }, + { + "epoch": 0.25203162749835273, + "grad_norm": 2.518660545349121, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7036684155464172, + "num_tokens": 58816889.0, + "step": 2295 + }, + { + "epoch": 0.2521414452009664, + "grad_norm": 2.2353999614715576, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6980942487716675, + "num_tokens": 58843804.0, + "step": 2296 + }, + { + "epoch": 0.2522512629035801, + "grad_norm": 2.360490560531616, + "learning_rate": 1e-06, + "loss": 1.1157, + "mean_token_accuracy": 0.6691930294036865, + "num_tokens": 58872439.0, + "step": 2297 + }, + { + "epoch": 0.2523610806061937, + "grad_norm": 2.0537607669830322, + "learning_rate": 1e-06, + "loss": 1.1264, + "mean_token_accuracy": 0.6619155406951904, + "num_tokens": 58903778.0, + "step": 2298 + }, + { + "epoch": 0.25247089830880737, + "grad_norm": 2.133699655532837, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6878175735473633, + "num_tokens": 58932005.0, + "step": 2299 + }, + { + "epoch": 0.252580716011421, + "grad_norm": 2.1729841232299805, + "learning_rate": 1e-06, + "loss": 1.041, + "mean_token_accuracy": 0.6898459196090698, + "num_tokens": 58961142.0, + "step": 2300 + }, + { + "epoch": 0.2526905337140347, + "grad_norm": 2.1451120376586914, + "learning_rate": 1e-06, + "loss": 1.0585, + "mean_token_accuracy": 0.6884509325027466, + "num_tokens": 58988206.0, + "step": 2301 + }, + { + "epoch": 0.25280035141664836, + "grad_norm": 2.083704948425293, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7091196775436401, + "num_tokens": 59015655.0, + "step": 2302 + }, + { + "epoch": 0.252910169119262, + "grad_norm": 2.242614984512329, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7307704091072083, + "num_tokens": 59040272.0, + "step": 2303 + }, + { + "epoch": 0.2530199868218757, + "grad_norm": 1.9630216360092163, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7026635408401489, + "num_tokens": 59070686.0, + "step": 2304 + }, + { + "epoch": 0.25312980452448935, + "grad_norm": 2.1594460010528564, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.7007200717926025, + "num_tokens": 59095916.0, + "step": 2305 + }, + { + "epoch": 0.253239622227103, + "grad_norm": 2.2509820461273193, + "learning_rate": 1e-06, + "loss": 1.0642, + "mean_token_accuracy": 0.6783993244171143, + "num_tokens": 59123751.0, + "step": 2306 + }, + { + "epoch": 0.25334943992971665, + "grad_norm": 2.438075065612793, + "learning_rate": 1e-06, + "loss": 1.0792, + "mean_token_accuracy": 0.6865098476409912, + "num_tokens": 59146072.0, + "step": 2307 + }, + { + "epoch": 0.25345925763233035, + "grad_norm": 2.224562168121338, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6961129903793335, + "num_tokens": 59174021.0, + "step": 2308 + }, + { + "epoch": 0.253569075334944, + "grad_norm": 2.1510350704193115, + "learning_rate": 1e-06, + "loss": 0.9913, + "mean_token_accuracy": 0.6991631388664246, + "num_tokens": 59201773.0, + "step": 2309 + }, + { + "epoch": 0.25367889303755764, + "grad_norm": 1.9923679828643799, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7086121439933777, + "num_tokens": 59230981.0, + "step": 2310 + }, + { + "epoch": 0.25378871074017134, + "grad_norm": 2.2542150020599365, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.7004170417785645, + "num_tokens": 59257373.0, + "step": 2311 + }, + { + "epoch": 0.253898528442785, + "grad_norm": 2.1171252727508545, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7096027731895447, + "num_tokens": 59283490.0, + "step": 2312 + }, + { + "epoch": 0.25400834614539863, + "grad_norm": 2.0066630840301514, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.696661114692688, + "num_tokens": 59314248.0, + "step": 2313 + }, + { + "epoch": 0.2541181638480123, + "grad_norm": 2.3892242908477783, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7137441635131836, + "num_tokens": 59336690.0, + "step": 2314 + }, + { + "epoch": 0.254227981550626, + "grad_norm": 2.2105612754821777, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.6899112462997437, + "num_tokens": 59362647.0, + "step": 2315 + }, + { + "epoch": 0.2543377992532396, + "grad_norm": 2.365067958831787, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.7003456354141235, + "num_tokens": 59387158.0, + "step": 2316 + }, + { + "epoch": 0.2544476169558533, + "grad_norm": 2.478738307952881, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.6888128519058228, + "num_tokens": 59409018.0, + "step": 2317 + }, + { + "epoch": 0.2545574346584669, + "grad_norm": 2.268702268600464, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6988972425460815, + "num_tokens": 59433435.0, + "step": 2318 + }, + { + "epoch": 0.2546672523610806, + "grad_norm": 2.4795820713043213, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7225340604782104, + "num_tokens": 59454706.0, + "step": 2319 + }, + { + "epoch": 0.25477707006369427, + "grad_norm": 2.2062783241271973, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6981057524681091, + "num_tokens": 59479711.0, + "step": 2320 + }, + { + "epoch": 0.2548868877663079, + "grad_norm": 2.0854806900024414, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.6910101771354675, + "num_tokens": 59508017.0, + "step": 2321 + }, + { + "epoch": 0.2549967054689216, + "grad_norm": 2.479979991912842, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6994703412055969, + "num_tokens": 59528864.0, + "step": 2322 + }, + { + "epoch": 0.25510652317153526, + "grad_norm": 2.312978744506836, + "learning_rate": 1e-06, + "loss": 1.0747, + "mean_token_accuracy": 0.67405104637146, + "num_tokens": 59552363.0, + "step": 2323 + }, + { + "epoch": 0.2552163408741489, + "grad_norm": 2.2553670406341553, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7028284072875977, + "num_tokens": 59576842.0, + "step": 2324 + }, + { + "epoch": 0.25532615857676255, + "grad_norm": 2.101889133453369, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.6924639940261841, + "num_tokens": 59604662.0, + "step": 2325 + }, + { + "epoch": 0.25543597627937625, + "grad_norm": 2.405679702758789, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7231925129890442, + "num_tokens": 59625364.0, + "step": 2326 + }, + { + "epoch": 0.2555457939819899, + "grad_norm": 2.086650848388672, + "learning_rate": 1e-06, + "loss": 1.1313, + "mean_token_accuracy": 0.6654362678527832, + "num_tokens": 59654597.0, + "step": 2327 + }, + { + "epoch": 0.25565561168460355, + "grad_norm": 2.1263697147369385, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.696302056312561, + "num_tokens": 59681424.0, + "step": 2328 + }, + { + "epoch": 0.25576542938721725, + "grad_norm": 2.1969244480133057, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.726926326751709, + "num_tokens": 59705685.0, + "step": 2329 + }, + { + "epoch": 0.2558752470898309, + "grad_norm": 2.2839887142181396, + "learning_rate": 1e-06, + "loss": 1.0816, + "mean_token_accuracy": 0.6745402812957764, + "num_tokens": 59731865.0, + "step": 2330 + }, + { + "epoch": 0.25598506479244454, + "grad_norm": 2.295706272125244, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.7092477083206177, + "num_tokens": 59756230.0, + "step": 2331 + }, + { + "epoch": 0.2560948824950582, + "grad_norm": 2.069566249847412, + "learning_rate": 1e-06, + "loss": 1.0548, + "mean_token_accuracy": 0.6823797225952148, + "num_tokens": 59786703.0, + "step": 2332 + }, + { + "epoch": 0.2562047001976719, + "grad_norm": 2.4799091815948486, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.716633141040802, + "num_tokens": 59807581.0, + "step": 2333 + }, + { + "epoch": 0.25631451790028553, + "grad_norm": 2.3879921436309814, + "learning_rate": 1e-06, + "loss": 1.0521, + "mean_token_accuracy": 0.6900049448013306, + "num_tokens": 59831208.0, + "step": 2334 + }, + { + "epoch": 0.2564243356028992, + "grad_norm": 2.3528366088867188, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6832911968231201, + "num_tokens": 59856581.0, + "step": 2335 + }, + { + "epoch": 0.2565341533055128, + "grad_norm": 2.1725542545318604, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.7053265571594238, + "num_tokens": 59886064.0, + "step": 2336 + }, + { + "epoch": 0.2566439710081265, + "grad_norm": 2.459254741668701, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7047941088676453, + "num_tokens": 59908218.0, + "step": 2337 + }, + { + "epoch": 0.25675378871074017, + "grad_norm": 2.1609151363372803, + "learning_rate": 1e-06, + "loss": 1.0829, + "mean_token_accuracy": 0.6800839900970459, + "num_tokens": 59938119.0, + "step": 2338 + }, + { + "epoch": 0.2568636064133538, + "grad_norm": 2.2224106788635254, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7080684900283813, + "num_tokens": 59963732.0, + "step": 2339 + }, + { + "epoch": 0.2569734241159675, + "grad_norm": 2.3143773078918457, + "learning_rate": 1e-06, + "loss": 1.0252, + "mean_token_accuracy": 0.6885696649551392, + "num_tokens": 59987351.0, + "step": 2340 + }, + { + "epoch": 0.25708324181858117, + "grad_norm": 2.109097480773926, + "learning_rate": 1e-06, + "loss": 1.0952, + "mean_token_accuracy": 0.6714810132980347, + "num_tokens": 60015706.0, + "step": 2341 + }, + { + "epoch": 0.2571930595211948, + "grad_norm": 2.302903652191162, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.6889904141426086, + "num_tokens": 60039845.0, + "step": 2342 + }, + { + "epoch": 0.25730287722380846, + "grad_norm": 2.237830400466919, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7119042873382568, + "num_tokens": 60064069.0, + "step": 2343 + }, + { + "epoch": 0.25741269492642216, + "grad_norm": 2.3237080574035645, + "learning_rate": 1e-06, + "loss": 1.0482, + "mean_token_accuracy": 0.6820257306098938, + "num_tokens": 60088260.0, + "step": 2344 + }, + { + "epoch": 0.2575225126290358, + "grad_norm": 2.169665575027466, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7110462784767151, + "num_tokens": 60114242.0, + "step": 2345 + }, + { + "epoch": 0.25763233033164945, + "grad_norm": 2.356818675994873, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.6872990131378174, + "num_tokens": 60136936.0, + "step": 2346 + }, + { + "epoch": 0.2577421480342631, + "grad_norm": 2.5839014053344727, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7114120125770569, + "num_tokens": 60156115.0, + "step": 2347 + }, + { + "epoch": 0.2578519657368768, + "grad_norm": 2.3365001678466797, + "learning_rate": 1e-06, + "loss": 1.0387, + "mean_token_accuracy": 0.685562014579773, + "num_tokens": 60180284.0, + "step": 2348 + }, + { + "epoch": 0.25796178343949044, + "grad_norm": 2.0871996879577637, + "learning_rate": 1e-06, + "loss": 1.0943, + "mean_token_accuracy": 0.6735533475875854, + "num_tokens": 60210079.0, + "step": 2349 + }, + { + "epoch": 0.2580716011421041, + "grad_norm": 2.326953887939453, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.7001818418502808, + "num_tokens": 60233608.0, + "step": 2350 + }, + { + "epoch": 0.2581814188447178, + "grad_norm": 2.2266407012939453, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6923282146453857, + "num_tokens": 60259318.0, + "step": 2351 + }, + { + "epoch": 0.25829123654733144, + "grad_norm": 2.1974198818206787, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7029057145118713, + "num_tokens": 60286390.0, + "step": 2352 + }, + { + "epoch": 0.2584010542499451, + "grad_norm": 2.143967628479004, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.7062627077102661, + "num_tokens": 60316388.0, + "step": 2353 + }, + { + "epoch": 0.25851087195255873, + "grad_norm": 2.5960304737091064, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.73455411195755, + "num_tokens": 60334859.0, + "step": 2354 + }, + { + "epoch": 0.25862068965517243, + "grad_norm": 2.211332082748413, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6992648839950562, + "num_tokens": 60361420.0, + "step": 2355 + }, + { + "epoch": 0.2587305073577861, + "grad_norm": 2.3856735229492188, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.6958962678909302, + "num_tokens": 60382456.0, + "step": 2356 + }, + { + "epoch": 0.2588403250603997, + "grad_norm": 2.5685062408447266, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7091501355171204, + "num_tokens": 60402745.0, + "step": 2357 + }, + { + "epoch": 0.2589501427630134, + "grad_norm": 2.152604341506958, + "learning_rate": 1e-06, + "loss": 1.0382, + "mean_token_accuracy": 0.6831388473510742, + "num_tokens": 60429588.0, + "step": 2358 + }, + { + "epoch": 0.25905996046562707, + "grad_norm": 2.2013099193573, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6975699663162231, + "num_tokens": 60456128.0, + "step": 2359 + }, + { + "epoch": 0.2591697781682407, + "grad_norm": 1.9974479675292969, + "learning_rate": 1e-06, + "loss": 1.0222, + "mean_token_accuracy": 0.6891297101974487, + "num_tokens": 60485824.0, + "step": 2360 + }, + { + "epoch": 0.25927959587085436, + "grad_norm": 2.580003261566162, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7154499888420105, + "num_tokens": 60504079.0, + "step": 2361 + }, + { + "epoch": 0.25938941357346806, + "grad_norm": 2.148820400238037, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7136206030845642, + "num_tokens": 60528796.0, + "step": 2362 + }, + { + "epoch": 0.2594992312760817, + "grad_norm": 2.3542823791503906, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7017821073532104, + "num_tokens": 60552024.0, + "step": 2363 + }, + { + "epoch": 0.25960904897869536, + "grad_norm": 2.2409205436706543, + "learning_rate": 1e-06, + "loss": 1.038, + "mean_token_accuracy": 0.6938130855560303, + "num_tokens": 60577764.0, + "step": 2364 + }, + { + "epoch": 0.259718866681309, + "grad_norm": 2.2495322227478027, + "learning_rate": 1e-06, + "loss": 1.0816, + "mean_token_accuracy": 0.6732100248336792, + "num_tokens": 60606660.0, + "step": 2365 + }, + { + "epoch": 0.2598286843839227, + "grad_norm": 2.088935613632202, + "learning_rate": 1e-06, + "loss": 1.0619, + "mean_token_accuracy": 0.6846827268600464, + "num_tokens": 60635120.0, + "step": 2366 + }, + { + "epoch": 0.25993850208653635, + "grad_norm": 2.3227646350860596, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.7019991874694824, + "num_tokens": 60659028.0, + "step": 2367 + }, + { + "epoch": 0.26004831978915, + "grad_norm": 2.3260498046875, + "learning_rate": 1e-06, + "loss": 1.0536, + "mean_token_accuracy": 0.6815140247344971, + "num_tokens": 60684069.0, + "step": 2368 + }, + { + "epoch": 0.2601581374917637, + "grad_norm": 2.1984212398529053, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6942429542541504, + "num_tokens": 60709423.0, + "step": 2369 + }, + { + "epoch": 0.26026795519437734, + "grad_norm": 2.466120958328247, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6915633678436279, + "num_tokens": 60732140.0, + "step": 2370 + }, + { + "epoch": 0.260377772896991, + "grad_norm": 2.612635374069214, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7082453966140747, + "num_tokens": 60752518.0, + "step": 2371 + }, + { + "epoch": 0.26048759059960463, + "grad_norm": 1.99578058719635, + "learning_rate": 1e-06, + "loss": 1.1093, + "mean_token_accuracy": 0.6765052676200867, + "num_tokens": 60787321.0, + "step": 2372 + }, + { + "epoch": 0.26059740830221834, + "grad_norm": 2.180943250656128, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7237097024917603, + "num_tokens": 60810882.0, + "step": 2373 + }, + { + "epoch": 0.260707226004832, + "grad_norm": 2.3226478099823, + "learning_rate": 1e-06, + "loss": 1.0441, + "mean_token_accuracy": 0.6886590719223022, + "num_tokens": 60835019.0, + "step": 2374 + }, + { + "epoch": 0.26081704370744563, + "grad_norm": 2.3227195739746094, + "learning_rate": 1e-06, + "loss": 1.1126, + "mean_token_accuracy": 0.6678181886672974, + "num_tokens": 60858903.0, + "step": 2375 + }, + { + "epoch": 0.2609268614100593, + "grad_norm": 2.441596508026123, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7139039635658264, + "num_tokens": 60879740.0, + "step": 2376 + }, + { + "epoch": 0.261036679112673, + "grad_norm": 2.262998104095459, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6993316411972046, + "num_tokens": 60905797.0, + "step": 2377 + }, + { + "epoch": 0.2611464968152866, + "grad_norm": 2.5358049869537354, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7015582323074341, + "num_tokens": 60926606.0, + "step": 2378 + }, + { + "epoch": 0.26125631451790027, + "grad_norm": 2.2441768646240234, + "learning_rate": 1e-06, + "loss": 1.0663, + "mean_token_accuracy": 0.6810123920440674, + "num_tokens": 60953069.0, + "step": 2379 + }, + { + "epoch": 0.26136613222051397, + "grad_norm": 2.183584213256836, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6914584636688232, + "num_tokens": 60981598.0, + "step": 2380 + }, + { + "epoch": 0.2614759499231276, + "grad_norm": 2.573272228240967, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.7063609957695007, + "num_tokens": 61002336.0, + "step": 2381 + }, + { + "epoch": 0.26158576762574126, + "grad_norm": 2.3013761043548584, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.7036977410316467, + "num_tokens": 61029736.0, + "step": 2382 + }, + { + "epoch": 0.2616955853283549, + "grad_norm": 2.56611704826355, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6960076689720154, + "num_tokens": 61050389.0, + "step": 2383 + }, + { + "epoch": 0.2618054030309686, + "grad_norm": 1.979732632637024, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.700837254524231, + "num_tokens": 61083555.0, + "step": 2384 + }, + { + "epoch": 0.26191522073358225, + "grad_norm": 2.1952455043792725, + "learning_rate": 1e-06, + "loss": 1.045, + "mean_token_accuracy": 0.6863572001457214, + "num_tokens": 61109289.0, + "step": 2385 + }, + { + "epoch": 0.2620250384361959, + "grad_norm": 2.256621837615967, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7095719575881958, + "num_tokens": 61135348.0, + "step": 2386 + }, + { + "epoch": 0.2621348561388096, + "grad_norm": 2.4251439571380615, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7030149698257446, + "num_tokens": 61157532.0, + "step": 2387 + }, + { + "epoch": 0.26224467384142325, + "grad_norm": 2.256809711456299, + "learning_rate": 1e-06, + "loss": 1.0029, + "mean_token_accuracy": 0.6959054470062256, + "num_tokens": 61183432.0, + "step": 2388 + }, + { + "epoch": 0.2623544915440369, + "grad_norm": 2.0267574787139893, + "learning_rate": 1e-06, + "loss": 1.0353, + "mean_token_accuracy": 0.6898998022079468, + "num_tokens": 61215014.0, + "step": 2389 + }, + { + "epoch": 0.26246430924665054, + "grad_norm": 2.518319606781006, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6942816972732544, + "num_tokens": 61237415.0, + "step": 2390 + }, + { + "epoch": 0.26257412694926424, + "grad_norm": 2.2657344341278076, + "learning_rate": 1e-06, + "loss": 1.0296, + "mean_token_accuracy": 0.6898562908172607, + "num_tokens": 61262105.0, + "step": 2391 + }, + { + "epoch": 0.2626839446518779, + "grad_norm": 2.1454570293426514, + "learning_rate": 1e-06, + "loss": 1.0369, + "mean_token_accuracy": 0.684663712978363, + "num_tokens": 61291086.0, + "step": 2392 + }, + { + "epoch": 0.26279376235449153, + "grad_norm": 2.314784049987793, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6869709491729736, + "num_tokens": 61315152.0, + "step": 2393 + }, + { + "epoch": 0.2629035800571052, + "grad_norm": 2.205519437789917, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6997379064559937, + "num_tokens": 61341979.0, + "step": 2394 + }, + { + "epoch": 0.2630133977597189, + "grad_norm": 2.3044474124908447, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7112893462181091, + "num_tokens": 61365147.0, + "step": 2395 + }, + { + "epoch": 0.2631232154623325, + "grad_norm": 1.8888773918151855, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.6922296285629272, + "num_tokens": 61398043.0, + "step": 2396 + }, + { + "epoch": 0.2632330331649462, + "grad_norm": 2.1362597942352295, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6880785226821899, + "num_tokens": 61426262.0, + "step": 2397 + }, + { + "epoch": 0.2633428508675599, + "grad_norm": 2.1790671348571777, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.7030743360519409, + "num_tokens": 61452970.0, + "step": 2398 + }, + { + "epoch": 0.2634526685701735, + "grad_norm": 2.1680357456207275, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6982653737068176, + "num_tokens": 61481797.0, + "step": 2399 + }, + { + "epoch": 0.26356248627278717, + "grad_norm": 2.05440354347229, + "learning_rate": 1e-06, + "loss": 1.0913, + "mean_token_accuracy": 0.6730861067771912, + "num_tokens": 61513799.0, + "step": 2400 + }, + { + "epoch": 0.2636723039754008, + "grad_norm": 2.0009617805480957, + "learning_rate": 1e-06, + "loss": 1.0682, + "mean_token_accuracy": 0.6761223077774048, + "num_tokens": 61544108.0, + "step": 2401 + }, + { + "epoch": 0.2637821216780145, + "grad_norm": 2.1156184673309326, + "learning_rate": 1e-06, + "loss": 1.0398, + "mean_token_accuracy": 0.6941618323326111, + "num_tokens": 61571542.0, + "step": 2402 + }, + { + "epoch": 0.26389193938062816, + "grad_norm": 2.1111881732940674, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7163208723068237, + "num_tokens": 61599149.0, + "step": 2403 + }, + { + "epoch": 0.2640017570832418, + "grad_norm": 2.3791046142578125, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7057311534881592, + "num_tokens": 61621366.0, + "step": 2404 + }, + { + "epoch": 0.2641115747858555, + "grad_norm": 2.3499350547790527, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7070705890655518, + "num_tokens": 61645484.0, + "step": 2405 + }, + { + "epoch": 0.26422139248846915, + "grad_norm": 2.3270840644836426, + "learning_rate": 1e-06, + "loss": 1.0531, + "mean_token_accuracy": 0.6916138529777527, + "num_tokens": 61670722.0, + "step": 2406 + }, + { + "epoch": 0.2643312101910828, + "grad_norm": 2.570974588394165, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6995487213134766, + "num_tokens": 61693323.0, + "step": 2407 + }, + { + "epoch": 0.26444102789369645, + "grad_norm": 2.2440896034240723, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7232795357704163, + "num_tokens": 61717837.0, + "step": 2408 + }, + { + "epoch": 0.26455084559631015, + "grad_norm": 2.150885820388794, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.6888887882232666, + "num_tokens": 61745218.0, + "step": 2409 + }, + { + "epoch": 0.2646606632989238, + "grad_norm": 2.0550777912139893, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7118311524391174, + "num_tokens": 61774036.0, + "step": 2410 + }, + { + "epoch": 0.26477048100153744, + "grad_norm": 2.347766399383545, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.693031907081604, + "num_tokens": 61799112.0, + "step": 2411 + }, + { + "epoch": 0.2648802987041511, + "grad_norm": 2.6846184730529785, + "learning_rate": 1e-06, + "loss": 1.0739, + "mean_token_accuracy": 0.6783883571624756, + "num_tokens": 61819787.0, + "step": 2412 + }, + { + "epoch": 0.2649901164067648, + "grad_norm": 2.39752459526062, + "learning_rate": 1e-06, + "loss": 1.0889, + "mean_token_accuracy": 0.6731582283973694, + "num_tokens": 61846507.0, + "step": 2413 + }, + { + "epoch": 0.26509993410937843, + "grad_norm": 2.4189276695251465, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6950652599334717, + "num_tokens": 61869769.0, + "step": 2414 + }, + { + "epoch": 0.2652097518119921, + "grad_norm": 2.645021915435791, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7009418606758118, + "num_tokens": 61888703.0, + "step": 2415 + }, + { + "epoch": 0.2653195695146058, + "grad_norm": 2.405613422393799, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.7058600783348083, + "num_tokens": 61912879.0, + "step": 2416 + }, + { + "epoch": 0.2654293872172194, + "grad_norm": 2.1289050579071045, + "learning_rate": 1e-06, + "loss": 1.0681, + "mean_token_accuracy": 0.6759078502655029, + "num_tokens": 61942259.0, + "step": 2417 + }, + { + "epoch": 0.26553920491983307, + "grad_norm": 1.9040073156356812, + "learning_rate": 1e-06, + "loss": 1.0429, + "mean_token_accuracy": 0.6932021379470825, + "num_tokens": 61977151.0, + "step": 2418 + }, + { + "epoch": 0.2656490226224467, + "grad_norm": 2.1772186756134033, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6921558380126953, + "num_tokens": 62004679.0, + "step": 2419 + }, + { + "epoch": 0.2657588403250604, + "grad_norm": 2.1705942153930664, + "learning_rate": 1e-06, + "loss": 1.1649, + "mean_token_accuracy": 0.6651187539100647, + "num_tokens": 62033532.0, + "step": 2420 + }, + { + "epoch": 0.26586865802767407, + "grad_norm": 2.140892267227173, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6986551880836487, + "num_tokens": 62059763.0, + "step": 2421 + }, + { + "epoch": 0.2659784757302877, + "grad_norm": 2.2174229621887207, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6935747265815735, + "num_tokens": 62089930.0, + "step": 2422 + }, + { + "epoch": 0.26608829343290136, + "grad_norm": 2.7482542991638184, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7266836166381836, + "num_tokens": 62105918.0, + "step": 2423 + }, + { + "epoch": 0.26619811113551506, + "grad_norm": 2.225987672805786, + "learning_rate": 1e-06, + "loss": 1.0523, + "mean_token_accuracy": 0.6840243935585022, + "num_tokens": 62131497.0, + "step": 2424 + }, + { + "epoch": 0.2663079288381287, + "grad_norm": 2.525909900665283, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7118495106697083, + "num_tokens": 62153530.0, + "step": 2425 + }, + { + "epoch": 0.26641774654074235, + "grad_norm": 2.099976062774658, + "learning_rate": 1e-06, + "loss": 1.0775, + "mean_token_accuracy": 0.6747172474861145, + "num_tokens": 62185562.0, + "step": 2426 + }, + { + "epoch": 0.26652756424335605, + "grad_norm": 2.3028061389923096, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.706612229347229, + "num_tokens": 62209523.0, + "step": 2427 + }, + { + "epoch": 0.2666373819459697, + "grad_norm": 1.9428669214248657, + "learning_rate": 1e-06, + "loss": 1.1134, + "mean_token_accuracy": 0.6730587482452393, + "num_tokens": 62244950.0, + "step": 2428 + }, + { + "epoch": 0.26674719964858334, + "grad_norm": 2.3357882499694824, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.7005451321601868, + "num_tokens": 62268181.0, + "step": 2429 + }, + { + "epoch": 0.266857017351197, + "grad_norm": 2.1731464862823486, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6922974586486816, + "num_tokens": 62292950.0, + "step": 2430 + }, + { + "epoch": 0.2669668350538107, + "grad_norm": 2.3455657958984375, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6941997408866882, + "num_tokens": 62317700.0, + "step": 2431 + }, + { + "epoch": 0.26707665275642434, + "grad_norm": 2.1372201442718506, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7060751914978027, + "num_tokens": 62343009.0, + "step": 2432 + }, + { + "epoch": 0.267186470459038, + "grad_norm": 2.3760387897491455, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7322235107421875, + "num_tokens": 62363447.0, + "step": 2433 + }, + { + "epoch": 0.2672962881616517, + "grad_norm": 2.0183358192443848, + "learning_rate": 1e-06, + "loss": 1.056, + "mean_token_accuracy": 0.6852149367332458, + "num_tokens": 62397010.0, + "step": 2434 + }, + { + "epoch": 0.26740610586426533, + "grad_norm": 2.2561967372894287, + "learning_rate": 1e-06, + "loss": 1.1136, + "mean_token_accuracy": 0.6700138449668884, + "num_tokens": 62425401.0, + "step": 2435 + }, + { + "epoch": 0.267515923566879, + "grad_norm": 2.4852120876312256, + "learning_rate": 1e-06, + "loss": 1.0614, + "mean_token_accuracy": 0.6836233139038086, + "num_tokens": 62450183.0, + "step": 2436 + }, + { + "epoch": 0.2676257412694926, + "grad_norm": 2.2981374263763428, + "learning_rate": 1e-06, + "loss": 1.0793, + "mean_token_accuracy": 0.677322268486023, + "num_tokens": 62474613.0, + "step": 2437 + }, + { + "epoch": 0.2677355589721063, + "grad_norm": 2.082087278366089, + "learning_rate": 1e-06, + "loss": 1.0474, + "mean_token_accuracy": 0.6852821707725525, + "num_tokens": 62503611.0, + "step": 2438 + }, + { + "epoch": 0.26784537667471997, + "grad_norm": 2.3321802616119385, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.716416597366333, + "num_tokens": 62527416.0, + "step": 2439 + }, + { + "epoch": 0.2679551943773336, + "grad_norm": 2.491736888885498, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.7065891027450562, + "num_tokens": 62548942.0, + "step": 2440 + }, + { + "epoch": 0.26806501207994726, + "grad_norm": 2.276463508605957, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6873574256896973, + "num_tokens": 62574346.0, + "step": 2441 + }, + { + "epoch": 0.26817482978256096, + "grad_norm": 2.102029800415039, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7105737924575806, + "num_tokens": 62602183.0, + "step": 2442 + }, + { + "epoch": 0.2682846474851746, + "grad_norm": 2.152651309967041, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.7003573179244995, + "num_tokens": 62628169.0, + "step": 2443 + }, + { + "epoch": 0.26839446518778826, + "grad_norm": 2.204988718032837, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7137086391448975, + "num_tokens": 62654727.0, + "step": 2444 + }, + { + "epoch": 0.26850428289040196, + "grad_norm": 2.181196451187134, + "learning_rate": 1e-06, + "loss": 1.0303, + "mean_token_accuracy": 0.6859154105186462, + "num_tokens": 62682548.0, + "step": 2445 + }, + { + "epoch": 0.2686141005930156, + "grad_norm": 2.6544229984283447, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7260599732398987, + "num_tokens": 62701920.0, + "step": 2446 + }, + { + "epoch": 0.26872391829562925, + "grad_norm": 2.0419676303863525, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6952767372131348, + "num_tokens": 62731995.0, + "step": 2447 + }, + { + "epoch": 0.2688337359982429, + "grad_norm": 2.1538634300231934, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6951959133148193, + "num_tokens": 62758742.0, + "step": 2448 + }, + { + "epoch": 0.2689435537008566, + "grad_norm": 2.4020071029663086, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7102779150009155, + "num_tokens": 62781159.0, + "step": 2449 + }, + { + "epoch": 0.26905337140347024, + "grad_norm": 2.3000383377075195, + "learning_rate": 1e-06, + "loss": 1.0662, + "mean_token_accuracy": 0.6842652559280396, + "num_tokens": 62807454.0, + "step": 2450 + }, + { + "epoch": 0.2691631891060839, + "grad_norm": 2.299257755279541, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.7051622867584229, + "num_tokens": 62833384.0, + "step": 2451 + }, + { + "epoch": 0.26927300680869753, + "grad_norm": 2.1556553840637207, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.7002125382423401, + "num_tokens": 62860738.0, + "step": 2452 + }, + { + "epoch": 0.26938282451131124, + "grad_norm": 2.2204089164733887, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.7005693316459656, + "num_tokens": 62887579.0, + "step": 2453 + }, + { + "epoch": 0.2694926422139249, + "grad_norm": 2.1755242347717285, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7065516710281372, + "num_tokens": 62912940.0, + "step": 2454 + }, + { + "epoch": 0.26960245991653853, + "grad_norm": 2.544889211654663, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7087737917900085, + "num_tokens": 62933443.0, + "step": 2455 + }, + { + "epoch": 0.26971227761915223, + "grad_norm": 2.0520179271698, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.6988067626953125, + "num_tokens": 62963707.0, + "step": 2456 + }, + { + "epoch": 0.2698220953217659, + "grad_norm": 2.342627763748169, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.6996113657951355, + "num_tokens": 62987011.0, + "step": 2457 + }, + { + "epoch": 0.2699319130243795, + "grad_norm": 2.3609519004821777, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7147989273071289, + "num_tokens": 63008626.0, + "step": 2458 + }, + { + "epoch": 0.27004173072699317, + "grad_norm": 2.505601406097412, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7022435665130615, + "num_tokens": 63028048.0, + "step": 2459 + }, + { + "epoch": 0.27015154842960687, + "grad_norm": 1.9172191619873047, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6927202939987183, + "num_tokens": 63059208.0, + "step": 2460 + }, + { + "epoch": 0.2702613661322205, + "grad_norm": 2.2047722339630127, + "learning_rate": 1e-06, + "loss": 1.0281, + "mean_token_accuracy": 0.6887925863265991, + "num_tokens": 63086174.0, + "step": 2461 + }, + { + "epoch": 0.27037118383483416, + "grad_norm": 2.2217700481414795, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7053009271621704, + "num_tokens": 63111422.0, + "step": 2462 + }, + { + "epoch": 0.27048100153744786, + "grad_norm": 2.182464838027954, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6921002864837646, + "num_tokens": 63136750.0, + "step": 2463 + }, + { + "epoch": 0.2705908192400615, + "grad_norm": 2.22052264213562, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.6928413510322571, + "num_tokens": 63162997.0, + "step": 2464 + }, + { + "epoch": 0.27070063694267515, + "grad_norm": 2.4421722888946533, + "learning_rate": 1e-06, + "loss": 1.0408, + "mean_token_accuracy": 0.6873328685760498, + "num_tokens": 63184082.0, + "step": 2465 + }, + { + "epoch": 0.2708104546452888, + "grad_norm": 2.5715420246124268, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6984807252883911, + "num_tokens": 63204132.0, + "step": 2466 + }, + { + "epoch": 0.2709202723479025, + "grad_norm": 2.3676035404205322, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.6962706446647644, + "num_tokens": 63228089.0, + "step": 2467 + }, + { + "epoch": 0.27103009005051615, + "grad_norm": 2.1944377422332764, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.6962496042251587, + "num_tokens": 63256098.0, + "step": 2468 + }, + { + "epoch": 0.2711399077531298, + "grad_norm": 2.093417167663574, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.71286940574646, + "num_tokens": 63283006.0, + "step": 2469 + }, + { + "epoch": 0.27124972545574344, + "grad_norm": 2.1208572387695312, + "learning_rate": 1e-06, + "loss": 1.0478, + "mean_token_accuracy": 0.6887779235839844, + "num_tokens": 63312602.0, + "step": 2470 + }, + { + "epoch": 0.27135954315835714, + "grad_norm": 2.43609881401062, + "learning_rate": 1e-06, + "loss": 1.0921, + "mean_token_accuracy": 0.6673942804336548, + "num_tokens": 63336684.0, + "step": 2471 + }, + { + "epoch": 0.2714693608609708, + "grad_norm": 2.218790054321289, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7065026760101318, + "num_tokens": 63362586.0, + "step": 2472 + }, + { + "epoch": 0.27157917856358443, + "grad_norm": 2.022475004196167, + "learning_rate": 1e-06, + "loss": 1.1051, + "mean_token_accuracy": 0.665701150894165, + "num_tokens": 63395714.0, + "step": 2473 + }, + { + "epoch": 0.27168899626619814, + "grad_norm": 2.453874349594116, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.6949924230575562, + "num_tokens": 63419647.0, + "step": 2474 + }, + { + "epoch": 0.2717988139688118, + "grad_norm": 2.122070074081421, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.696587085723877, + "num_tokens": 63447644.0, + "step": 2475 + }, + { + "epoch": 0.2719086316714254, + "grad_norm": 2.3778269290924072, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6912282705307007, + "num_tokens": 63471680.0, + "step": 2476 + }, + { + "epoch": 0.2720184493740391, + "grad_norm": 2.1302900314331055, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.7053030729293823, + "num_tokens": 63499447.0, + "step": 2477 + }, + { + "epoch": 0.2721282670766528, + "grad_norm": 2.235295295715332, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.7036216855049133, + "num_tokens": 63522879.0, + "step": 2478 + }, + { + "epoch": 0.2722380847792664, + "grad_norm": 2.1794774532318115, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6958955526351929, + "num_tokens": 63550965.0, + "step": 2479 + }, + { + "epoch": 0.27234790248188007, + "grad_norm": 2.3760628700256348, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.7006064057350159, + "num_tokens": 63575058.0, + "step": 2480 + }, + { + "epoch": 0.27245772018449377, + "grad_norm": 2.200052261352539, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6997182369232178, + "num_tokens": 63603170.0, + "step": 2481 + }, + { + "epoch": 0.2725675378871074, + "grad_norm": 1.9535892009735107, + "learning_rate": 1e-06, + "loss": 1.0351, + "mean_token_accuracy": 0.69745272397995, + "num_tokens": 63636313.0, + "step": 2482 + }, + { + "epoch": 0.27267735558972106, + "grad_norm": 2.468653917312622, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6954785585403442, + "num_tokens": 63658240.0, + "step": 2483 + }, + { + "epoch": 0.2727871732923347, + "grad_norm": 2.001567840576172, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6974878907203674, + "num_tokens": 63688448.0, + "step": 2484 + }, + { + "epoch": 0.2728969909949484, + "grad_norm": 2.476693630218506, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7131625413894653, + "num_tokens": 63709081.0, + "step": 2485 + }, + { + "epoch": 0.27300680869756205, + "grad_norm": 2.2124695777893066, + "learning_rate": 1e-06, + "loss": 1.0564, + "mean_token_accuracy": 0.6847926378250122, + "num_tokens": 63734122.0, + "step": 2486 + }, + { + "epoch": 0.2731166264001757, + "grad_norm": 1.9219214916229248, + "learning_rate": 1e-06, + "loss": 1.082, + "mean_token_accuracy": 0.6780920028686523, + "num_tokens": 63768898.0, + "step": 2487 + }, + { + "epoch": 0.27322644410278935, + "grad_norm": 2.2686002254486084, + "learning_rate": 1e-06, + "loss": 1.0513, + "mean_token_accuracy": 0.6873461604118347, + "num_tokens": 63795872.0, + "step": 2488 + }, + { + "epoch": 0.27333626180540305, + "grad_norm": 2.1026618480682373, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.7027609348297119, + "num_tokens": 63824239.0, + "step": 2489 + }, + { + "epoch": 0.2734460795080167, + "grad_norm": 2.0972907543182373, + "learning_rate": 1e-06, + "loss": 1.0846, + "mean_token_accuracy": 0.6816850900650024, + "num_tokens": 63852952.0, + "step": 2490 + }, + { + "epoch": 0.27355589721063034, + "grad_norm": 2.5609660148620605, + "learning_rate": 1e-06, + "loss": 1.0729, + "mean_token_accuracy": 0.6736879348754883, + "num_tokens": 63873317.0, + "step": 2491 + }, + { + "epoch": 0.27366571491324404, + "grad_norm": 2.3628122806549072, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7325959801673889, + "num_tokens": 63895638.0, + "step": 2492 + }, + { + "epoch": 0.2737755326158577, + "grad_norm": 2.642406940460205, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.6985554695129395, + "num_tokens": 63916340.0, + "step": 2493 + }, + { + "epoch": 0.27388535031847133, + "grad_norm": 2.263200521469116, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7223315238952637, + "num_tokens": 63940784.0, + "step": 2494 + }, + { + "epoch": 0.273995168021085, + "grad_norm": 2.5762805938720703, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7205013036727905, + "num_tokens": 63960934.0, + "step": 2495 + }, + { + "epoch": 0.2741049857236987, + "grad_norm": 2.1767356395721436, + "learning_rate": 1e-06, + "loss": 1.0256, + "mean_token_accuracy": 0.6862571239471436, + "num_tokens": 63990685.0, + "step": 2496 + }, + { + "epoch": 0.2742148034263123, + "grad_norm": 2.271568775177002, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.7136809825897217, + "num_tokens": 64015728.0, + "step": 2497 + }, + { + "epoch": 0.27432462112892597, + "grad_norm": 2.035649538040161, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.7016057372093201, + "num_tokens": 64046819.0, + "step": 2498 + }, + { + "epoch": 0.2744344388315396, + "grad_norm": 2.246263265609741, + "learning_rate": 1e-06, + "loss": 1.0569, + "mean_token_accuracy": 0.6861909627914429, + "num_tokens": 64075015.0, + "step": 2499 + }, + { + "epoch": 0.2745442565341533, + "grad_norm": 2.2876687049865723, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.704647421836853, + "num_tokens": 64099666.0, + "step": 2500 + }, + { + "epoch": 0.27465407423676697, + "grad_norm": 2.3472418785095215, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7059268951416016, + "num_tokens": 64125363.0, + "step": 2501 + }, + { + "epoch": 0.2747638919393806, + "grad_norm": 2.602536201477051, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7136068344116211, + "num_tokens": 64144314.0, + "step": 2502 + }, + { + "epoch": 0.2748737096419943, + "grad_norm": 2.4404122829437256, + "learning_rate": 1e-06, + "loss": 1.0736, + "mean_token_accuracy": 0.6827595233917236, + "num_tokens": 64167733.0, + "step": 2503 + }, + { + "epoch": 0.27498352734460796, + "grad_norm": 2.1764838695526123, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.6917089223861694, + "num_tokens": 64195115.0, + "step": 2504 + }, + { + "epoch": 0.2750933450472216, + "grad_norm": 2.1531901359558105, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.7033172845840454, + "num_tokens": 64220898.0, + "step": 2505 + }, + { + "epoch": 0.27520316274983525, + "grad_norm": 2.1733205318450928, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.7017790079116821, + "num_tokens": 64249062.0, + "step": 2506 + }, + { + "epoch": 0.27531298045244895, + "grad_norm": 2.3270554542541504, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7031326293945312, + "num_tokens": 64273423.0, + "step": 2507 + }, + { + "epoch": 0.2754227981550626, + "grad_norm": 2.5417532920837402, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.6934515237808228, + "num_tokens": 64294474.0, + "step": 2508 + }, + { + "epoch": 0.27553261585767624, + "grad_norm": 2.297229051589966, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.6976020336151123, + "num_tokens": 64319271.0, + "step": 2509 + }, + { + "epoch": 0.27564243356028995, + "grad_norm": 2.428680896759033, + "learning_rate": 1e-06, + "loss": 1.0783, + "mean_token_accuracy": 0.6744135618209839, + "num_tokens": 64341224.0, + "step": 2510 + }, + { + "epoch": 0.2757522512629036, + "grad_norm": 1.9513174295425415, + "learning_rate": 1e-06, + "loss": 1.0584, + "mean_token_accuracy": 0.6812140345573425, + "num_tokens": 64372910.0, + "step": 2511 + }, + { + "epoch": 0.27586206896551724, + "grad_norm": 2.3153998851776123, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6973429918289185, + "num_tokens": 64397208.0, + "step": 2512 + }, + { + "epoch": 0.2759718866681309, + "grad_norm": 2.275897979736328, + "learning_rate": 1e-06, + "loss": 1.0254, + "mean_token_accuracy": 0.6978297829627991, + "num_tokens": 64422589.0, + "step": 2513 + }, + { + "epoch": 0.2760817043707446, + "grad_norm": 2.4210045337677, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.6969451308250427, + "num_tokens": 64447592.0, + "step": 2514 + }, + { + "epoch": 0.27619152207335823, + "grad_norm": 2.60369610786438, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7149565815925598, + "num_tokens": 64468609.0, + "step": 2515 + }, + { + "epoch": 0.2763013397759719, + "grad_norm": 2.2346038818359375, + "learning_rate": 1e-06, + "loss": 1.0574, + "mean_token_accuracy": 0.6924113631248474, + "num_tokens": 64495551.0, + "step": 2516 + }, + { + "epoch": 0.2764111574785855, + "grad_norm": 2.5178754329681396, + "learning_rate": 1e-06, + "loss": 1.0398, + "mean_token_accuracy": 0.6902115345001221, + "num_tokens": 64517760.0, + "step": 2517 + }, + { + "epoch": 0.2765209751811992, + "grad_norm": 2.1289560794830322, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.699407696723938, + "num_tokens": 64551955.0, + "step": 2518 + }, + { + "epoch": 0.27663079288381287, + "grad_norm": 2.3009355068206787, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7010039687156677, + "num_tokens": 64577341.0, + "step": 2519 + }, + { + "epoch": 0.2767406105864265, + "grad_norm": 2.178487539291382, + "learning_rate": 1e-06, + "loss": 1.0418, + "mean_token_accuracy": 0.690735936164856, + "num_tokens": 64605571.0, + "step": 2520 + }, + { + "epoch": 0.2768504282890402, + "grad_norm": 2.3488245010375977, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.6943689584732056, + "num_tokens": 64629723.0, + "step": 2521 + }, + { + "epoch": 0.27696024599165386, + "grad_norm": 2.3609790802001953, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7147682905197144, + "num_tokens": 64652523.0, + "step": 2522 + }, + { + "epoch": 0.2770700636942675, + "grad_norm": 2.164844274520874, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.696770191192627, + "num_tokens": 64680858.0, + "step": 2523 + }, + { + "epoch": 0.27717988139688116, + "grad_norm": 2.5243265628814697, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7040741443634033, + "num_tokens": 64701135.0, + "step": 2524 + }, + { + "epoch": 0.27728969909949486, + "grad_norm": 2.275801658630371, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7093092799186707, + "num_tokens": 64725028.0, + "step": 2525 + }, + { + "epoch": 0.2773995168021085, + "grad_norm": 2.592322826385498, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6914048194885254, + "num_tokens": 64743917.0, + "step": 2526 + }, + { + "epoch": 0.27750933450472215, + "grad_norm": 2.0399489402770996, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.6982650756835938, + "num_tokens": 64775817.0, + "step": 2527 + }, + { + "epoch": 0.2776191522073358, + "grad_norm": 2.0483298301696777, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.7036833167076111, + "num_tokens": 64806635.0, + "step": 2528 + }, + { + "epoch": 0.2777289699099495, + "grad_norm": 2.1567606925964355, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7046247720718384, + "num_tokens": 64834990.0, + "step": 2529 + }, + { + "epoch": 0.27783878761256314, + "grad_norm": 2.1560165882110596, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7129971981048584, + "num_tokens": 64860335.0, + "step": 2530 + }, + { + "epoch": 0.2779486053151768, + "grad_norm": 2.552349805831909, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7120548486709595, + "num_tokens": 64879367.0, + "step": 2531 + }, + { + "epoch": 0.2780584230177905, + "grad_norm": 2.183189630508423, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7031203508377075, + "num_tokens": 64905197.0, + "step": 2532 + }, + { + "epoch": 0.27816824072040414, + "grad_norm": 2.3566625118255615, + "learning_rate": 1e-06, + "loss": 1.1081, + "mean_token_accuracy": 0.6651195883750916, + "num_tokens": 64930357.0, + "step": 2533 + }, + { + "epoch": 0.2782780584230178, + "grad_norm": 2.5466878414154053, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7082844972610474, + "num_tokens": 64953362.0, + "step": 2534 + }, + { + "epoch": 0.27838787612563143, + "grad_norm": 2.0689730644226074, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7273086309432983, + "num_tokens": 64981288.0, + "step": 2535 + }, + { + "epoch": 0.27849769382824513, + "grad_norm": 2.2108049392700195, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7205426692962646, + "num_tokens": 65006052.0, + "step": 2536 + }, + { + "epoch": 0.2786075115308588, + "grad_norm": 2.245647668838501, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7074156999588013, + "num_tokens": 65029367.0, + "step": 2537 + }, + { + "epoch": 0.2787173292334724, + "grad_norm": 2.320446014404297, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.7030611634254456, + "num_tokens": 65053892.0, + "step": 2538 + }, + { + "epoch": 0.2788271469360861, + "grad_norm": 1.9821759462356567, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6882196664810181, + "num_tokens": 65085814.0, + "step": 2539 + }, + { + "epoch": 0.27893696463869977, + "grad_norm": 2.4100522994995117, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6950153708457947, + "num_tokens": 65108919.0, + "step": 2540 + }, + { + "epoch": 0.2790467823413134, + "grad_norm": 2.0444300174713135, + "learning_rate": 1e-06, + "loss": 1.0793, + "mean_token_accuracy": 0.6828361749649048, + "num_tokens": 65138564.0, + "step": 2541 + }, + { + "epoch": 0.27915660004392706, + "grad_norm": 2.359593152999878, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6827456951141357, + "num_tokens": 65162599.0, + "step": 2542 + }, + { + "epoch": 0.27926641774654076, + "grad_norm": 2.1118807792663574, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7089157700538635, + "num_tokens": 65189994.0, + "step": 2543 + }, + { + "epoch": 0.2793762354491544, + "grad_norm": 2.4812748432159424, + "learning_rate": 1e-06, + "loss": 1.0256, + "mean_token_accuracy": 0.6898664236068726, + "num_tokens": 65213290.0, + "step": 2544 + }, + { + "epoch": 0.27948605315176805, + "grad_norm": 2.0818681716918945, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.6988836526870728, + "num_tokens": 65242043.0, + "step": 2545 + }, + { + "epoch": 0.2795958708543817, + "grad_norm": 2.421725273132324, + "learning_rate": 1e-06, + "loss": 1.0513, + "mean_token_accuracy": 0.6824566721916199, + "num_tokens": 65264010.0, + "step": 2546 + }, + { + "epoch": 0.2797056885569954, + "grad_norm": 2.347019910812378, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.6862891912460327, + "num_tokens": 65287449.0, + "step": 2547 + }, + { + "epoch": 0.27981550625960905, + "grad_norm": 2.0551211833953857, + "learning_rate": 1e-06, + "loss": 1.104, + "mean_token_accuracy": 0.6688084006309509, + "num_tokens": 65319568.0, + "step": 2548 + }, + { + "epoch": 0.2799253239622227, + "grad_norm": 2.2521400451660156, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7081721425056458, + "num_tokens": 65345505.0, + "step": 2549 + }, + { + "epoch": 0.2800351416648364, + "grad_norm": 2.3970160484313965, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.7024707198143005, + "num_tokens": 65368775.0, + "step": 2550 + }, + { + "epoch": 0.28014495936745004, + "grad_norm": 2.58842134475708, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.691830039024353, + "num_tokens": 65389222.0, + "step": 2551 + }, + { + "epoch": 0.2802547770700637, + "grad_norm": 2.355393648147583, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6924788355827332, + "num_tokens": 65412434.0, + "step": 2552 + }, + { + "epoch": 0.28036459477267733, + "grad_norm": 2.166747808456421, + "learning_rate": 1e-06, + "loss": 1.1041, + "mean_token_accuracy": 0.6723120212554932, + "num_tokens": 65441413.0, + "step": 2553 + }, + { + "epoch": 0.28047441247529104, + "grad_norm": 2.0424916744232178, + "learning_rate": 1e-06, + "loss": 1.0381, + "mean_token_accuracy": 0.6867715120315552, + "num_tokens": 65472275.0, + "step": 2554 + }, + { + "epoch": 0.2805842301779047, + "grad_norm": 2.3520121574401855, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7090801000595093, + "num_tokens": 65495272.0, + "step": 2555 + }, + { + "epoch": 0.2806940478805183, + "grad_norm": 2.3902618885040283, + "learning_rate": 1e-06, + "loss": 1.0264, + "mean_token_accuracy": 0.6917374134063721, + "num_tokens": 65517781.0, + "step": 2556 + }, + { + "epoch": 0.28080386558313203, + "grad_norm": 2.1434905529022217, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.7006205320358276, + "num_tokens": 65546293.0, + "step": 2557 + }, + { + "epoch": 0.2809136832857457, + "grad_norm": 2.5160210132598877, + "learning_rate": 1e-06, + "loss": 1.0806, + "mean_token_accuracy": 0.6736239194869995, + "num_tokens": 65569212.0, + "step": 2558 + }, + { + "epoch": 0.2810235009883593, + "grad_norm": 2.1843268871307373, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7146735191345215, + "num_tokens": 65593678.0, + "step": 2559 + }, + { + "epoch": 0.28113331869097297, + "grad_norm": 2.408625841140747, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7082594633102417, + "num_tokens": 65615959.0, + "step": 2560 + }, + { + "epoch": 0.28124313639358667, + "grad_norm": 2.1899847984313965, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7063568830490112, + "num_tokens": 65641216.0, + "step": 2561 + }, + { + "epoch": 0.2813529540962003, + "grad_norm": 2.1604464054107666, + "learning_rate": 1e-06, + "loss": 1.0663, + "mean_token_accuracy": 0.6823466420173645, + "num_tokens": 65668321.0, + "step": 2562 + }, + { + "epoch": 0.28146277179881396, + "grad_norm": 2.671013593673706, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7110413312911987, + "num_tokens": 65687468.0, + "step": 2563 + }, + { + "epoch": 0.2815725895014276, + "grad_norm": 2.193765878677368, + "learning_rate": 1e-06, + "loss": 1.0686, + "mean_token_accuracy": 0.6779945492744446, + "num_tokens": 65716658.0, + "step": 2564 + }, + { + "epoch": 0.2816824072040413, + "grad_norm": 2.2843666076660156, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7164064645767212, + "num_tokens": 65740727.0, + "step": 2565 + }, + { + "epoch": 0.28179222490665495, + "grad_norm": 2.350698947906494, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7193347215652466, + "num_tokens": 65762709.0, + "step": 2566 + }, + { + "epoch": 0.2819020426092686, + "grad_norm": 2.380690097808838, + "learning_rate": 1e-06, + "loss": 1.0976, + "mean_token_accuracy": 0.6811671257019043, + "num_tokens": 65786403.0, + "step": 2567 + }, + { + "epoch": 0.2820118603118823, + "grad_norm": 1.9790866374969482, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.6895407438278198, + "num_tokens": 65816536.0, + "step": 2568 + }, + { + "epoch": 0.28212167801449595, + "grad_norm": 2.473193645477295, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7151817083358765, + "num_tokens": 65837455.0, + "step": 2569 + }, + { + "epoch": 0.2822314957171096, + "grad_norm": 2.3082072734832764, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6964923739433289, + "num_tokens": 65861931.0, + "step": 2570 + }, + { + "epoch": 0.28234131341972324, + "grad_norm": 2.269379138946533, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.7063024044036865, + "num_tokens": 65886067.0, + "step": 2571 + }, + { + "epoch": 0.28245113112233694, + "grad_norm": 2.339938163757324, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7054762244224548, + "num_tokens": 65910188.0, + "step": 2572 + }, + { + "epoch": 0.2825609488249506, + "grad_norm": 2.230468988418579, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7293176651000977, + "num_tokens": 65933081.0, + "step": 2573 + }, + { + "epoch": 0.28267076652756423, + "grad_norm": 2.567782163619995, + "learning_rate": 1e-06, + "loss": 1.0495, + "mean_token_accuracy": 0.6841006875038147, + "num_tokens": 65955141.0, + "step": 2574 + }, + { + "epoch": 0.2827805842301779, + "grad_norm": 2.1222894191741943, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6896244883537292, + "num_tokens": 65981845.0, + "step": 2575 + }, + { + "epoch": 0.2828904019327916, + "grad_norm": 2.3443965911865234, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7018798589706421, + "num_tokens": 66005000.0, + "step": 2576 + }, + { + "epoch": 0.2830002196354052, + "grad_norm": 2.3918893337249756, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.7062444686889648, + "num_tokens": 66027750.0, + "step": 2577 + }, + { + "epoch": 0.28311003733801887, + "grad_norm": 2.1865110397338867, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.7017630338668823, + "num_tokens": 66054600.0, + "step": 2578 + }, + { + "epoch": 0.2832198550406326, + "grad_norm": 2.406277894973755, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7052571773529053, + "num_tokens": 66078625.0, + "step": 2579 + }, + { + "epoch": 0.2833296727432462, + "grad_norm": 2.302581548690796, + "learning_rate": 1e-06, + "loss": 1.0651, + "mean_token_accuracy": 0.684383749961853, + "num_tokens": 66105522.0, + "step": 2580 + }, + { + "epoch": 0.28343949044585987, + "grad_norm": 2.000251293182373, + "learning_rate": 1e-06, + "loss": 1.084, + "mean_token_accuracy": 0.6789394021034241, + "num_tokens": 66137259.0, + "step": 2581 + }, + { + "epoch": 0.2835493081484735, + "grad_norm": 2.1673059463500977, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7082539796829224, + "num_tokens": 66166896.0, + "step": 2582 + }, + { + "epoch": 0.2836591258510872, + "grad_norm": 2.2819983959198, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7162806987762451, + "num_tokens": 66190769.0, + "step": 2583 + }, + { + "epoch": 0.28376894355370086, + "grad_norm": 2.2577431201934814, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7058217525482178, + "num_tokens": 66218102.0, + "step": 2584 + }, + { + "epoch": 0.2838787612563145, + "grad_norm": 2.5661375522613525, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.6945816874504089, + "num_tokens": 66238856.0, + "step": 2585 + }, + { + "epoch": 0.2839885789589282, + "grad_norm": 2.1660478115081787, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7134556770324707, + "num_tokens": 66266937.0, + "step": 2586 + }, + { + "epoch": 0.28409839666154185, + "grad_norm": 1.8782870769500732, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7103801965713501, + "num_tokens": 66300734.0, + "step": 2587 + }, + { + "epoch": 0.2842082143641555, + "grad_norm": 2.316174030303955, + "learning_rate": 1e-06, + "loss": 1.0247, + "mean_token_accuracy": 0.6960300207138062, + "num_tokens": 66324952.0, + "step": 2588 + }, + { + "epoch": 0.28431803206676914, + "grad_norm": 2.43361759185791, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.6993529796600342, + "num_tokens": 66347031.0, + "step": 2589 + }, + { + "epoch": 0.28442784976938285, + "grad_norm": 2.087006092071533, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.7002002596855164, + "num_tokens": 66377233.0, + "step": 2590 + }, + { + "epoch": 0.2845376674719965, + "grad_norm": 2.2008330821990967, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6965930461883545, + "num_tokens": 66402441.0, + "step": 2591 + }, + { + "epoch": 0.28464748517461014, + "grad_norm": 2.2051796913146973, + "learning_rate": 1e-06, + "loss": 1.0552, + "mean_token_accuracy": 0.6853726506233215, + "num_tokens": 66428711.0, + "step": 2592 + }, + { + "epoch": 0.2847573028772238, + "grad_norm": 2.222007989883423, + "learning_rate": 1e-06, + "loss": 1.0951, + "mean_token_accuracy": 0.674849808216095, + "num_tokens": 66458301.0, + "step": 2593 + }, + { + "epoch": 0.2848671205798375, + "grad_norm": 2.3400702476501465, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.6970381140708923, + "num_tokens": 66481561.0, + "step": 2594 + }, + { + "epoch": 0.28497693828245113, + "grad_norm": 2.396690607070923, + "learning_rate": 1e-06, + "loss": 1.0477, + "mean_token_accuracy": 0.6934667825698853, + "num_tokens": 66505492.0, + "step": 2595 + }, + { + "epoch": 0.2850867559850648, + "grad_norm": 2.3338708877563477, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.6946927309036255, + "num_tokens": 66530034.0, + "step": 2596 + }, + { + "epoch": 0.2851965736876785, + "grad_norm": 2.2673282623291016, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.7124946713447571, + "num_tokens": 66555303.0, + "step": 2597 + }, + { + "epoch": 0.2853063913902921, + "grad_norm": 2.1181278228759766, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6999936103820801, + "num_tokens": 66583450.0, + "step": 2598 + }, + { + "epoch": 0.28541620909290577, + "grad_norm": 2.3502047061920166, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7129718661308289, + "num_tokens": 66607039.0, + "step": 2599 + }, + { + "epoch": 0.2855260267955194, + "grad_norm": 2.1955833435058594, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6896428465843201, + "num_tokens": 66635149.0, + "step": 2600 + }, + { + "epoch": 0.2856358444981331, + "grad_norm": 2.1646602153778076, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.709389328956604, + "num_tokens": 66661322.0, + "step": 2601 + }, + { + "epoch": 0.28574566220074676, + "grad_norm": 2.2008633613586426, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.7017449140548706, + "num_tokens": 66685895.0, + "step": 2602 + }, + { + "epoch": 0.2858554799033604, + "grad_norm": 2.068944215774536, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7266967296600342, + "num_tokens": 66712934.0, + "step": 2603 + }, + { + "epoch": 0.28596529760597406, + "grad_norm": 2.476720094680786, + "learning_rate": 1e-06, + "loss": 1.0942, + "mean_token_accuracy": 0.6777265071868896, + "num_tokens": 66735547.0, + "step": 2604 + }, + { + "epoch": 0.28607511530858776, + "grad_norm": 2.1195058822631836, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.7019757032394409, + "num_tokens": 66763327.0, + "step": 2605 + }, + { + "epoch": 0.2861849330112014, + "grad_norm": 2.231240749359131, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7060142159461975, + "num_tokens": 66788856.0, + "step": 2606 + }, + { + "epoch": 0.28629475071381505, + "grad_norm": 2.260206937789917, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7115767002105713, + "num_tokens": 66811949.0, + "step": 2607 + }, + { + "epoch": 0.28640456841642875, + "grad_norm": 2.3076906204223633, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6912311315536499, + "num_tokens": 66836979.0, + "step": 2608 + }, + { + "epoch": 0.2865143861190424, + "grad_norm": 2.4302544593811035, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7202039361000061, + "num_tokens": 66858460.0, + "step": 2609 + }, + { + "epoch": 0.28662420382165604, + "grad_norm": 2.486807107925415, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7066266536712646, + "num_tokens": 66879362.0, + "step": 2610 + }, + { + "epoch": 0.2867340215242697, + "grad_norm": 2.1606764793395996, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.6948830485343933, + "num_tokens": 66907810.0, + "step": 2611 + }, + { + "epoch": 0.2868438392268834, + "grad_norm": 1.959529995918274, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7070941925048828, + "num_tokens": 66937165.0, + "step": 2612 + }, + { + "epoch": 0.28695365692949704, + "grad_norm": 2.549440860748291, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7082536816596985, + "num_tokens": 66958314.0, + "step": 2613 + }, + { + "epoch": 0.2870634746321107, + "grad_norm": 2.385171890258789, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.6899771094322205, + "num_tokens": 66982530.0, + "step": 2614 + }, + { + "epoch": 0.2871732923347244, + "grad_norm": 2.3786513805389404, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.7026170492172241, + "num_tokens": 67007278.0, + "step": 2615 + }, + { + "epoch": 0.28728311003733803, + "grad_norm": 2.368868827819824, + "learning_rate": 1e-06, + "loss": 1.0884, + "mean_token_accuracy": 0.6819926500320435, + "num_tokens": 67032702.0, + "step": 2616 + }, + { + "epoch": 0.2873929277399517, + "grad_norm": 2.5669548511505127, + "learning_rate": 1e-06, + "loss": 1.0337, + "mean_token_accuracy": 0.6905133128166199, + "num_tokens": 67052840.0, + "step": 2617 + }, + { + "epoch": 0.2875027454425653, + "grad_norm": 2.4898977279663086, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6909563541412354, + "num_tokens": 67074347.0, + "step": 2618 + }, + { + "epoch": 0.287612563145179, + "grad_norm": 2.610100746154785, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6949360966682434, + "num_tokens": 67097029.0, + "step": 2619 + }, + { + "epoch": 0.28772238084779267, + "grad_norm": 2.2256834506988525, + "learning_rate": 1e-06, + "loss": 1.0374, + "mean_token_accuracy": 0.6869096159934998, + "num_tokens": 67121854.0, + "step": 2620 + }, + { + "epoch": 0.2878321985504063, + "grad_norm": 2.433772325515747, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7193912267684937, + "num_tokens": 67144223.0, + "step": 2621 + }, + { + "epoch": 0.28794201625301996, + "grad_norm": 2.4118378162384033, + "learning_rate": 1e-06, + "loss": 1.0909, + "mean_token_accuracy": 0.6794079542160034, + "num_tokens": 67170340.0, + "step": 2622 + }, + { + "epoch": 0.28805183395563366, + "grad_norm": 2.33851957321167, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7249513864517212, + "num_tokens": 67192618.0, + "step": 2623 + }, + { + "epoch": 0.2881616516582473, + "grad_norm": 2.5030949115753174, + "learning_rate": 1e-06, + "loss": 1.034, + "mean_token_accuracy": 0.6937094926834106, + "num_tokens": 67215651.0, + "step": 2624 + }, + { + "epoch": 0.28827146936086095, + "grad_norm": 2.082824230194092, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.7011436223983765, + "num_tokens": 67243162.0, + "step": 2625 + }, + { + "epoch": 0.28838128706347466, + "grad_norm": 2.272132635116577, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.6883271336555481, + "num_tokens": 67268539.0, + "step": 2626 + }, + { + "epoch": 0.2884911047660883, + "grad_norm": 2.384673595428467, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.7080007791519165, + "num_tokens": 67291639.0, + "step": 2627 + }, + { + "epoch": 0.28860092246870195, + "grad_norm": 2.207425117492676, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.7058176398277283, + "num_tokens": 67317037.0, + "step": 2628 + }, + { + "epoch": 0.2887107401713156, + "grad_norm": 2.1543221473693848, + "learning_rate": 1e-06, + "loss": 1.0507, + "mean_token_accuracy": 0.6840207576751709, + "num_tokens": 67345174.0, + "step": 2629 + }, + { + "epoch": 0.2888205578739293, + "grad_norm": 2.452854871749878, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7173951268196106, + "num_tokens": 67365836.0, + "step": 2630 + }, + { + "epoch": 0.28893037557654294, + "grad_norm": 2.219791889190674, + "learning_rate": 1e-06, + "loss": 1.043, + "mean_token_accuracy": 0.6816378831863403, + "num_tokens": 67391973.0, + "step": 2631 + }, + { + "epoch": 0.2890401932791566, + "grad_norm": 2.2316389083862305, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6955045461654663, + "num_tokens": 67419188.0, + "step": 2632 + }, + { + "epoch": 0.2891500109817703, + "grad_norm": 2.2268216609954834, + "learning_rate": 1e-06, + "loss": 1.1082, + "mean_token_accuracy": 0.6758730411529541, + "num_tokens": 67447074.0, + "step": 2633 + }, + { + "epoch": 0.28925982868438394, + "grad_norm": 2.06103777885437, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.7003018260002136, + "num_tokens": 67477867.0, + "step": 2634 + }, + { + "epoch": 0.2893696463869976, + "grad_norm": 2.229897975921631, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.697199285030365, + "num_tokens": 67503564.0, + "step": 2635 + }, + { + "epoch": 0.2894794640896112, + "grad_norm": 2.4168944358825684, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.6963071227073669, + "num_tokens": 67526592.0, + "step": 2636 + }, + { + "epoch": 0.28958928179222493, + "grad_norm": 2.3248677253723145, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.7017989754676819, + "num_tokens": 67550984.0, + "step": 2637 + }, + { + "epoch": 0.2896990994948386, + "grad_norm": 2.368501901626587, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.7009091973304749, + "num_tokens": 67574576.0, + "step": 2638 + }, + { + "epoch": 0.2898089171974522, + "grad_norm": 2.2304651737213135, + "learning_rate": 1e-06, + "loss": 1.0745, + "mean_token_accuracy": 0.6798215508460999, + "num_tokens": 67602362.0, + "step": 2639 + }, + { + "epoch": 0.28991873490006587, + "grad_norm": 2.210132598876953, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.7021220326423645, + "num_tokens": 67628337.0, + "step": 2640 + }, + { + "epoch": 0.29002855260267957, + "grad_norm": 2.20235538482666, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7013130784034729, + "num_tokens": 67653698.0, + "step": 2641 + }, + { + "epoch": 0.2901383703052932, + "grad_norm": 1.9290130138397217, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7205613851547241, + "num_tokens": 67685970.0, + "step": 2642 + }, + { + "epoch": 0.29024818800790686, + "grad_norm": 2.1981663703918457, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7194898128509521, + "num_tokens": 67711133.0, + "step": 2643 + }, + { + "epoch": 0.29035800571052056, + "grad_norm": 2.363986015319824, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7116812467575073, + "num_tokens": 67733294.0, + "step": 2644 + }, + { + "epoch": 0.2904678234131342, + "grad_norm": 2.1687822341918945, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.695644736289978, + "num_tokens": 67761077.0, + "step": 2645 + }, + { + "epoch": 0.29057764111574785, + "grad_norm": 2.3310139179229736, + "learning_rate": 1e-06, + "loss": 1.0726, + "mean_token_accuracy": 0.6825960278511047, + "num_tokens": 67784737.0, + "step": 2646 + }, + { + "epoch": 0.2906874588183615, + "grad_norm": 2.0602071285247803, + "learning_rate": 1e-06, + "loss": 1.0681, + "mean_token_accuracy": 0.6858008503913879, + "num_tokens": 67815646.0, + "step": 2647 + }, + { + "epoch": 0.2907972765209752, + "grad_norm": 2.1300644874572754, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7090540528297424, + "num_tokens": 67841474.0, + "step": 2648 + }, + { + "epoch": 0.29090709422358885, + "grad_norm": 2.644777536392212, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7044399976730347, + "num_tokens": 67862152.0, + "step": 2649 + }, + { + "epoch": 0.2910169119262025, + "grad_norm": 2.1320905685424805, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6919313073158264, + "num_tokens": 67889839.0, + "step": 2650 + }, + { + "epoch": 0.29112672962881614, + "grad_norm": 2.3958446979522705, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6892440915107727, + "num_tokens": 67915339.0, + "step": 2651 + }, + { + "epoch": 0.29123654733142984, + "grad_norm": 2.3652236461639404, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7010098695755005, + "num_tokens": 67939062.0, + "step": 2652 + }, + { + "epoch": 0.2913463650340435, + "grad_norm": 2.394603729248047, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.7007250785827637, + "num_tokens": 67960744.0, + "step": 2653 + }, + { + "epoch": 0.29145618273665713, + "grad_norm": 2.536593437194824, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7228614687919617, + "num_tokens": 67980889.0, + "step": 2654 + }, + { + "epoch": 0.29156600043927083, + "grad_norm": 2.1753997802734375, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.7029775381088257, + "num_tokens": 68007041.0, + "step": 2655 + }, + { + "epoch": 0.2916758181418845, + "grad_norm": 2.6307668685913086, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7058249115943909, + "num_tokens": 68027420.0, + "step": 2656 + }, + { + "epoch": 0.2917856358444981, + "grad_norm": 2.2902746200561523, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.6919615268707275, + "num_tokens": 68052813.0, + "step": 2657 + }, + { + "epoch": 0.29189545354711177, + "grad_norm": 2.394535541534424, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.7019195556640625, + "num_tokens": 68074904.0, + "step": 2658 + }, + { + "epoch": 0.2920052712497255, + "grad_norm": 2.360907554626465, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.696847677230835, + "num_tokens": 68097573.0, + "step": 2659 + }, + { + "epoch": 0.2921150889523391, + "grad_norm": 2.323061466217041, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.719161868095398, + "num_tokens": 68119914.0, + "step": 2660 + }, + { + "epoch": 0.29222490665495277, + "grad_norm": 2.17901349067688, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6892935633659363, + "num_tokens": 68143513.0, + "step": 2661 + }, + { + "epoch": 0.29233472435756647, + "grad_norm": 2.3165106773376465, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7107927799224854, + "num_tokens": 68168487.0, + "step": 2662 + }, + { + "epoch": 0.2924445420601801, + "grad_norm": 2.3322582244873047, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6904667615890503, + "num_tokens": 68192483.0, + "step": 2663 + }, + { + "epoch": 0.29255435976279376, + "grad_norm": 2.2194628715515137, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7241271734237671, + "num_tokens": 68217500.0, + "step": 2664 + }, + { + "epoch": 0.2926641774654074, + "grad_norm": 2.4362781047821045, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.702970027923584, + "num_tokens": 68239745.0, + "step": 2665 + }, + { + "epoch": 0.2927739951680211, + "grad_norm": 2.376675844192505, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7126359343528748, + "num_tokens": 68261014.0, + "step": 2666 + }, + { + "epoch": 0.29288381287063475, + "grad_norm": 2.0796375274658203, + "learning_rate": 1e-06, + "loss": 1.0529, + "mean_token_accuracy": 0.6860125064849854, + "num_tokens": 68288964.0, + "step": 2667 + }, + { + "epoch": 0.2929936305732484, + "grad_norm": 2.1211700439453125, + "learning_rate": 1e-06, + "loss": 1.0536, + "mean_token_accuracy": 0.6936238408088684, + "num_tokens": 68317811.0, + "step": 2668 + }, + { + "epoch": 0.29310344827586204, + "grad_norm": 2.0363876819610596, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6989939212799072, + "num_tokens": 68347543.0, + "step": 2669 + }, + { + "epoch": 0.29321326597847575, + "grad_norm": 2.7080111503601074, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.709581732749939, + "num_tokens": 68365383.0, + "step": 2670 + }, + { + "epoch": 0.2933230836810894, + "grad_norm": 2.2471792697906494, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7050052285194397, + "num_tokens": 68390875.0, + "step": 2671 + }, + { + "epoch": 0.29343290138370304, + "grad_norm": 2.084528684616089, + "learning_rate": 1e-06, + "loss": 1.049, + "mean_token_accuracy": 0.6871029138565063, + "num_tokens": 68419460.0, + "step": 2672 + }, + { + "epoch": 0.29354271908631674, + "grad_norm": 2.5685389041900635, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7040817737579346, + "num_tokens": 68437816.0, + "step": 2673 + }, + { + "epoch": 0.2936525367889304, + "grad_norm": 2.1308910846710205, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.7011973857879639, + "num_tokens": 68465145.0, + "step": 2674 + }, + { + "epoch": 0.29376235449154403, + "grad_norm": 2.3106067180633545, + "learning_rate": 1e-06, + "loss": 1.1119, + "mean_token_accuracy": 0.672584593296051, + "num_tokens": 68491706.0, + "step": 2675 + }, + { + "epoch": 0.2938721721941577, + "grad_norm": 2.4320712089538574, + "learning_rate": 1e-06, + "loss": 1.0176, + "mean_token_accuracy": 0.6939617991447449, + "num_tokens": 68515204.0, + "step": 2676 + }, + { + "epoch": 0.2939819898967714, + "grad_norm": 2.282162666320801, + "learning_rate": 1e-06, + "loss": 1.0782, + "mean_token_accuracy": 0.6890614032745361, + "num_tokens": 68544409.0, + "step": 2677 + }, + { + "epoch": 0.294091807599385, + "grad_norm": 2.54738712310791, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7052555680274963, + "num_tokens": 68564925.0, + "step": 2678 + }, + { + "epoch": 0.29420162530199867, + "grad_norm": 2.3905296325683594, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7035491466522217, + "num_tokens": 68586051.0, + "step": 2679 + }, + { + "epoch": 0.2943114430046123, + "grad_norm": 2.506438732147217, + "learning_rate": 1e-06, + "loss": 1.0293, + "mean_token_accuracy": 0.6861369609832764, + "num_tokens": 68608144.0, + "step": 2680 + }, + { + "epoch": 0.294421260707226, + "grad_norm": 2.164391279220581, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6895278692245483, + "num_tokens": 68636422.0, + "step": 2681 + }, + { + "epoch": 0.29453107840983966, + "grad_norm": 2.4883906841278076, + "learning_rate": 1e-06, + "loss": 1.0246, + "mean_token_accuracy": 0.6916260719299316, + "num_tokens": 68656369.0, + "step": 2682 + }, + { + "epoch": 0.2946408961124533, + "grad_norm": 2.235178232192993, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.7079934477806091, + "num_tokens": 68682445.0, + "step": 2683 + }, + { + "epoch": 0.294750713815067, + "grad_norm": 2.290273904800415, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.6953433156013489, + "num_tokens": 68706743.0, + "step": 2684 + }, + { + "epoch": 0.29486053151768066, + "grad_norm": 2.3614397048950195, + "learning_rate": 1e-06, + "loss": 1.0471, + "mean_token_accuracy": 0.6751103401184082, + "num_tokens": 68728719.0, + "step": 2685 + }, + { + "epoch": 0.2949703492202943, + "grad_norm": 2.360271453857422, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7200996279716492, + "num_tokens": 68751490.0, + "step": 2686 + }, + { + "epoch": 0.29508016692290795, + "grad_norm": 2.370154619216919, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7066283226013184, + "num_tokens": 68774620.0, + "step": 2687 + }, + { + "epoch": 0.29518998462552165, + "grad_norm": 2.221001148223877, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.6967158913612366, + "num_tokens": 68800291.0, + "step": 2688 + }, + { + "epoch": 0.2952998023281353, + "grad_norm": 2.1186130046844482, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.7026459574699402, + "num_tokens": 68826895.0, + "step": 2689 + }, + { + "epoch": 0.29540962003074894, + "grad_norm": 2.79752516746521, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6942383050918579, + "num_tokens": 68844417.0, + "step": 2690 + }, + { + "epoch": 0.29551943773336264, + "grad_norm": 2.215812921524048, + "learning_rate": 1e-06, + "loss": 1.0516, + "mean_token_accuracy": 0.6882070302963257, + "num_tokens": 68871937.0, + "step": 2691 + }, + { + "epoch": 0.2956292554359763, + "grad_norm": 2.159154176712036, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7055532932281494, + "num_tokens": 68897720.0, + "step": 2692 + }, + { + "epoch": 0.29573907313858994, + "grad_norm": 2.161057949066162, + "learning_rate": 1e-06, + "loss": 1.0562, + "mean_token_accuracy": 0.6831163167953491, + "num_tokens": 68926879.0, + "step": 2693 + }, + { + "epoch": 0.2958488908412036, + "grad_norm": 2.207385301589966, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.700961172580719, + "num_tokens": 68953663.0, + "step": 2694 + }, + { + "epoch": 0.2959587085438173, + "grad_norm": 2.3605544567108154, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.701730489730835, + "num_tokens": 68976514.0, + "step": 2695 + }, + { + "epoch": 0.29606852624643093, + "grad_norm": 2.25291109085083, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7300532460212708, + "num_tokens": 69001124.0, + "step": 2696 + }, + { + "epoch": 0.2961783439490446, + "grad_norm": 2.2653348445892334, + "learning_rate": 1e-06, + "loss": 1.028, + "mean_token_accuracy": 0.7001834511756897, + "num_tokens": 69026396.0, + "step": 2697 + }, + { + "epoch": 0.2962881616516582, + "grad_norm": 2.038266181945801, + "learning_rate": 1e-06, + "loss": 1.0341, + "mean_token_accuracy": 0.6940964460372925, + "num_tokens": 69055417.0, + "step": 2698 + }, + { + "epoch": 0.2963979793542719, + "grad_norm": 2.6231513023376465, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6954193115234375, + "num_tokens": 69076103.0, + "step": 2699 + }, + { + "epoch": 0.29650779705688557, + "grad_norm": 2.239000082015991, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.6964102983474731, + "num_tokens": 69101240.0, + "step": 2700 + }, + { + "epoch": 0.2966176147594992, + "grad_norm": 2.1323137283325195, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.6868356466293335, + "num_tokens": 69128503.0, + "step": 2701 + }, + { + "epoch": 0.2967274324621129, + "grad_norm": 2.2366342544555664, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.719706654548645, + "num_tokens": 69153150.0, + "step": 2702 + }, + { + "epoch": 0.29683725016472656, + "grad_norm": 3.0395524501800537, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7292567491531372, + "num_tokens": 69167996.0, + "step": 2703 + }, + { + "epoch": 0.2969470678673402, + "grad_norm": 2.0273430347442627, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7231011986732483, + "num_tokens": 69194444.0, + "step": 2704 + }, + { + "epoch": 0.29705688556995385, + "grad_norm": 2.4412007331848145, + "learning_rate": 1e-06, + "loss": 1.0479, + "mean_token_accuracy": 0.6823304891586304, + "num_tokens": 69218506.0, + "step": 2705 + }, + { + "epoch": 0.29716670327256756, + "grad_norm": 2.606452703475952, + "learning_rate": 1e-06, + "loss": 1.0374, + "mean_token_accuracy": 0.6869847178459167, + "num_tokens": 69239168.0, + "step": 2706 + }, + { + "epoch": 0.2972765209751812, + "grad_norm": 2.325324535369873, + "learning_rate": 1e-06, + "loss": 1.0628, + "mean_token_accuracy": 0.6849586963653564, + "num_tokens": 69264485.0, + "step": 2707 + }, + { + "epoch": 0.29738633867779485, + "grad_norm": 2.3231122493743896, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6998949646949768, + "num_tokens": 69289944.0, + "step": 2708 + }, + { + "epoch": 0.29749615638040855, + "grad_norm": 2.447679281234741, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7112960815429688, + "num_tokens": 69310858.0, + "step": 2709 + }, + { + "epoch": 0.2976059740830222, + "grad_norm": 2.2191946506500244, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.6876577734947205, + "num_tokens": 69338214.0, + "step": 2710 + }, + { + "epoch": 0.29771579178563584, + "grad_norm": 2.1636130809783936, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6983034610748291, + "num_tokens": 69368673.0, + "step": 2711 + }, + { + "epoch": 0.2978256094882495, + "grad_norm": 2.1186330318450928, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.6946398019790649, + "num_tokens": 69396369.0, + "step": 2712 + }, + { + "epoch": 0.2979354271908632, + "grad_norm": 2.6423349380493164, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.719146728515625, + "num_tokens": 69414283.0, + "step": 2713 + }, + { + "epoch": 0.29804524489347684, + "grad_norm": 2.4829604625701904, + "learning_rate": 1e-06, + "loss": 1.071, + "mean_token_accuracy": 0.682335376739502, + "num_tokens": 69436546.0, + "step": 2714 + }, + { + "epoch": 0.2981550625960905, + "grad_norm": 2.1665403842926025, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7059996128082275, + "num_tokens": 69463520.0, + "step": 2715 + }, + { + "epoch": 0.2982648802987041, + "grad_norm": 2.5366475582122803, + "learning_rate": 1e-06, + "loss": 1.0611, + "mean_token_accuracy": 0.6808410286903381, + "num_tokens": 69487519.0, + "step": 2716 + }, + { + "epoch": 0.29837469800131783, + "grad_norm": 2.0703258514404297, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6864300966262817, + "num_tokens": 69518717.0, + "step": 2717 + }, + { + "epoch": 0.2984845157039315, + "grad_norm": 2.422443151473999, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7124596238136292, + "num_tokens": 69540071.0, + "step": 2718 + }, + { + "epoch": 0.2985943334065451, + "grad_norm": 2.5601704120635986, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7121923565864563, + "num_tokens": 69561175.0, + "step": 2719 + }, + { + "epoch": 0.2987041511091588, + "grad_norm": 2.471226930618286, + "learning_rate": 1e-06, + "loss": 1.0509, + "mean_token_accuracy": 0.6815121173858643, + "num_tokens": 69585349.0, + "step": 2720 + }, + { + "epoch": 0.29881396881177247, + "grad_norm": 2.2585668563842773, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.6958890557289124, + "num_tokens": 69613055.0, + "step": 2721 + }, + { + "epoch": 0.2989237865143861, + "grad_norm": 2.3919215202331543, + "learning_rate": 1e-06, + "loss": 1.0114, + "mean_token_accuracy": 0.6919739246368408, + "num_tokens": 69637489.0, + "step": 2722 + }, + { + "epoch": 0.29903360421699976, + "grad_norm": 2.2431037425994873, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7179856300354004, + "num_tokens": 69660580.0, + "step": 2723 + }, + { + "epoch": 0.29914342191961346, + "grad_norm": 2.169532537460327, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7151689529418945, + "num_tokens": 69686903.0, + "step": 2724 + }, + { + "epoch": 0.2992532396222271, + "grad_norm": 2.2921316623687744, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7003347873687744, + "num_tokens": 69714427.0, + "step": 2725 + }, + { + "epoch": 0.29936305732484075, + "grad_norm": 2.467735528945923, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.699588418006897, + "num_tokens": 69736597.0, + "step": 2726 + }, + { + "epoch": 0.2994728750274544, + "grad_norm": 2.8683481216430664, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7134132385253906, + "num_tokens": 69753679.0, + "step": 2727 + }, + { + "epoch": 0.2995826927300681, + "grad_norm": 2.428955554962158, + "learning_rate": 1e-06, + "loss": 1.0416, + "mean_token_accuracy": 0.6901091933250427, + "num_tokens": 69778329.0, + "step": 2728 + }, + { + "epoch": 0.29969251043268175, + "grad_norm": 2.508054733276367, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6946766376495361, + "num_tokens": 69800716.0, + "step": 2729 + }, + { + "epoch": 0.2998023281352954, + "grad_norm": 2.263535499572754, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7020784020423889, + "num_tokens": 69826098.0, + "step": 2730 + }, + { + "epoch": 0.2999121458379091, + "grad_norm": 2.3086743354797363, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.7039368748664856, + "num_tokens": 69852669.0, + "step": 2731 + }, + { + "epoch": 0.30002196354052274, + "grad_norm": 2.2910263538360596, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.7045056819915771, + "num_tokens": 69878846.0, + "step": 2732 + }, + { + "epoch": 0.3001317812431364, + "grad_norm": 2.4508626461029053, + "learning_rate": 1e-06, + "loss": 1.0377, + "mean_token_accuracy": 0.6976564526557922, + "num_tokens": 69902253.0, + "step": 2733 + }, + { + "epoch": 0.30024159894575003, + "grad_norm": 2.514502763748169, + "learning_rate": 1e-06, + "loss": 1.0357, + "mean_token_accuracy": 0.6947339773178101, + "num_tokens": 69926864.0, + "step": 2734 + }, + { + "epoch": 0.30035141664836373, + "grad_norm": 1.8613486289978027, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.7027002573013306, + "num_tokens": 69963994.0, + "step": 2735 + }, + { + "epoch": 0.3004612343509774, + "grad_norm": 2.154764175415039, + "learning_rate": 1e-06, + "loss": 1.0587, + "mean_token_accuracy": 0.6866186261177063, + "num_tokens": 69993217.0, + "step": 2736 + }, + { + "epoch": 0.300571052053591, + "grad_norm": 2.039027452468872, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7286697626113892, + "num_tokens": 70021762.0, + "step": 2737 + }, + { + "epoch": 0.3006808697562047, + "grad_norm": 2.362339496612549, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7294749021530151, + "num_tokens": 70043746.0, + "step": 2738 + }, + { + "epoch": 0.3007906874588184, + "grad_norm": 2.2288949489593506, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6984391212463379, + "num_tokens": 70067366.0, + "step": 2739 + }, + { + "epoch": 0.300900505161432, + "grad_norm": 2.154998302459717, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6933973431587219, + "num_tokens": 70095428.0, + "step": 2740 + }, + { + "epoch": 0.30101032286404567, + "grad_norm": 2.200754404067993, + "learning_rate": 1e-06, + "loss": 1.0526, + "mean_token_accuracy": 0.6807023286819458, + "num_tokens": 70124161.0, + "step": 2741 + }, + { + "epoch": 0.30112014056665937, + "grad_norm": 2.2852938175201416, + "learning_rate": 1e-06, + "loss": 1.0683, + "mean_token_accuracy": 0.6719629168510437, + "num_tokens": 70149850.0, + "step": 2742 + }, + { + "epoch": 0.301229958269273, + "grad_norm": 2.0714123249053955, + "learning_rate": 1e-06, + "loss": 1.0322, + "mean_token_accuracy": 0.6900376081466675, + "num_tokens": 70178287.0, + "step": 2743 + }, + { + "epoch": 0.30133977597188666, + "grad_norm": 2.4105281829833984, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7047341465950012, + "num_tokens": 70201399.0, + "step": 2744 + }, + { + "epoch": 0.3014495936745003, + "grad_norm": 1.9910203218460083, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6874673366546631, + "num_tokens": 70230859.0, + "step": 2745 + }, + { + "epoch": 0.301559411377114, + "grad_norm": 2.614820957183838, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.7173469066619873, + "num_tokens": 70251437.0, + "step": 2746 + }, + { + "epoch": 0.30166922907972765, + "grad_norm": 2.6512796878814697, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.70643150806427, + "num_tokens": 70271547.0, + "step": 2747 + }, + { + "epoch": 0.3017790467823413, + "grad_norm": 2.5431933403015137, + "learning_rate": 1e-06, + "loss": 1.1056, + "mean_token_accuracy": 0.6808602809906006, + "num_tokens": 70295430.0, + "step": 2748 + }, + { + "epoch": 0.301888864484955, + "grad_norm": 2.4301674365997314, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6865103244781494, + "num_tokens": 70318878.0, + "step": 2749 + }, + { + "epoch": 0.30199868218756865, + "grad_norm": 2.219835042953491, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6965452432632446, + "num_tokens": 70348451.0, + "step": 2750 + }, + { + "epoch": 0.3021084998901823, + "grad_norm": 2.3448359966278076, + "learning_rate": 1e-06, + "loss": 1.0317, + "mean_token_accuracy": 0.6839351654052734, + "num_tokens": 70371727.0, + "step": 2751 + }, + { + "epoch": 0.30221831759279594, + "grad_norm": 2.319458484649658, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.6874902248382568, + "num_tokens": 70395561.0, + "step": 2752 + }, + { + "epoch": 0.30232813529540964, + "grad_norm": 2.0616819858551025, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.692769467830658, + "num_tokens": 70427183.0, + "step": 2753 + }, + { + "epoch": 0.3024379529980233, + "grad_norm": 2.1698479652404785, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.7090818285942078, + "num_tokens": 70453753.0, + "step": 2754 + }, + { + "epoch": 0.30254777070063693, + "grad_norm": 2.063490152359009, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7180188894271851, + "num_tokens": 70483257.0, + "step": 2755 + }, + { + "epoch": 0.3026575884032506, + "grad_norm": 2.2084028720855713, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6930286884307861, + "num_tokens": 70509579.0, + "step": 2756 + }, + { + "epoch": 0.3027674061058643, + "grad_norm": 2.295809745788574, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6953785419464111, + "num_tokens": 70533319.0, + "step": 2757 + }, + { + "epoch": 0.3028772238084779, + "grad_norm": 1.9719170331954956, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.7030174732208252, + "num_tokens": 70563299.0, + "step": 2758 + }, + { + "epoch": 0.30298704151109157, + "grad_norm": 2.155402898788452, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7176016569137573, + "num_tokens": 70591873.0, + "step": 2759 + }, + { + "epoch": 0.30309685921370527, + "grad_norm": 2.661980390548706, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7197957038879395, + "num_tokens": 70609996.0, + "step": 2760 + }, + { + "epoch": 0.3032066769163189, + "grad_norm": 2.2213592529296875, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7108955383300781, + "num_tokens": 70635799.0, + "step": 2761 + }, + { + "epoch": 0.30331649461893256, + "grad_norm": 2.2747883796691895, + "learning_rate": 1e-06, + "loss": 1.0681, + "mean_token_accuracy": 0.6863242983818054, + "num_tokens": 70662382.0, + "step": 2762 + }, + { + "epoch": 0.3034263123215462, + "grad_norm": 2.300034761428833, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7121802568435669, + "num_tokens": 70684497.0, + "step": 2763 + }, + { + "epoch": 0.3035361300241599, + "grad_norm": 2.4083049297332764, + "learning_rate": 1e-06, + "loss": 1.0285, + "mean_token_accuracy": 0.6838868856430054, + "num_tokens": 70707153.0, + "step": 2764 + }, + { + "epoch": 0.30364594772677356, + "grad_norm": 2.2204325199127197, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7125944495201111, + "num_tokens": 70732828.0, + "step": 2765 + }, + { + "epoch": 0.3037557654293872, + "grad_norm": 2.4688985347747803, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.691704273223877, + "num_tokens": 70757607.0, + "step": 2766 + }, + { + "epoch": 0.3038655831320009, + "grad_norm": 2.16597580909729, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6943452954292297, + "num_tokens": 70786941.0, + "step": 2767 + }, + { + "epoch": 0.30397540083461455, + "grad_norm": 2.3138999938964844, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7184176445007324, + "num_tokens": 70812655.0, + "step": 2768 + }, + { + "epoch": 0.3040852185372282, + "grad_norm": 2.2917139530181885, + "learning_rate": 1e-06, + "loss": 1.0353, + "mean_token_accuracy": 0.6943790912628174, + "num_tokens": 70838651.0, + "step": 2769 + }, + { + "epoch": 0.30419503623984184, + "grad_norm": 2.483250379562378, + "learning_rate": 1e-06, + "loss": 1.0777, + "mean_token_accuracy": 0.6822527647018433, + "num_tokens": 70861239.0, + "step": 2770 + }, + { + "epoch": 0.30430485394245554, + "grad_norm": 2.5755796432495117, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.6990912556648254, + "num_tokens": 70882365.0, + "step": 2771 + }, + { + "epoch": 0.3044146716450692, + "grad_norm": 2.5573813915252686, + "learning_rate": 1e-06, + "loss": 1.0442, + "mean_token_accuracy": 0.6874129772186279, + "num_tokens": 70904876.0, + "step": 2772 + }, + { + "epoch": 0.30452448934768284, + "grad_norm": 2.125112295150757, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.7039007544517517, + "num_tokens": 70933661.0, + "step": 2773 + }, + { + "epoch": 0.3046343070502965, + "grad_norm": 2.7265048027038574, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7071146965026855, + "num_tokens": 70955775.0, + "step": 2774 + }, + { + "epoch": 0.3047441247529102, + "grad_norm": 2.0465097427368164, + "learning_rate": 1e-06, + "loss": 1.0471, + "mean_token_accuracy": 0.6831754446029663, + "num_tokens": 70986552.0, + "step": 2775 + }, + { + "epoch": 0.30485394245552383, + "grad_norm": 1.9601693153381348, + "learning_rate": 1e-06, + "loss": 1.0629, + "mean_token_accuracy": 0.6856894493103027, + "num_tokens": 71019291.0, + "step": 2776 + }, + { + "epoch": 0.3049637601581375, + "grad_norm": 2.686978816986084, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7265021800994873, + "num_tokens": 71038652.0, + "step": 2777 + }, + { + "epoch": 0.3050735778607512, + "grad_norm": 2.253901720046997, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7029929757118225, + "num_tokens": 71062900.0, + "step": 2778 + }, + { + "epoch": 0.3051833955633648, + "grad_norm": 1.9690994024276733, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7033965587615967, + "num_tokens": 71092418.0, + "step": 2779 + }, + { + "epoch": 0.30529321326597847, + "grad_norm": 1.92755925655365, + "learning_rate": 1e-06, + "loss": 1.0871, + "mean_token_accuracy": 0.6814607977867126, + "num_tokens": 71126384.0, + "step": 2780 + }, + { + "epoch": 0.3054030309685921, + "grad_norm": 2.1298270225524902, + "learning_rate": 1e-06, + "loss": 1.0529, + "mean_token_accuracy": 0.6809748411178589, + "num_tokens": 71154154.0, + "step": 2781 + }, + { + "epoch": 0.3055128486712058, + "grad_norm": 2.295090436935425, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6920581459999084, + "num_tokens": 71181153.0, + "step": 2782 + }, + { + "epoch": 0.30562266637381946, + "grad_norm": 2.39864182472229, + "learning_rate": 1e-06, + "loss": 1.0497, + "mean_token_accuracy": 0.6856892108917236, + "num_tokens": 71204853.0, + "step": 2783 + }, + { + "epoch": 0.3057324840764331, + "grad_norm": 2.0297462940216064, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7151907682418823, + "num_tokens": 71231306.0, + "step": 2784 + }, + { + "epoch": 0.30584230177904675, + "grad_norm": 2.0358731746673584, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7105709910392761, + "num_tokens": 71260254.0, + "step": 2785 + }, + { + "epoch": 0.30595211948166046, + "grad_norm": 1.9132882356643677, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.7044751644134521, + "num_tokens": 71295416.0, + "step": 2786 + }, + { + "epoch": 0.3060619371842741, + "grad_norm": 2.253669500350952, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6919466257095337, + "num_tokens": 71321024.0, + "step": 2787 + }, + { + "epoch": 0.30617175488688775, + "grad_norm": 2.1934878826141357, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7265141606330872, + "num_tokens": 71347080.0, + "step": 2788 + }, + { + "epoch": 0.30628157258950145, + "grad_norm": 2.3261542320251465, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6961207389831543, + "num_tokens": 71372513.0, + "step": 2789 + }, + { + "epoch": 0.3063913902921151, + "grad_norm": 2.2494006156921387, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7278406620025635, + "num_tokens": 71396118.0, + "step": 2790 + }, + { + "epoch": 0.30650120799472874, + "grad_norm": 2.468846559524536, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7281750440597534, + "num_tokens": 71417039.0, + "step": 2791 + }, + { + "epoch": 0.3066110256973424, + "grad_norm": 2.180495500564575, + "learning_rate": 1e-06, + "loss": 1.0217, + "mean_token_accuracy": 0.6973667740821838, + "num_tokens": 71442159.0, + "step": 2792 + }, + { + "epoch": 0.3067208433999561, + "grad_norm": 2.2212371826171875, + "learning_rate": 1e-06, + "loss": 1.0427, + "mean_token_accuracy": 0.6900134682655334, + "num_tokens": 71469028.0, + "step": 2793 + }, + { + "epoch": 0.30683066110256974, + "grad_norm": 2.4344770908355713, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7111905813217163, + "num_tokens": 71490029.0, + "step": 2794 + }, + { + "epoch": 0.3069404788051834, + "grad_norm": 2.4262120723724365, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7041275501251221, + "num_tokens": 71512574.0, + "step": 2795 + }, + { + "epoch": 0.3070502965077971, + "grad_norm": 2.3280727863311768, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.7014831900596619, + "num_tokens": 71537917.0, + "step": 2796 + }, + { + "epoch": 0.30716011421041073, + "grad_norm": 2.2655582427978516, + "learning_rate": 1e-06, + "loss": 1.0694, + "mean_token_accuracy": 0.6808855533599854, + "num_tokens": 71564820.0, + "step": 2797 + }, + { + "epoch": 0.3072699319130244, + "grad_norm": 2.210282564163208, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.7093492150306702, + "num_tokens": 71592655.0, + "step": 2798 + }, + { + "epoch": 0.307379749615638, + "grad_norm": 2.3444013595581055, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.7045708894729614, + "num_tokens": 71616757.0, + "step": 2799 + }, + { + "epoch": 0.3074895673182517, + "grad_norm": 2.177027463912964, + "learning_rate": 1e-06, + "loss": 1.0807, + "mean_token_accuracy": 0.6899767518043518, + "num_tokens": 71645413.0, + "step": 2800 + }, + { + "epoch": 0.30759938502086537, + "grad_norm": 1.9955004453659058, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6969026327133179, + "num_tokens": 71676698.0, + "step": 2801 + }, + { + "epoch": 0.307709202723479, + "grad_norm": 2.153543472290039, + "learning_rate": 1e-06, + "loss": 1.0668, + "mean_token_accuracy": 0.6776629686355591, + "num_tokens": 71704873.0, + "step": 2802 + }, + { + "epoch": 0.30781902042609266, + "grad_norm": 2.280853033065796, + "learning_rate": 1e-06, + "loss": 1.1367, + "mean_token_accuracy": 0.6662915945053101, + "num_tokens": 71731365.0, + "step": 2803 + }, + { + "epoch": 0.30792883812870636, + "grad_norm": 1.9899909496307373, + "learning_rate": 1e-06, + "loss": 1.0622, + "mean_token_accuracy": 0.6846244931221008, + "num_tokens": 71762362.0, + "step": 2804 + }, + { + "epoch": 0.30803865583132, + "grad_norm": 2.492661476135254, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7049205303192139, + "num_tokens": 71782620.0, + "step": 2805 + }, + { + "epoch": 0.30814847353393365, + "grad_norm": 2.035297155380249, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.7014736533164978, + "num_tokens": 71812514.0, + "step": 2806 + }, + { + "epoch": 0.30825829123654735, + "grad_norm": 2.3596203327178955, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.723454475402832, + "num_tokens": 71834621.0, + "step": 2807 + }, + { + "epoch": 0.308368108939161, + "grad_norm": 2.056018114089966, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7272682785987854, + "num_tokens": 71861510.0, + "step": 2808 + }, + { + "epoch": 0.30847792664177465, + "grad_norm": 2.3815441131591797, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7038096189498901, + "num_tokens": 71882793.0, + "step": 2809 + }, + { + "epoch": 0.3085877443443883, + "grad_norm": 2.230695962905884, + "learning_rate": 1e-06, + "loss": 1.0308, + "mean_token_accuracy": 0.6930520534515381, + "num_tokens": 71908211.0, + "step": 2810 + }, + { + "epoch": 0.308697562047002, + "grad_norm": 2.1518001556396484, + "learning_rate": 1e-06, + "loss": 1.0483, + "mean_token_accuracy": 0.688552975654602, + "num_tokens": 71935779.0, + "step": 2811 + }, + { + "epoch": 0.30880737974961564, + "grad_norm": 2.1353487968444824, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7133692502975464, + "num_tokens": 71963952.0, + "step": 2812 + }, + { + "epoch": 0.3089171974522293, + "grad_norm": 2.0883700847625732, + "learning_rate": 1e-06, + "loss": 1.0709, + "mean_token_accuracy": 0.676409900188446, + "num_tokens": 71995401.0, + "step": 2813 + }, + { + "epoch": 0.309027015154843, + "grad_norm": 1.9305918216705322, + "learning_rate": 1e-06, + "loss": 1.0925, + "mean_token_accuracy": 0.6739307045936584, + "num_tokens": 72030528.0, + "step": 2814 + }, + { + "epoch": 0.30913683285745663, + "grad_norm": 2.2438175678253174, + "learning_rate": 1e-06, + "loss": 1.0505, + "mean_token_accuracy": 0.6870137453079224, + "num_tokens": 72055948.0, + "step": 2815 + }, + { + "epoch": 0.3092466505600703, + "grad_norm": 2.4047203063964844, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.7046857476234436, + "num_tokens": 72081751.0, + "step": 2816 + }, + { + "epoch": 0.3093564682626839, + "grad_norm": 2.193112850189209, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.7012161016464233, + "num_tokens": 72109423.0, + "step": 2817 + }, + { + "epoch": 0.3094662859652976, + "grad_norm": 2.3989157676696777, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.698671281337738, + "num_tokens": 72131286.0, + "step": 2818 + }, + { + "epoch": 0.3095761036679113, + "grad_norm": 2.402348518371582, + "learning_rate": 1e-06, + "loss": 1.0774, + "mean_token_accuracy": 0.6813856959342957, + "num_tokens": 72155639.0, + "step": 2819 + }, + { + "epoch": 0.3096859213705249, + "grad_norm": 2.338754177093506, + "learning_rate": 1e-06, + "loss": 1.0029, + "mean_token_accuracy": 0.6937756538391113, + "num_tokens": 72179617.0, + "step": 2820 + }, + { + "epoch": 0.30979573907313857, + "grad_norm": 1.9762498140335083, + "learning_rate": 1e-06, + "loss": 1.0489, + "mean_token_accuracy": 0.6881823539733887, + "num_tokens": 72213281.0, + "step": 2821 + }, + { + "epoch": 0.30990555677575227, + "grad_norm": 2.2417590618133545, + "learning_rate": 1e-06, + "loss": 1.0635, + "mean_token_accuracy": 0.6768896579742432, + "num_tokens": 72239829.0, + "step": 2822 + }, + { + "epoch": 0.3100153744783659, + "grad_norm": 2.3312551975250244, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7145538330078125, + "num_tokens": 72262426.0, + "step": 2823 + }, + { + "epoch": 0.31012519218097956, + "grad_norm": 2.153252124786377, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.6974007487297058, + "num_tokens": 72288102.0, + "step": 2824 + }, + { + "epoch": 0.31023500988359326, + "grad_norm": 2.1671924591064453, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7061954140663147, + "num_tokens": 72316164.0, + "step": 2825 + }, + { + "epoch": 0.3103448275862069, + "grad_norm": 2.015537738800049, + "learning_rate": 1e-06, + "loss": 1.0782, + "mean_token_accuracy": 0.6734449863433838, + "num_tokens": 72348002.0, + "step": 2826 + }, + { + "epoch": 0.31045464528882055, + "grad_norm": 2.596045970916748, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7215123176574707, + "num_tokens": 72364741.0, + "step": 2827 + }, + { + "epoch": 0.3105644629914342, + "grad_norm": 2.185622453689575, + "learning_rate": 1e-06, + "loss": 1.06, + "mean_token_accuracy": 0.6844154596328735, + "num_tokens": 72392522.0, + "step": 2828 + }, + { + "epoch": 0.3106742806940479, + "grad_norm": 2.239426612854004, + "learning_rate": 1e-06, + "loss": 1.0429, + "mean_token_accuracy": 0.6893544793128967, + "num_tokens": 72417044.0, + "step": 2829 + }, + { + "epoch": 0.31078409839666155, + "grad_norm": 2.1026322841644287, + "learning_rate": 1e-06, + "loss": 1.0629, + "mean_token_accuracy": 0.6812787055969238, + "num_tokens": 72445901.0, + "step": 2830 + }, + { + "epoch": 0.3108939160992752, + "grad_norm": 2.8681561946868896, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7094141840934753, + "num_tokens": 72471000.0, + "step": 2831 + }, + { + "epoch": 0.31100373380188884, + "grad_norm": 2.2162392139434814, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.7077246904373169, + "num_tokens": 72496138.0, + "step": 2832 + }, + { + "epoch": 0.31111355150450254, + "grad_norm": 2.3844969272613525, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6967929005622864, + "num_tokens": 72521833.0, + "step": 2833 + }, + { + "epoch": 0.3112233692071162, + "grad_norm": 2.412278652191162, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.688315749168396, + "num_tokens": 72546645.0, + "step": 2834 + }, + { + "epoch": 0.31133318690972983, + "grad_norm": 2.278700590133667, + "learning_rate": 1e-06, + "loss": 1.0543, + "mean_token_accuracy": 0.6785383820533752, + "num_tokens": 72571474.0, + "step": 2835 + }, + { + "epoch": 0.31144300461234353, + "grad_norm": 2.150944471359253, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7235264778137207, + "num_tokens": 72598113.0, + "step": 2836 + }, + { + "epoch": 0.3115528223149572, + "grad_norm": 2.2702748775482178, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.6971497535705566, + "num_tokens": 72623567.0, + "step": 2837 + }, + { + "epoch": 0.3116626400175708, + "grad_norm": 2.2889461517333984, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6909672021865845, + "num_tokens": 72648561.0, + "step": 2838 + }, + { + "epoch": 0.31177245772018447, + "grad_norm": 2.2505860328674316, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6962955594062805, + "num_tokens": 72674250.0, + "step": 2839 + }, + { + "epoch": 0.31188227542279817, + "grad_norm": 2.324331521987915, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.6969953775405884, + "num_tokens": 72696487.0, + "step": 2840 + }, + { + "epoch": 0.3119920931254118, + "grad_norm": 2.2116825580596924, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.7047473192214966, + "num_tokens": 72721405.0, + "step": 2841 + }, + { + "epoch": 0.31210191082802546, + "grad_norm": 2.2282192707061768, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.6983352899551392, + "num_tokens": 72748049.0, + "step": 2842 + }, + { + "epoch": 0.31221172853063917, + "grad_norm": 1.8419842720031738, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.6947928667068481, + "num_tokens": 72782227.0, + "step": 2843 + }, + { + "epoch": 0.3123215462332528, + "grad_norm": 2.4039275646209717, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6992571949958801, + "num_tokens": 72805148.0, + "step": 2844 + }, + { + "epoch": 0.31243136393586646, + "grad_norm": 2.1797258853912354, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6937251091003418, + "num_tokens": 72833448.0, + "step": 2845 + }, + { + "epoch": 0.3125411816384801, + "grad_norm": 2.2313289642333984, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7122212648391724, + "num_tokens": 72857758.0, + "step": 2846 + }, + { + "epoch": 0.3126509993410938, + "grad_norm": 2.4236807823181152, + "learning_rate": 1e-06, + "loss": 1.025, + "mean_token_accuracy": 0.7019844055175781, + "num_tokens": 72878695.0, + "step": 2847 + }, + { + "epoch": 0.31276081704370745, + "grad_norm": 2.308748245239258, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7026515603065491, + "num_tokens": 72905965.0, + "step": 2848 + }, + { + "epoch": 0.3128706347463211, + "grad_norm": 2.2519304752349854, + "learning_rate": 1e-06, + "loss": 1.0562, + "mean_token_accuracy": 0.6853592991828918, + "num_tokens": 72932364.0, + "step": 2849 + }, + { + "epoch": 0.31298045244893474, + "grad_norm": 2.2954556941986084, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.6983800530433655, + "num_tokens": 72956649.0, + "step": 2850 + }, + { + "epoch": 0.31309027015154844, + "grad_norm": 2.2284047603607178, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7034544944763184, + "num_tokens": 72982120.0, + "step": 2851 + }, + { + "epoch": 0.3132000878541621, + "grad_norm": 1.9965462684631348, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.7020002007484436, + "num_tokens": 73010328.0, + "step": 2852 + }, + { + "epoch": 0.31330990555677574, + "grad_norm": 2.3350489139556885, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6852995157241821, + "num_tokens": 73034030.0, + "step": 2853 + }, + { + "epoch": 0.31341972325938944, + "grad_norm": 2.1423914432525635, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.6917693018913269, + "num_tokens": 73062862.0, + "step": 2854 + }, + { + "epoch": 0.3135295409620031, + "grad_norm": 2.041896343231201, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.716014564037323, + "num_tokens": 73092784.0, + "step": 2855 + }, + { + "epoch": 0.31363935866461673, + "grad_norm": 2.0860087871551514, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6998193860054016, + "num_tokens": 73123419.0, + "step": 2856 + }, + { + "epoch": 0.3137491763672304, + "grad_norm": 2.27884840965271, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6935670375823975, + "num_tokens": 73148098.0, + "step": 2857 + }, + { + "epoch": 0.3138589940698441, + "grad_norm": 1.8960187435150146, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7174453139305115, + "num_tokens": 73179630.0, + "step": 2858 + }, + { + "epoch": 0.3139688117724577, + "grad_norm": 2.337021589279175, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7274795174598694, + "num_tokens": 73204300.0, + "step": 2859 + }, + { + "epoch": 0.31407862947507137, + "grad_norm": 2.317490577697754, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7073744535446167, + "num_tokens": 73227187.0, + "step": 2860 + }, + { + "epoch": 0.314188447177685, + "grad_norm": 2.070984125137329, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6945163011550903, + "num_tokens": 73257678.0, + "step": 2861 + }, + { + "epoch": 0.3142982648802987, + "grad_norm": 2.563225507736206, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7050408124923706, + "num_tokens": 73280618.0, + "step": 2862 + }, + { + "epoch": 0.31440808258291236, + "grad_norm": 2.267101287841797, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6959448456764221, + "num_tokens": 73305692.0, + "step": 2863 + }, + { + "epoch": 0.314517900285526, + "grad_norm": 2.2971885204315186, + "learning_rate": 1e-06, + "loss": 1.032, + "mean_token_accuracy": 0.6875357627868652, + "num_tokens": 73330452.0, + "step": 2864 + }, + { + "epoch": 0.3146277179881397, + "grad_norm": 2.247187376022339, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7150274515151978, + "num_tokens": 73355494.0, + "step": 2865 + }, + { + "epoch": 0.31473753569075336, + "grad_norm": 2.2701501846313477, + "learning_rate": 1e-06, + "loss": 1.0371, + "mean_token_accuracy": 0.682557225227356, + "num_tokens": 73380518.0, + "step": 2866 + }, + { + "epoch": 0.314847353393367, + "grad_norm": 2.339737892150879, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7164047956466675, + "num_tokens": 73403846.0, + "step": 2867 + }, + { + "epoch": 0.31495717109598065, + "grad_norm": 2.22544264793396, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.6959092617034912, + "num_tokens": 73427150.0, + "step": 2868 + }, + { + "epoch": 0.31506698879859435, + "grad_norm": 2.581350564956665, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7029657363891602, + "num_tokens": 73448079.0, + "step": 2869 + }, + { + "epoch": 0.315176806501208, + "grad_norm": 2.1977970600128174, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6916331052780151, + "num_tokens": 73472017.0, + "step": 2870 + }, + { + "epoch": 0.31528662420382164, + "grad_norm": 1.8996225595474243, + "learning_rate": 1e-06, + "loss": 1.0841, + "mean_token_accuracy": 0.6784127354621887, + "num_tokens": 73506663.0, + "step": 2871 + }, + { + "epoch": 0.31539644190643534, + "grad_norm": 2.2416515350341797, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6888129115104675, + "num_tokens": 73532776.0, + "step": 2872 + }, + { + "epoch": 0.315506259609049, + "grad_norm": 2.1217281818389893, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7074622511863708, + "num_tokens": 73558806.0, + "step": 2873 + }, + { + "epoch": 0.31561607731166264, + "grad_norm": 2.096047878265381, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7173488140106201, + "num_tokens": 73586664.0, + "step": 2874 + }, + { + "epoch": 0.3157258950142763, + "grad_norm": 2.354849100112915, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7058566808700562, + "num_tokens": 73610525.0, + "step": 2875 + }, + { + "epoch": 0.31583571271689, + "grad_norm": 2.1613810062408447, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6953511834144592, + "num_tokens": 73636445.0, + "step": 2876 + }, + { + "epoch": 0.31594553041950363, + "grad_norm": 2.3465259075164795, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6957421898841858, + "num_tokens": 73660304.0, + "step": 2877 + }, + { + "epoch": 0.3160553481221173, + "grad_norm": 2.160538911819458, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6902569532394409, + "num_tokens": 73688305.0, + "step": 2878 + }, + { + "epoch": 0.3161651658247309, + "grad_norm": 2.362330675125122, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7228678464889526, + "num_tokens": 73711535.0, + "step": 2879 + }, + { + "epoch": 0.3162749835273446, + "grad_norm": 2.512916088104248, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7288817763328552, + "num_tokens": 73733439.0, + "step": 2880 + }, + { + "epoch": 0.31638480122995827, + "grad_norm": 2.2524333000183105, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7175567150115967, + "num_tokens": 73758571.0, + "step": 2881 + }, + { + "epoch": 0.3164946189325719, + "grad_norm": 2.1130008697509766, + "learning_rate": 1e-06, + "loss": 1.0738, + "mean_token_accuracy": 0.6758434772491455, + "num_tokens": 73788312.0, + "step": 2882 + }, + { + "epoch": 0.3166044366351856, + "grad_norm": 2.3309621810913086, + "learning_rate": 1e-06, + "loss": 1.0843, + "mean_token_accuracy": 0.6741666793823242, + "num_tokens": 73812836.0, + "step": 2883 + }, + { + "epoch": 0.31671425433779926, + "grad_norm": 2.307866096496582, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7174903154373169, + "num_tokens": 73836452.0, + "step": 2884 + }, + { + "epoch": 0.3168240720404129, + "grad_norm": 2.319272041320801, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7216283082962036, + "num_tokens": 73858393.0, + "step": 2885 + }, + { + "epoch": 0.31693388974302655, + "grad_norm": 2.0420987606048584, + "learning_rate": 1e-06, + "loss": 1.0235, + "mean_token_accuracy": 0.6929274797439575, + "num_tokens": 73889748.0, + "step": 2886 + }, + { + "epoch": 0.31704370744564025, + "grad_norm": 2.2555761337280273, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.7011076807975769, + "num_tokens": 73915217.0, + "step": 2887 + }, + { + "epoch": 0.3171535251482539, + "grad_norm": 2.2854654788970947, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7087159156799316, + "num_tokens": 73940580.0, + "step": 2888 + }, + { + "epoch": 0.31726334285086755, + "grad_norm": 2.298689603805542, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.6904944181442261, + "num_tokens": 73964267.0, + "step": 2889 + }, + { + "epoch": 0.31737316055348125, + "grad_norm": 2.024265766143799, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.694308876991272, + "num_tokens": 73995320.0, + "step": 2890 + }, + { + "epoch": 0.3174829782560949, + "grad_norm": 2.120370626449585, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.7007547616958618, + "num_tokens": 74021607.0, + "step": 2891 + }, + { + "epoch": 0.31759279595870854, + "grad_norm": 2.634827136993408, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.7061429023742676, + "num_tokens": 74041908.0, + "step": 2892 + }, + { + "epoch": 0.3177026136613222, + "grad_norm": 2.2666966915130615, + "learning_rate": 1e-06, + "loss": 1.0316, + "mean_token_accuracy": 0.6845477819442749, + "num_tokens": 74070318.0, + "step": 2893 + }, + { + "epoch": 0.3178124313639359, + "grad_norm": 2.1112611293792725, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.7069555521011353, + "num_tokens": 74096455.0, + "step": 2894 + }, + { + "epoch": 0.31792224906654953, + "grad_norm": 2.411700963973999, + "learning_rate": 1e-06, + "loss": 1.053, + "mean_token_accuracy": 0.6860358715057373, + "num_tokens": 74122695.0, + "step": 2895 + }, + { + "epoch": 0.3180320667691632, + "grad_norm": 2.138392925262451, + "learning_rate": 1e-06, + "loss": 1.115, + "mean_token_accuracy": 0.6781230568885803, + "num_tokens": 74152555.0, + "step": 2896 + }, + { + "epoch": 0.3181418844717768, + "grad_norm": 2.0897114276885986, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6845060586929321, + "num_tokens": 74180500.0, + "step": 2897 + }, + { + "epoch": 0.3182517021743905, + "grad_norm": 2.148660898208618, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.7004010677337646, + "num_tokens": 74205868.0, + "step": 2898 + }, + { + "epoch": 0.3183615198770042, + "grad_norm": 2.1773414611816406, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7166944742202759, + "num_tokens": 74231565.0, + "step": 2899 + }, + { + "epoch": 0.3184713375796178, + "grad_norm": 2.3878679275512695, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.7047189474105835, + "num_tokens": 74254590.0, + "step": 2900 + }, + { + "epoch": 0.3185811552822315, + "grad_norm": 2.226653814315796, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7069982290267944, + "num_tokens": 74281674.0, + "step": 2901 + }, + { + "epoch": 0.31869097298484517, + "grad_norm": 2.2030835151672363, + "learning_rate": 1e-06, + "loss": 1.0357, + "mean_token_accuracy": 0.686184287071228, + "num_tokens": 74308768.0, + "step": 2902 + }, + { + "epoch": 0.3188007906874588, + "grad_norm": 2.230942726135254, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.6941601037979126, + "num_tokens": 74335347.0, + "step": 2903 + }, + { + "epoch": 0.31891060839007246, + "grad_norm": 2.915403127670288, + "learning_rate": 1e-06, + "loss": 1.0325, + "mean_token_accuracy": 0.6865118145942688, + "num_tokens": 74353379.0, + "step": 2904 + }, + { + "epoch": 0.31902042609268616, + "grad_norm": 2.4478297233581543, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7005363702774048, + "num_tokens": 74376473.0, + "step": 2905 + }, + { + "epoch": 0.3191302437952998, + "grad_norm": 2.0232269763946533, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.6969640254974365, + "num_tokens": 74404748.0, + "step": 2906 + }, + { + "epoch": 0.31924006149791345, + "grad_norm": 2.211486577987671, + "learning_rate": 1e-06, + "loss": 1.1091, + "mean_token_accuracy": 0.664374828338623, + "num_tokens": 74433512.0, + "step": 2907 + }, + { + "epoch": 0.3193498792005271, + "grad_norm": 2.3234851360321045, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7215579748153687, + "num_tokens": 74457554.0, + "step": 2908 + }, + { + "epoch": 0.3194596969031408, + "grad_norm": 2.480813503265381, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7034989595413208, + "num_tokens": 74479352.0, + "step": 2909 + }, + { + "epoch": 0.31956951460575445, + "grad_norm": 2.3807272911071777, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6985341310501099, + "num_tokens": 74502834.0, + "step": 2910 + }, + { + "epoch": 0.3196793323083681, + "grad_norm": 2.2221271991729736, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7136598229408264, + "num_tokens": 74525965.0, + "step": 2911 + }, + { + "epoch": 0.3197891500109818, + "grad_norm": 2.3826231956481934, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.707202136516571, + "num_tokens": 74549038.0, + "step": 2912 + }, + { + "epoch": 0.31989896771359544, + "grad_norm": 2.270415782928467, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7070530652999878, + "num_tokens": 74573983.0, + "step": 2913 + }, + { + "epoch": 0.3200087854162091, + "grad_norm": 2.2397007942199707, + "learning_rate": 1e-06, + "loss": 1.0029, + "mean_token_accuracy": 0.699209988117218, + "num_tokens": 74602156.0, + "step": 2914 + }, + { + "epoch": 0.32011860311882273, + "grad_norm": 2.0249786376953125, + "learning_rate": 1e-06, + "loss": 1.0929, + "mean_token_accuracy": 0.6730824708938599, + "num_tokens": 74631953.0, + "step": 2915 + }, + { + "epoch": 0.32022842082143643, + "grad_norm": 2.538273572921753, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7174757719039917, + "num_tokens": 74652520.0, + "step": 2916 + }, + { + "epoch": 0.3203382385240501, + "grad_norm": 2.3407819271087646, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7071832418441772, + "num_tokens": 74676226.0, + "step": 2917 + }, + { + "epoch": 0.3204480562266637, + "grad_norm": 2.151099681854248, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7038167715072632, + "num_tokens": 74703782.0, + "step": 2918 + }, + { + "epoch": 0.3205578739292774, + "grad_norm": 2.037179470062256, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7221076488494873, + "num_tokens": 74730397.0, + "step": 2919 + }, + { + "epoch": 0.32066769163189107, + "grad_norm": 2.16180419921875, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6901655793190002, + "num_tokens": 74756140.0, + "step": 2920 + }, + { + "epoch": 0.3207775093345047, + "grad_norm": 2.474613904953003, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.6996559500694275, + "num_tokens": 74777201.0, + "step": 2921 + }, + { + "epoch": 0.32088732703711836, + "grad_norm": 1.9716721773147583, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.699993371963501, + "num_tokens": 74806934.0, + "step": 2922 + }, + { + "epoch": 0.32099714473973207, + "grad_norm": 1.9795751571655273, + "learning_rate": 1e-06, + "loss": 1.0521, + "mean_token_accuracy": 0.6949832439422607, + "num_tokens": 74838614.0, + "step": 2923 + }, + { + "epoch": 0.3211069624423457, + "grad_norm": 2.198859691619873, + "learning_rate": 1e-06, + "loss": 1.0498, + "mean_token_accuracy": 0.6864608526229858, + "num_tokens": 74865504.0, + "step": 2924 + }, + { + "epoch": 0.32121678014495936, + "grad_norm": 2.069974660873413, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.7048088908195496, + "num_tokens": 74894749.0, + "step": 2925 + }, + { + "epoch": 0.321326597847573, + "grad_norm": 2.3299238681793213, + "learning_rate": 1e-06, + "loss": 1.1257, + "mean_token_accuracy": 0.6674047708511353, + "num_tokens": 74922100.0, + "step": 2926 + }, + { + "epoch": 0.3214364155501867, + "grad_norm": 2.4949328899383545, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7090831398963928, + "num_tokens": 74942020.0, + "step": 2927 + }, + { + "epoch": 0.32154623325280035, + "grad_norm": 2.1142311096191406, + "learning_rate": 1e-06, + "loss": 1.07, + "mean_token_accuracy": 0.680640459060669, + "num_tokens": 74973028.0, + "step": 2928 + }, + { + "epoch": 0.321656050955414, + "grad_norm": 2.409803867340088, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7250481843948364, + "num_tokens": 74996375.0, + "step": 2929 + }, + { + "epoch": 0.3217658686580277, + "grad_norm": 2.3644652366638184, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.7054066061973572, + "num_tokens": 75020534.0, + "step": 2930 + }, + { + "epoch": 0.32187568636064134, + "grad_norm": 2.3105409145355225, + "learning_rate": 1e-06, + "loss": 1.0648, + "mean_token_accuracy": 0.6889679431915283, + "num_tokens": 75045298.0, + "step": 2931 + }, + { + "epoch": 0.321985504063255, + "grad_norm": 2.44992733001709, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.6982918977737427, + "num_tokens": 75065885.0, + "step": 2932 + }, + { + "epoch": 0.32209532176586864, + "grad_norm": 2.5838170051574707, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7060070037841797, + "num_tokens": 75085096.0, + "step": 2933 + }, + { + "epoch": 0.32220513946848234, + "grad_norm": 2.2720673084259033, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7088425159454346, + "num_tokens": 75108675.0, + "step": 2934 + }, + { + "epoch": 0.322314957171096, + "grad_norm": 2.4494924545288086, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7153849601745605, + "num_tokens": 75130113.0, + "step": 2935 + }, + { + "epoch": 0.32242477487370963, + "grad_norm": 2.2442333698272705, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6996278166770935, + "num_tokens": 75157984.0, + "step": 2936 + }, + { + "epoch": 0.3225345925763233, + "grad_norm": 2.0096261501312256, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.6913567185401917, + "num_tokens": 75191347.0, + "step": 2937 + }, + { + "epoch": 0.322644410278937, + "grad_norm": 2.146890640258789, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6975740790367126, + "num_tokens": 75218006.0, + "step": 2938 + }, + { + "epoch": 0.3227542279815506, + "grad_norm": 2.204333543777466, + "learning_rate": 1e-06, + "loss": 1.0611, + "mean_token_accuracy": 0.6828559637069702, + "num_tokens": 75247727.0, + "step": 2939 + }, + { + "epoch": 0.32286404568416427, + "grad_norm": 2.22816801071167, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7196767330169678, + "num_tokens": 75271972.0, + "step": 2940 + }, + { + "epoch": 0.32297386338677797, + "grad_norm": 2.383671283721924, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7125613689422607, + "num_tokens": 75294796.0, + "step": 2941 + }, + { + "epoch": 0.3230836810893916, + "grad_norm": 2.135284662246704, + "learning_rate": 1e-06, + "loss": 1.0501, + "mean_token_accuracy": 0.6855567693710327, + "num_tokens": 75322626.0, + "step": 2942 + }, + { + "epoch": 0.32319349879200526, + "grad_norm": 2.1371567249298096, + "learning_rate": 1e-06, + "loss": 1.1234, + "mean_token_accuracy": 0.6646714210510254, + "num_tokens": 75353759.0, + "step": 2943 + }, + { + "epoch": 0.3233033164946189, + "grad_norm": 2.4963033199310303, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7084991335868835, + "num_tokens": 75375439.0, + "step": 2944 + }, + { + "epoch": 0.3234131341972326, + "grad_norm": 2.5173499584198, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.6960521936416626, + "num_tokens": 75397069.0, + "step": 2945 + }, + { + "epoch": 0.32352295189984626, + "grad_norm": 2.1122078895568848, + "learning_rate": 1e-06, + "loss": 1.0397, + "mean_token_accuracy": 0.6923180818557739, + "num_tokens": 75422744.0, + "step": 2946 + }, + { + "epoch": 0.3236327696024599, + "grad_norm": 2.089015483856201, + "learning_rate": 1e-06, + "loss": 1.0625, + "mean_token_accuracy": 0.6827645301818848, + "num_tokens": 75453250.0, + "step": 2947 + }, + { + "epoch": 0.3237425873050736, + "grad_norm": 2.6166739463806152, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.7044202089309692, + "num_tokens": 75473140.0, + "step": 2948 + }, + { + "epoch": 0.32385240500768725, + "grad_norm": 2.4765868186950684, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.7039179801940918, + "num_tokens": 75495319.0, + "step": 2949 + }, + { + "epoch": 0.3239622227103009, + "grad_norm": 2.2764780521392822, + "learning_rate": 1e-06, + "loss": 1.0301, + "mean_token_accuracy": 0.6915906667709351, + "num_tokens": 75520238.0, + "step": 2950 + }, + { + "epoch": 0.32407204041291454, + "grad_norm": 2.5315279960632324, + "learning_rate": 1e-06, + "loss": 1.0316, + "mean_token_accuracy": 0.6858729124069214, + "num_tokens": 75541053.0, + "step": 2951 + }, + { + "epoch": 0.32418185811552824, + "grad_norm": 2.5744097232818604, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.6954903602600098, + "num_tokens": 75561537.0, + "step": 2952 + }, + { + "epoch": 0.3242916758181419, + "grad_norm": 2.1177754402160645, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7065255045890808, + "num_tokens": 75591285.0, + "step": 2953 + }, + { + "epoch": 0.32440149352075554, + "grad_norm": 2.2252912521362305, + "learning_rate": 1e-06, + "loss": 1.0427, + "mean_token_accuracy": 0.6807703971862793, + "num_tokens": 75618690.0, + "step": 2954 + }, + { + "epoch": 0.3245113112233692, + "grad_norm": 2.11972975730896, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7057889699935913, + "num_tokens": 75648387.0, + "step": 2955 + }, + { + "epoch": 0.3246211289259829, + "grad_norm": 2.178032398223877, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7094207406044006, + "num_tokens": 75675594.0, + "step": 2956 + }, + { + "epoch": 0.32473094662859653, + "grad_norm": 1.9453984498977661, + "learning_rate": 1e-06, + "loss": 1.0673, + "mean_token_accuracy": 0.6889500617980957, + "num_tokens": 75705257.0, + "step": 2957 + }, + { + "epoch": 0.3248407643312102, + "grad_norm": 2.1121251583099365, + "learning_rate": 1e-06, + "loss": 1.0798, + "mean_token_accuracy": 0.67750483751297, + "num_tokens": 75734521.0, + "step": 2958 + }, + { + "epoch": 0.3249505820338239, + "grad_norm": 2.5391547679901123, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7040608525276184, + "num_tokens": 75754363.0, + "step": 2959 + }, + { + "epoch": 0.3250603997364375, + "grad_norm": 2.2709429264068604, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.7064452171325684, + "num_tokens": 75778195.0, + "step": 2960 + }, + { + "epoch": 0.32517021743905117, + "grad_norm": 2.5950326919555664, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7019426822662354, + "num_tokens": 75798144.0, + "step": 2961 + }, + { + "epoch": 0.3252800351416648, + "grad_norm": 1.9983881711959839, + "learning_rate": 1e-06, + "loss": 1.0247, + "mean_token_accuracy": 0.696122407913208, + "num_tokens": 75827819.0, + "step": 2962 + }, + { + "epoch": 0.3253898528442785, + "grad_norm": 2.1148526668548584, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.6986258625984192, + "num_tokens": 75854882.0, + "step": 2963 + }, + { + "epoch": 0.32549967054689216, + "grad_norm": 2.085268497467041, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6880648136138916, + "num_tokens": 75883143.0, + "step": 2964 + }, + { + "epoch": 0.3256094882495058, + "grad_norm": 2.1968867778778076, + "learning_rate": 1e-06, + "loss": 1.0526, + "mean_token_accuracy": 0.6802915334701538, + "num_tokens": 75911485.0, + "step": 2965 + }, + { + "epoch": 0.3257193059521195, + "grad_norm": 2.020907402038574, + "learning_rate": 1e-06, + "loss": 1.0705, + "mean_token_accuracy": 0.6862974166870117, + "num_tokens": 75942639.0, + "step": 2966 + }, + { + "epoch": 0.32582912365473315, + "grad_norm": 2.1785366535186768, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7313238382339478, + "num_tokens": 75967659.0, + "step": 2967 + }, + { + "epoch": 0.3259389413573468, + "grad_norm": 2.1972806453704834, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7399151921272278, + "num_tokens": 75992899.0, + "step": 2968 + }, + { + "epoch": 0.32604875905996045, + "grad_norm": 2.254488706588745, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7163939476013184, + "num_tokens": 76017579.0, + "step": 2969 + }, + { + "epoch": 0.32615857676257415, + "grad_norm": 2.135106325149536, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.7100385427474976, + "num_tokens": 76045425.0, + "step": 2970 + }, + { + "epoch": 0.3262683944651878, + "grad_norm": 2.2191851139068604, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7244164943695068, + "num_tokens": 76068787.0, + "step": 2971 + }, + { + "epoch": 0.32637821216780144, + "grad_norm": 2.4813730716705322, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6997570395469666, + "num_tokens": 76089881.0, + "step": 2972 + }, + { + "epoch": 0.3264880298704151, + "grad_norm": 2.1435141563415527, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7091584205627441, + "num_tokens": 76116252.0, + "step": 2973 + }, + { + "epoch": 0.3265978475730288, + "grad_norm": 2.3204736709594727, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.6959209442138672, + "num_tokens": 76138064.0, + "step": 2974 + }, + { + "epoch": 0.32670766527564243, + "grad_norm": 2.1182072162628174, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.6964098215103149, + "num_tokens": 76165082.0, + "step": 2975 + }, + { + "epoch": 0.3268174829782561, + "grad_norm": 2.736834764480591, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6909323930740356, + "num_tokens": 76185314.0, + "step": 2976 + }, + { + "epoch": 0.3269273006808698, + "grad_norm": 2.4699699878692627, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.6921745538711548, + "num_tokens": 76208204.0, + "step": 2977 + }, + { + "epoch": 0.3270371183834834, + "grad_norm": 2.282134771347046, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.7027179598808289, + "num_tokens": 76234943.0, + "step": 2978 + }, + { + "epoch": 0.3271469360860971, + "grad_norm": 2.419128894805908, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6963592171669006, + "num_tokens": 76255759.0, + "step": 2979 + }, + { + "epoch": 0.3272567537887107, + "grad_norm": 2.402451992034912, + "learning_rate": 1e-06, + "loss": 1.0272, + "mean_token_accuracy": 0.6887059211730957, + "num_tokens": 76278555.0, + "step": 2980 + }, + { + "epoch": 0.3273665714913244, + "grad_norm": 2.1698081493377686, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7002424001693726, + "num_tokens": 76303905.0, + "step": 2981 + }, + { + "epoch": 0.32747638919393807, + "grad_norm": 2.176572322845459, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.7019565105438232, + "num_tokens": 76330346.0, + "step": 2982 + }, + { + "epoch": 0.3275862068965517, + "grad_norm": 1.9887555837631226, + "learning_rate": 1e-06, + "loss": 1.0607, + "mean_token_accuracy": 0.6797968745231628, + "num_tokens": 76363153.0, + "step": 2983 + }, + { + "epoch": 0.32769602459916536, + "grad_norm": 2.2900617122650146, + "learning_rate": 1e-06, + "loss": 1.0376, + "mean_token_accuracy": 0.6860732436180115, + "num_tokens": 76387069.0, + "step": 2984 + }, + { + "epoch": 0.32780584230177906, + "grad_norm": 2.2269840240478516, + "learning_rate": 1e-06, + "loss": 1.0629, + "mean_token_accuracy": 0.6845624446868896, + "num_tokens": 76411821.0, + "step": 2985 + }, + { + "epoch": 0.3279156600043927, + "grad_norm": 2.213057518005371, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7057111263275146, + "num_tokens": 76435827.0, + "step": 2986 + }, + { + "epoch": 0.32802547770700635, + "grad_norm": 1.9615380764007568, + "learning_rate": 1e-06, + "loss": 1.0588, + "mean_token_accuracy": 0.6840654015541077, + "num_tokens": 76468904.0, + "step": 2987 + }, + { + "epoch": 0.32813529540962005, + "grad_norm": 2.208221673965454, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.700387179851532, + "num_tokens": 76493372.0, + "step": 2988 + }, + { + "epoch": 0.3282451131122337, + "grad_norm": 2.3964056968688965, + "learning_rate": 1e-06, + "loss": 1.038, + "mean_token_accuracy": 0.6878856420516968, + "num_tokens": 76516452.0, + "step": 2989 + }, + { + "epoch": 0.32835493081484735, + "grad_norm": 2.2721900939941406, + "learning_rate": 1e-06, + "loss": 1.0562, + "mean_token_accuracy": 0.6870713233947754, + "num_tokens": 76540993.0, + "step": 2990 + }, + { + "epoch": 0.328464748517461, + "grad_norm": 2.2822704315185547, + "learning_rate": 1e-06, + "loss": 1.0851, + "mean_token_accuracy": 0.6774150133132935, + "num_tokens": 76564957.0, + "step": 2991 + }, + { + "epoch": 0.3285745662200747, + "grad_norm": 2.2196779251098633, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.706173300743103, + "num_tokens": 76589454.0, + "step": 2992 + }, + { + "epoch": 0.32868438392268834, + "grad_norm": 2.316330671310425, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7275235652923584, + "num_tokens": 76610364.0, + "step": 2993 + }, + { + "epoch": 0.328794201625302, + "grad_norm": 2.248666763305664, + "learning_rate": 1e-06, + "loss": 1.0792, + "mean_token_accuracy": 0.6793043613433838, + "num_tokens": 76638075.0, + "step": 2994 + }, + { + "epoch": 0.3289040193279157, + "grad_norm": 2.221764326095581, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7071290612220764, + "num_tokens": 76663489.0, + "step": 2995 + }, + { + "epoch": 0.32901383703052933, + "grad_norm": 2.1367435455322266, + "learning_rate": 1e-06, + "loss": 1.012, + "mean_token_accuracy": 0.6914585828781128, + "num_tokens": 76690824.0, + "step": 2996 + }, + { + "epoch": 0.329123654733143, + "grad_norm": 2.1491777896881104, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7021563649177551, + "num_tokens": 76717641.0, + "step": 2997 + }, + { + "epoch": 0.3292334724357566, + "grad_norm": 2.308641195297241, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6890177130699158, + "num_tokens": 76740915.0, + "step": 2998 + }, + { + "epoch": 0.3293432901383703, + "grad_norm": 2.570802688598633, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6934628486633301, + "num_tokens": 76762882.0, + "step": 2999 + }, + { + "epoch": 0.32945310784098397, + "grad_norm": 2.7449069023132324, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7092230319976807, + "num_tokens": 76781628.0, + "step": 3000 + }, + { + "epoch": 0.3295629255435976, + "grad_norm": 2.2835540771484375, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7068403959274292, + "num_tokens": 76805876.0, + "step": 3001 + }, + { + "epoch": 0.32967274324621126, + "grad_norm": 2.122249126434326, + "learning_rate": 1e-06, + "loss": 1.0345, + "mean_token_accuracy": 0.6875838041305542, + "num_tokens": 76835706.0, + "step": 3002 + }, + { + "epoch": 0.32978256094882497, + "grad_norm": 2.2840769290924072, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7141232490539551, + "num_tokens": 76861677.0, + "step": 3003 + }, + { + "epoch": 0.3298923786514386, + "grad_norm": 2.2789196968078613, + "learning_rate": 1e-06, + "loss": 1.0351, + "mean_token_accuracy": 0.6880353093147278, + "num_tokens": 76886193.0, + "step": 3004 + }, + { + "epoch": 0.33000219635405226, + "grad_norm": 2.079664945602417, + "learning_rate": 1e-06, + "loss": 1.0614, + "mean_token_accuracy": 0.6964491605758667, + "num_tokens": 76913825.0, + "step": 3005 + }, + { + "epoch": 0.33011201405666596, + "grad_norm": 2.124851703643799, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.6938729882240295, + "num_tokens": 76942587.0, + "step": 3006 + }, + { + "epoch": 0.3302218317592796, + "grad_norm": 2.17277193069458, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6893502473831177, + "num_tokens": 76968073.0, + "step": 3007 + }, + { + "epoch": 0.33033164946189325, + "grad_norm": 2.2180795669555664, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.7107406854629517, + "num_tokens": 76995246.0, + "step": 3008 + }, + { + "epoch": 0.3304414671645069, + "grad_norm": 2.4389758110046387, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7035679817199707, + "num_tokens": 77017353.0, + "step": 3009 + }, + { + "epoch": 0.3305512848671206, + "grad_norm": 2.0668458938598633, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6966516375541687, + "num_tokens": 77045974.0, + "step": 3010 + }, + { + "epoch": 0.33066110256973424, + "grad_norm": 2.153402805328369, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.6931779980659485, + "num_tokens": 77073006.0, + "step": 3011 + }, + { + "epoch": 0.3307709202723479, + "grad_norm": 2.218611717224121, + "learning_rate": 1e-06, + "loss": 1.0167, + "mean_token_accuracy": 0.6959884166717529, + "num_tokens": 77100179.0, + "step": 3012 + }, + { + "epoch": 0.33088073797496154, + "grad_norm": 2.46979022026062, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6929450035095215, + "num_tokens": 77120172.0, + "step": 3013 + }, + { + "epoch": 0.33099055567757524, + "grad_norm": 2.4670846462249756, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.7039101123809814, + "num_tokens": 77141821.0, + "step": 3014 + }, + { + "epoch": 0.3311003733801889, + "grad_norm": 2.1580801010131836, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7190753817558289, + "num_tokens": 77167543.0, + "step": 3015 + }, + { + "epoch": 0.33121019108280253, + "grad_norm": 2.358938455581665, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6965892314910889, + "num_tokens": 77193832.0, + "step": 3016 + }, + { + "epoch": 0.33132000878541623, + "grad_norm": 2.543354034423828, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7235453128814697, + "num_tokens": 77216449.0, + "step": 3017 + }, + { + "epoch": 0.3314298264880299, + "grad_norm": 2.226787567138672, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.7012989521026611, + "num_tokens": 77242603.0, + "step": 3018 + }, + { + "epoch": 0.3315396441906435, + "grad_norm": 2.158191442489624, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7115314602851868, + "num_tokens": 77270288.0, + "step": 3019 + }, + { + "epoch": 0.33164946189325717, + "grad_norm": 2.199510335922241, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6943084597587585, + "num_tokens": 77298129.0, + "step": 3020 + }, + { + "epoch": 0.33175927959587087, + "grad_norm": 2.1334614753723145, + "learning_rate": 1e-06, + "loss": 1.0238, + "mean_token_accuracy": 0.6923489570617676, + "num_tokens": 77325740.0, + "step": 3021 + }, + { + "epoch": 0.3318690972984845, + "grad_norm": 2.245142936706543, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.7121134996414185, + "num_tokens": 77350905.0, + "step": 3022 + }, + { + "epoch": 0.33197891500109816, + "grad_norm": 2.357659101486206, + "learning_rate": 1e-06, + "loss": 1.0305, + "mean_token_accuracy": 0.6878758668899536, + "num_tokens": 77376334.0, + "step": 3023 + }, + { + "epoch": 0.33208873270371186, + "grad_norm": 2.3694941997528076, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7032468914985657, + "num_tokens": 77401859.0, + "step": 3024 + }, + { + "epoch": 0.3321985504063255, + "grad_norm": 2.1781115531921387, + "learning_rate": 1e-06, + "loss": 1.0568, + "mean_token_accuracy": 0.6845402717590332, + "num_tokens": 77430202.0, + "step": 3025 + }, + { + "epoch": 0.33230836810893916, + "grad_norm": 2.039682149887085, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7022798657417297, + "num_tokens": 77457247.0, + "step": 3026 + }, + { + "epoch": 0.3324181858115528, + "grad_norm": 2.2315640449523926, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.6985331177711487, + "num_tokens": 77482734.0, + "step": 3027 + }, + { + "epoch": 0.3325280035141665, + "grad_norm": 2.173706293106079, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7096378207206726, + "num_tokens": 77509465.0, + "step": 3028 + }, + { + "epoch": 0.33263782121678015, + "grad_norm": 1.8888922929763794, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.7019757032394409, + "num_tokens": 77543399.0, + "step": 3029 + }, + { + "epoch": 0.3327476389193938, + "grad_norm": 2.17742657661438, + "learning_rate": 1e-06, + "loss": 1.0531, + "mean_token_accuracy": 0.688247799873352, + "num_tokens": 77570749.0, + "step": 3030 + }, + { + "epoch": 0.33285745662200744, + "grad_norm": 2.321479320526123, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.6996420621871948, + "num_tokens": 77594231.0, + "step": 3031 + }, + { + "epoch": 0.33296727432462114, + "grad_norm": 2.2059993743896484, + "learning_rate": 1e-06, + "loss": 1.0459, + "mean_token_accuracy": 0.6832085847854614, + "num_tokens": 77621769.0, + "step": 3032 + }, + { + "epoch": 0.3330770920272348, + "grad_norm": 2.051020860671997, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.7012797594070435, + "num_tokens": 77652018.0, + "step": 3033 + }, + { + "epoch": 0.33318690972984844, + "grad_norm": 2.251709461212158, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6983374357223511, + "num_tokens": 77677221.0, + "step": 3034 + }, + { + "epoch": 0.33329672743246214, + "grad_norm": 2.0965583324432373, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6941372752189636, + "num_tokens": 77705397.0, + "step": 3035 + }, + { + "epoch": 0.3334065451350758, + "grad_norm": 2.099076986312866, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7101715803146362, + "num_tokens": 77733707.0, + "step": 3036 + }, + { + "epoch": 0.33351636283768943, + "grad_norm": 2.268681526184082, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7053951025009155, + "num_tokens": 77758414.0, + "step": 3037 + }, + { + "epoch": 0.3336261805403031, + "grad_norm": 2.4452850818634033, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.700494647026062, + "num_tokens": 77781440.0, + "step": 3038 + }, + { + "epoch": 0.3337359982429168, + "grad_norm": 2.4406635761260986, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6924887299537659, + "num_tokens": 77804078.0, + "step": 3039 + }, + { + "epoch": 0.3338458159455304, + "grad_norm": 2.1495840549468994, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7013109922409058, + "num_tokens": 77834272.0, + "step": 3040 + }, + { + "epoch": 0.33395563364814407, + "grad_norm": 1.9760929346084595, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.6999438405036926, + "num_tokens": 77866139.0, + "step": 3041 + }, + { + "epoch": 0.33406545135075777, + "grad_norm": 2.0637102127075195, + "learning_rate": 1e-06, + "loss": 1.0826, + "mean_token_accuracy": 0.6742896437644958, + "num_tokens": 77895134.0, + "step": 3042 + }, + { + "epoch": 0.3341752690533714, + "grad_norm": 2.3210608959198, + "learning_rate": 1e-06, + "loss": 1.061, + "mean_token_accuracy": 0.6851823925971985, + "num_tokens": 77922110.0, + "step": 3043 + }, + { + "epoch": 0.33428508675598506, + "grad_norm": 2.0068318843841553, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6886602640151978, + "num_tokens": 77954036.0, + "step": 3044 + }, + { + "epoch": 0.3343949044585987, + "grad_norm": 2.352825880050659, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6972362995147705, + "num_tokens": 77977377.0, + "step": 3045 + }, + { + "epoch": 0.3345047221612124, + "grad_norm": 2.7037241458892822, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.7011721134185791, + "num_tokens": 77996451.0, + "step": 3046 + }, + { + "epoch": 0.33461453986382605, + "grad_norm": 2.288641929626465, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.692683756351471, + "num_tokens": 78025580.0, + "step": 3047 + }, + { + "epoch": 0.3347243575664397, + "grad_norm": 2.407041311264038, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7253854274749756, + "num_tokens": 78047909.0, + "step": 3048 + }, + { + "epoch": 0.33483417526905335, + "grad_norm": 2.223397970199585, + "learning_rate": 1e-06, + "loss": 1.0531, + "mean_token_accuracy": 0.6776180863380432, + "num_tokens": 78072893.0, + "step": 3049 + }, + { + "epoch": 0.33494399297166705, + "grad_norm": 1.9829518795013428, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.6895252466201782, + "num_tokens": 78104393.0, + "step": 3050 + }, + { + "epoch": 0.3350538106742807, + "grad_norm": 2.3963820934295654, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7250658273696899, + "num_tokens": 78125919.0, + "step": 3051 + }, + { + "epoch": 0.33516362837689434, + "grad_norm": 2.3036458492279053, + "learning_rate": 1e-06, + "loss": 1.0309, + "mean_token_accuracy": 0.6899014711380005, + "num_tokens": 78149661.0, + "step": 3052 + }, + { + "epoch": 0.33527344607950804, + "grad_norm": 2.1480484008789062, + "learning_rate": 1e-06, + "loss": 1.0729, + "mean_token_accuracy": 0.6849918961524963, + "num_tokens": 78177556.0, + "step": 3053 + }, + { + "epoch": 0.3353832637821217, + "grad_norm": 2.285968542098999, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6980418562889099, + "num_tokens": 78200727.0, + "step": 3054 + }, + { + "epoch": 0.33549308148473533, + "grad_norm": 2.125624895095825, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.715123176574707, + "num_tokens": 78226411.0, + "step": 3055 + }, + { + "epoch": 0.335602899187349, + "grad_norm": 2.4178621768951416, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7204015254974365, + "num_tokens": 78251511.0, + "step": 3056 + }, + { + "epoch": 0.3357127168899627, + "grad_norm": 2.108588218688965, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7122155427932739, + "num_tokens": 78278613.0, + "step": 3057 + }, + { + "epoch": 0.3358225345925763, + "grad_norm": 2.0416669845581055, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.6915847063064575, + "num_tokens": 78308781.0, + "step": 3058 + }, + { + "epoch": 0.33593235229519, + "grad_norm": 2.359571933746338, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7080381512641907, + "num_tokens": 78332005.0, + "step": 3059 + }, + { + "epoch": 0.3360421699978036, + "grad_norm": 2.2555012702941895, + "learning_rate": 1e-06, + "loss": 1.0307, + "mean_token_accuracy": 0.6918102502822876, + "num_tokens": 78356692.0, + "step": 3060 + }, + { + "epoch": 0.3361519877004173, + "grad_norm": 2.5615086555480957, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7169548273086548, + "num_tokens": 78378287.0, + "step": 3061 + }, + { + "epoch": 0.33626180540303097, + "grad_norm": 1.9887901544570923, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7064196467399597, + "num_tokens": 78407694.0, + "step": 3062 + }, + { + "epoch": 0.3363716231056446, + "grad_norm": 2.34791898727417, + "learning_rate": 1e-06, + "loss": 1.0473, + "mean_token_accuracy": 0.6856867074966431, + "num_tokens": 78431785.0, + "step": 3063 + }, + { + "epoch": 0.3364814408082583, + "grad_norm": 2.286473512649536, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.685705304145813, + "num_tokens": 78456175.0, + "step": 3064 + }, + { + "epoch": 0.33659125851087196, + "grad_norm": 1.9469847679138184, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6836868524551392, + "num_tokens": 78486970.0, + "step": 3065 + }, + { + "epoch": 0.3367010762134856, + "grad_norm": 2.0225541591644287, + "learning_rate": 1e-06, + "loss": 1.0295, + "mean_token_accuracy": 0.6957520246505737, + "num_tokens": 78516813.0, + "step": 3066 + }, + { + "epoch": 0.33681089391609925, + "grad_norm": 2.2024917602539062, + "learning_rate": 1e-06, + "loss": 1.046, + "mean_token_accuracy": 0.6857238411903381, + "num_tokens": 78541725.0, + "step": 3067 + }, + { + "epoch": 0.33692071161871295, + "grad_norm": 2.419909954071045, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7100693583488464, + "num_tokens": 78563047.0, + "step": 3068 + }, + { + "epoch": 0.3370305293213266, + "grad_norm": 2.1862800121307373, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7193808555603027, + "num_tokens": 78589700.0, + "step": 3069 + }, + { + "epoch": 0.33714034702394025, + "grad_norm": 2.2818796634674072, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7251671552658081, + "num_tokens": 78612378.0, + "step": 3070 + }, + { + "epoch": 0.33725016472655395, + "grad_norm": 2.3211472034454346, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.69376540184021, + "num_tokens": 78636954.0, + "step": 3071 + }, + { + "epoch": 0.3373599824291676, + "grad_norm": 2.0649242401123047, + "learning_rate": 1e-06, + "loss": 1.103, + "mean_token_accuracy": 0.6687459349632263, + "num_tokens": 78667246.0, + "step": 3072 + }, + { + "epoch": 0.33746980013178124, + "grad_norm": 2.1572751998901367, + "learning_rate": 1e-06, + "loss": 1.0961, + "mean_token_accuracy": 0.6795269846916199, + "num_tokens": 78694529.0, + "step": 3073 + }, + { + "epoch": 0.3375796178343949, + "grad_norm": 2.6825127601623535, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7187474966049194, + "num_tokens": 78714814.0, + "step": 3074 + }, + { + "epoch": 0.3376894355370086, + "grad_norm": 1.995581030845642, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7190479040145874, + "num_tokens": 78744147.0, + "step": 3075 + }, + { + "epoch": 0.33779925323962223, + "grad_norm": 2.156761884689331, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6941993236541748, + "num_tokens": 78770181.0, + "step": 3076 + }, + { + "epoch": 0.3379090709422359, + "grad_norm": 2.30842661857605, + "learning_rate": 1e-06, + "loss": 1.0221, + "mean_token_accuracy": 0.6931757926940918, + "num_tokens": 78798518.0, + "step": 3077 + }, + { + "epoch": 0.3380188886448495, + "grad_norm": 2.222764492034912, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.6983643770217896, + "num_tokens": 78825242.0, + "step": 3078 + }, + { + "epoch": 0.3381287063474632, + "grad_norm": 2.2886974811553955, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7186145782470703, + "num_tokens": 78847421.0, + "step": 3079 + }, + { + "epoch": 0.33823852405007687, + "grad_norm": 2.1475467681884766, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.6987633109092712, + "num_tokens": 78874719.0, + "step": 3080 + }, + { + "epoch": 0.3383483417526905, + "grad_norm": 2.2439093589782715, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6892318725585938, + "num_tokens": 78900361.0, + "step": 3081 + }, + { + "epoch": 0.3384581594553042, + "grad_norm": 2.572190284729004, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6947267055511475, + "num_tokens": 78920337.0, + "step": 3082 + }, + { + "epoch": 0.33856797715791787, + "grad_norm": 2.0833685398101807, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7161816358566284, + "num_tokens": 78950602.0, + "step": 3083 + }, + { + "epoch": 0.3386777948605315, + "grad_norm": 2.0293359756469727, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7082264423370361, + "num_tokens": 78980791.0, + "step": 3084 + }, + { + "epoch": 0.33878761256314516, + "grad_norm": 2.28578519821167, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.7037127017974854, + "num_tokens": 79003807.0, + "step": 3085 + }, + { + "epoch": 0.33889743026575886, + "grad_norm": 2.1438639163970947, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.7019911408424377, + "num_tokens": 79030886.0, + "step": 3086 + }, + { + "epoch": 0.3390072479683725, + "grad_norm": 2.4422335624694824, + "learning_rate": 1e-06, + "loss": 1.0786, + "mean_token_accuracy": 0.6822119355201721, + "num_tokens": 79053525.0, + "step": 3087 + }, + { + "epoch": 0.33911706567098615, + "grad_norm": 2.238863945007324, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7189245223999023, + "num_tokens": 79079178.0, + "step": 3088 + }, + { + "epoch": 0.3392268833735998, + "grad_norm": 2.2150418758392334, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7045971155166626, + "num_tokens": 79105314.0, + "step": 3089 + }, + { + "epoch": 0.3393367010762135, + "grad_norm": 2.4491677284240723, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6921916604042053, + "num_tokens": 79125769.0, + "step": 3090 + }, + { + "epoch": 0.33944651877882714, + "grad_norm": 2.235713005065918, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7157245874404907, + "num_tokens": 79151755.0, + "step": 3091 + }, + { + "epoch": 0.3395563364814408, + "grad_norm": 2.2646405696868896, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6935725212097168, + "num_tokens": 79175183.0, + "step": 3092 + }, + { + "epoch": 0.3396661541840545, + "grad_norm": 2.1934103965759277, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6927427053451538, + "num_tokens": 79200812.0, + "step": 3093 + }, + { + "epoch": 0.33977597188666814, + "grad_norm": 2.0811805725097656, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.7101932764053345, + "num_tokens": 79227903.0, + "step": 3094 + }, + { + "epoch": 0.3398857895892818, + "grad_norm": 2.3960554599761963, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7074777483940125, + "num_tokens": 79249041.0, + "step": 3095 + }, + { + "epoch": 0.33999560729189543, + "grad_norm": 2.2386703491210938, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.7008367776870728, + "num_tokens": 79274202.0, + "step": 3096 + }, + { + "epoch": 0.34010542499450913, + "grad_norm": 2.4646975994110107, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.699051022529602, + "num_tokens": 79298336.0, + "step": 3097 + }, + { + "epoch": 0.3402152426971228, + "grad_norm": 2.2594363689422607, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6922644376754761, + "num_tokens": 79323031.0, + "step": 3098 + }, + { + "epoch": 0.3403250603997364, + "grad_norm": 2.0350582599639893, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7070440649986267, + "num_tokens": 79356433.0, + "step": 3099 + }, + { + "epoch": 0.3404348781023501, + "grad_norm": 2.0029215812683105, + "learning_rate": 1e-06, + "loss": 1.1491, + "mean_token_accuracy": 0.6568443179130554, + "num_tokens": 79388958.0, + "step": 3100 + }, + { + "epoch": 0.34054469580496377, + "grad_norm": 2.331836462020874, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7190110683441162, + "num_tokens": 79412830.0, + "step": 3101 + }, + { + "epoch": 0.3406545135075774, + "grad_norm": 2.3010878562927246, + "learning_rate": 1e-06, + "loss": 1.0545, + "mean_token_accuracy": 0.6832247972488403, + "num_tokens": 79436142.0, + "step": 3102 + }, + { + "epoch": 0.34076433121019106, + "grad_norm": 2.2985689640045166, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.6927527785301208, + "num_tokens": 79461153.0, + "step": 3103 + }, + { + "epoch": 0.34087414891280476, + "grad_norm": 2.2149956226348877, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.6972825527191162, + "num_tokens": 79485270.0, + "step": 3104 + }, + { + "epoch": 0.3409839666154184, + "grad_norm": 2.2423174381256104, + "learning_rate": 1e-06, + "loss": 1.0226, + "mean_token_accuracy": 0.6939179301261902, + "num_tokens": 79510996.0, + "step": 3105 + }, + { + "epoch": 0.34109378431803206, + "grad_norm": 2.4264278411865234, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7242255806922913, + "num_tokens": 79532111.0, + "step": 3106 + }, + { + "epoch": 0.3412036020206457, + "grad_norm": 2.3070578575134277, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6963716745376587, + "num_tokens": 79556727.0, + "step": 3107 + }, + { + "epoch": 0.3413134197232594, + "grad_norm": 2.3718883991241455, + "learning_rate": 1e-06, + "loss": 1.0781, + "mean_token_accuracy": 0.6766782999038696, + "num_tokens": 79583443.0, + "step": 3108 + }, + { + "epoch": 0.34142323742587305, + "grad_norm": 2.294649124145508, + "learning_rate": 1e-06, + "loss": 1.0223, + "mean_token_accuracy": 0.6906867623329163, + "num_tokens": 79609238.0, + "step": 3109 + }, + { + "epoch": 0.3415330551284867, + "grad_norm": 2.0108790397644043, + "learning_rate": 1e-06, + "loss": 1.1194, + "mean_token_accuracy": 0.6657503247261047, + "num_tokens": 79640777.0, + "step": 3110 + }, + { + "epoch": 0.3416428728311004, + "grad_norm": 2.3042447566986084, + "learning_rate": 1e-06, + "loss": 1.1047, + "mean_token_accuracy": 0.6696303486824036, + "num_tokens": 79664623.0, + "step": 3111 + }, + { + "epoch": 0.34175269053371404, + "grad_norm": 2.379575490951538, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7202697992324829, + "num_tokens": 79686169.0, + "step": 3112 + }, + { + "epoch": 0.3418625082363277, + "grad_norm": 2.1481740474700928, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.6874300241470337, + "num_tokens": 79715033.0, + "step": 3113 + }, + { + "epoch": 0.34197232593894134, + "grad_norm": 2.5624032020568848, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7249366641044617, + "num_tokens": 79733567.0, + "step": 3114 + }, + { + "epoch": 0.34208214364155504, + "grad_norm": 2.32663631439209, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7108259201049805, + "num_tokens": 79754646.0, + "step": 3115 + }, + { + "epoch": 0.3421919613441687, + "grad_norm": 2.117675304412842, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6902151107788086, + "num_tokens": 79781789.0, + "step": 3116 + }, + { + "epoch": 0.34230177904678233, + "grad_norm": 2.0505690574645996, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6883646845817566, + "num_tokens": 79809715.0, + "step": 3117 + }, + { + "epoch": 0.34241159674939603, + "grad_norm": 2.1930973529815674, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7089553475379944, + "num_tokens": 79835054.0, + "step": 3118 + }, + { + "epoch": 0.3425214144520097, + "grad_norm": 2.191610813140869, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6998621225357056, + "num_tokens": 79860390.0, + "step": 3119 + }, + { + "epoch": 0.3426312321546233, + "grad_norm": 2.1777079105377197, + "learning_rate": 1e-06, + "loss": 1.0683, + "mean_token_accuracy": 0.6811099052429199, + "num_tokens": 79887343.0, + "step": 3120 + }, + { + "epoch": 0.34274104985723697, + "grad_norm": 1.9626623392105103, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7078995704650879, + "num_tokens": 79918293.0, + "step": 3121 + }, + { + "epoch": 0.34285086755985067, + "grad_norm": 2.2953269481658936, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.6971985101699829, + "num_tokens": 79943467.0, + "step": 3122 + }, + { + "epoch": 0.3429606852624643, + "grad_norm": 1.9752819538116455, + "learning_rate": 1e-06, + "loss": 1.0675, + "mean_token_accuracy": 0.677513837814331, + "num_tokens": 79974976.0, + "step": 3123 + }, + { + "epoch": 0.34307050296507796, + "grad_norm": 2.387883424758911, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7088993191719055, + "num_tokens": 79998722.0, + "step": 3124 + }, + { + "epoch": 0.3431803206676916, + "grad_norm": 2.000965118408203, + "learning_rate": 1e-06, + "loss": 1.0331, + "mean_token_accuracy": 0.6930078268051147, + "num_tokens": 80030529.0, + "step": 3125 + }, + { + "epoch": 0.3432901383703053, + "grad_norm": 2.420938014984131, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.6977322697639465, + "num_tokens": 80054320.0, + "step": 3126 + }, + { + "epoch": 0.34339995607291895, + "grad_norm": 2.558213233947754, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7118226289749146, + "num_tokens": 80074599.0, + "step": 3127 + }, + { + "epoch": 0.3435097737755326, + "grad_norm": 2.07009220123291, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.695298969745636, + "num_tokens": 80103603.0, + "step": 3128 + }, + { + "epoch": 0.3436195914781463, + "grad_norm": 2.368481159210205, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.7018366456031799, + "num_tokens": 80126704.0, + "step": 3129 + }, + { + "epoch": 0.34372940918075995, + "grad_norm": 2.0734784603118896, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6966708898544312, + "num_tokens": 80154198.0, + "step": 3130 + }, + { + "epoch": 0.3438392268833736, + "grad_norm": 2.4312491416931152, + "learning_rate": 1e-06, + "loss": 1.0812, + "mean_token_accuracy": 0.6754839420318604, + "num_tokens": 80178676.0, + "step": 3131 + }, + { + "epoch": 0.34394904458598724, + "grad_norm": 2.248589515686035, + "learning_rate": 1e-06, + "loss": 1.0665, + "mean_token_accuracy": 0.68267422914505, + "num_tokens": 80205184.0, + "step": 3132 + }, + { + "epoch": 0.34405886228860094, + "grad_norm": 2.3166871070861816, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7035375833511353, + "num_tokens": 80228588.0, + "step": 3133 + }, + { + "epoch": 0.3441686799912146, + "grad_norm": 2.1344823837280273, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.7015470266342163, + "num_tokens": 80255218.0, + "step": 3134 + }, + { + "epoch": 0.34427849769382823, + "grad_norm": 2.4898509979248047, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7314569354057312, + "num_tokens": 80274660.0, + "step": 3135 + }, + { + "epoch": 0.3443883153964419, + "grad_norm": 2.4219353199005127, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6953213810920715, + "num_tokens": 80297527.0, + "step": 3136 + }, + { + "epoch": 0.3444981330990556, + "grad_norm": 2.5911941528320312, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7024663686752319, + "num_tokens": 80318691.0, + "step": 3137 + }, + { + "epoch": 0.3446079508016692, + "grad_norm": 2.467061758041382, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.712602972984314, + "num_tokens": 80339707.0, + "step": 3138 + }, + { + "epoch": 0.3447177685042829, + "grad_norm": 2.0242018699645996, + "learning_rate": 1e-06, + "loss": 1.0313, + "mean_token_accuracy": 0.691396176815033, + "num_tokens": 80368028.0, + "step": 3139 + }, + { + "epoch": 0.3448275862068966, + "grad_norm": 2.0716705322265625, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7057195901870728, + "num_tokens": 80396063.0, + "step": 3140 + }, + { + "epoch": 0.3449374039095102, + "grad_norm": 2.3981375694274902, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6890004873275757, + "num_tokens": 80420122.0, + "step": 3141 + }, + { + "epoch": 0.34504722161212387, + "grad_norm": 2.180652618408203, + "learning_rate": 1e-06, + "loss": 1.1073, + "mean_token_accuracy": 0.6653589010238647, + "num_tokens": 80447768.0, + "step": 3142 + }, + { + "epoch": 0.3451570393147375, + "grad_norm": 2.1172444820404053, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.6871689558029175, + "num_tokens": 80474634.0, + "step": 3143 + }, + { + "epoch": 0.3452668570173512, + "grad_norm": 2.3365821838378906, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7331774234771729, + "num_tokens": 80496306.0, + "step": 3144 + }, + { + "epoch": 0.34537667471996486, + "grad_norm": 2.4125099182128906, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7157310247421265, + "num_tokens": 80517455.0, + "step": 3145 + }, + { + "epoch": 0.3454864924225785, + "grad_norm": 2.223203659057617, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7237275838851929, + "num_tokens": 80542239.0, + "step": 3146 + }, + { + "epoch": 0.3455963101251922, + "grad_norm": 2.118645191192627, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7190076112747192, + "num_tokens": 80568301.0, + "step": 3147 + }, + { + "epoch": 0.34570612782780585, + "grad_norm": 2.5243968963623047, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7344564199447632, + "num_tokens": 80587598.0, + "step": 3148 + }, + { + "epoch": 0.3458159455304195, + "grad_norm": 2.111879348754883, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.6972272992134094, + "num_tokens": 80615142.0, + "step": 3149 + }, + { + "epoch": 0.34592576323303315, + "grad_norm": 2.231578826904297, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7047541737556458, + "num_tokens": 80640301.0, + "step": 3150 + }, + { + "epoch": 0.34603558093564685, + "grad_norm": 2.375581979751587, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6987123489379883, + "num_tokens": 80664010.0, + "step": 3151 + }, + { + "epoch": 0.3461453986382605, + "grad_norm": 2.271721363067627, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7018997669219971, + "num_tokens": 80690042.0, + "step": 3152 + }, + { + "epoch": 0.34625521634087414, + "grad_norm": 2.022303342819214, + "learning_rate": 1e-06, + "loss": 1.0368, + "mean_token_accuracy": 0.6898661255836487, + "num_tokens": 80720550.0, + "step": 3153 + }, + { + "epoch": 0.3463650340434878, + "grad_norm": 2.427666187286377, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7184320688247681, + "num_tokens": 80743367.0, + "step": 3154 + }, + { + "epoch": 0.3464748517461015, + "grad_norm": 2.255084991455078, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7210550904273987, + "num_tokens": 80768328.0, + "step": 3155 + }, + { + "epoch": 0.34658466944871513, + "grad_norm": 2.2975821495056152, + "learning_rate": 1e-06, + "loss": 1.0521, + "mean_token_accuracy": 0.6823105812072754, + "num_tokens": 80792793.0, + "step": 3156 + }, + { + "epoch": 0.3466944871513288, + "grad_norm": 2.0242700576782227, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7175298929214478, + "num_tokens": 80821705.0, + "step": 3157 + }, + { + "epoch": 0.3468043048539425, + "grad_norm": 2.611078977584839, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6945120096206665, + "num_tokens": 80841631.0, + "step": 3158 + }, + { + "epoch": 0.3469141225565561, + "grad_norm": 2.057004690170288, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.6931639909744263, + "num_tokens": 80872688.0, + "step": 3159 + }, + { + "epoch": 0.34702394025916977, + "grad_norm": 2.0628607273101807, + "learning_rate": 1e-06, + "loss": 1.0593, + "mean_token_accuracy": 0.6832021474838257, + "num_tokens": 80904288.0, + "step": 3160 + }, + { + "epoch": 0.3471337579617834, + "grad_norm": 2.081254720687866, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7042050361633301, + "num_tokens": 80933199.0, + "step": 3161 + }, + { + "epoch": 0.3472435756643971, + "grad_norm": 2.1455845832824707, + "learning_rate": 1e-06, + "loss": 1.0962, + "mean_token_accuracy": 0.6750383377075195, + "num_tokens": 80964058.0, + "step": 3162 + }, + { + "epoch": 0.34735339336701077, + "grad_norm": 1.9795416593551636, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.6944903135299683, + "num_tokens": 80996169.0, + "step": 3163 + }, + { + "epoch": 0.3474632110696244, + "grad_norm": 2.306835889816284, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7115802764892578, + "num_tokens": 81019108.0, + "step": 3164 + }, + { + "epoch": 0.34757302877223806, + "grad_norm": 1.837167739868164, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.6993094086647034, + "num_tokens": 81054399.0, + "step": 3165 + }, + { + "epoch": 0.34768284647485176, + "grad_norm": 2.1137332916259766, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7078762650489807, + "num_tokens": 81080559.0, + "step": 3166 + }, + { + "epoch": 0.3477926641774654, + "grad_norm": 2.2160587310791016, + "learning_rate": 1e-06, + "loss": 1.0429, + "mean_token_accuracy": 0.6864601969718933, + "num_tokens": 81108269.0, + "step": 3167 + }, + { + "epoch": 0.34790248188007905, + "grad_norm": 2.4974513053894043, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7139103412628174, + "num_tokens": 81129075.0, + "step": 3168 + }, + { + "epoch": 0.34801229958269275, + "grad_norm": 2.008413076400757, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.7151566743850708, + "num_tokens": 81155806.0, + "step": 3169 + }, + { + "epoch": 0.3481221172853064, + "grad_norm": 2.608206272125244, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7105485200881958, + "num_tokens": 81174731.0, + "step": 3170 + }, + { + "epoch": 0.34823193498792004, + "grad_norm": 2.3264644145965576, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7041879892349243, + "num_tokens": 81196946.0, + "step": 3171 + }, + { + "epoch": 0.3483417526905337, + "grad_norm": 2.3309872150421143, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7135269641876221, + "num_tokens": 81220530.0, + "step": 3172 + }, + { + "epoch": 0.3484515703931474, + "grad_norm": 2.135664463043213, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7235934734344482, + "num_tokens": 81247517.0, + "step": 3173 + }, + { + "epoch": 0.34856138809576104, + "grad_norm": 2.428175926208496, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7053054571151733, + "num_tokens": 81268453.0, + "step": 3174 + }, + { + "epoch": 0.3486712057983747, + "grad_norm": 2.3087570667266846, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.701047956943512, + "num_tokens": 81292943.0, + "step": 3175 + }, + { + "epoch": 0.3487810235009884, + "grad_norm": 2.1926627159118652, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7051847577095032, + "num_tokens": 81317141.0, + "step": 3176 + }, + { + "epoch": 0.34889084120360203, + "grad_norm": 2.3945302963256836, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6956198215484619, + "num_tokens": 81340575.0, + "step": 3177 + }, + { + "epoch": 0.3490006589062157, + "grad_norm": 2.1380467414855957, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.6988615989685059, + "num_tokens": 81366122.0, + "step": 3178 + }, + { + "epoch": 0.3491104766088293, + "grad_norm": 2.459618330001831, + "learning_rate": 1e-06, + "loss": 1.0302, + "mean_token_accuracy": 0.685778021812439, + "num_tokens": 81388333.0, + "step": 3179 + }, + { + "epoch": 0.349220294311443, + "grad_norm": 2.3839075565338135, + "learning_rate": 1e-06, + "loss": 1.0517, + "mean_token_accuracy": 0.6885777711868286, + "num_tokens": 81411616.0, + "step": 3180 + }, + { + "epoch": 0.34933011201405667, + "grad_norm": 2.45806884765625, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7048465609550476, + "num_tokens": 81433122.0, + "step": 3181 + }, + { + "epoch": 0.3494399297166703, + "grad_norm": 2.4488449096679688, + "learning_rate": 1e-06, + "loss": 1.0487, + "mean_token_accuracy": 0.6827946901321411, + "num_tokens": 81459388.0, + "step": 3182 + }, + { + "epoch": 0.34954974741928396, + "grad_norm": 2.2399632930755615, + "learning_rate": 1e-06, + "loss": 1.0382, + "mean_token_accuracy": 0.6923984885215759, + "num_tokens": 81485794.0, + "step": 3183 + }, + { + "epoch": 0.34965956512189766, + "grad_norm": 1.91651451587677, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.694329559803009, + "num_tokens": 81518434.0, + "step": 3184 + }, + { + "epoch": 0.3497693828245113, + "grad_norm": 2.385611057281494, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.6990604400634766, + "num_tokens": 81542102.0, + "step": 3185 + }, + { + "epoch": 0.34987920052712496, + "grad_norm": 2.5913844108581543, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7090262174606323, + "num_tokens": 81561776.0, + "step": 3186 + }, + { + "epoch": 0.34998901822973866, + "grad_norm": 2.3275864124298096, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.703217625617981, + "num_tokens": 81586497.0, + "step": 3187 + }, + { + "epoch": 0.3500988359323523, + "grad_norm": 2.4811854362487793, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7032369375228882, + "num_tokens": 81608209.0, + "step": 3188 + }, + { + "epoch": 0.35020865363496595, + "grad_norm": 2.270252227783203, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7167061567306519, + "num_tokens": 81630666.0, + "step": 3189 + }, + { + "epoch": 0.3503184713375796, + "grad_norm": 2.5158329010009766, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.7001780867576599, + "num_tokens": 81650638.0, + "step": 3190 + }, + { + "epoch": 0.3504282890401933, + "grad_norm": 2.1034512519836426, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7041939496994019, + "num_tokens": 81678225.0, + "step": 3191 + }, + { + "epoch": 0.35053810674280694, + "grad_norm": 2.271483898162842, + "learning_rate": 1e-06, + "loss": 1.0418, + "mean_token_accuracy": 0.6941636800765991, + "num_tokens": 81702202.0, + "step": 3192 + }, + { + "epoch": 0.3506479244454206, + "grad_norm": 2.288428544998169, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7048532962799072, + "num_tokens": 81726937.0, + "step": 3193 + }, + { + "epoch": 0.3507577421480343, + "grad_norm": 2.301483392715454, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6941773891448975, + "num_tokens": 81753291.0, + "step": 3194 + }, + { + "epoch": 0.35086755985064794, + "grad_norm": 2.1906793117523193, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6926711201667786, + "num_tokens": 81781630.0, + "step": 3195 + }, + { + "epoch": 0.3509773775532616, + "grad_norm": 2.688676357269287, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7248802185058594, + "num_tokens": 81798609.0, + "step": 3196 + }, + { + "epoch": 0.35108719525587523, + "grad_norm": 2.3542490005493164, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.7046478986740112, + "num_tokens": 81822109.0, + "step": 3197 + }, + { + "epoch": 0.35119701295848893, + "grad_norm": 2.248908758163452, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.7007231712341309, + "num_tokens": 81847702.0, + "step": 3198 + }, + { + "epoch": 0.3513068306611026, + "grad_norm": 2.1814210414886475, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6907670497894287, + "num_tokens": 81872848.0, + "step": 3199 + }, + { + "epoch": 0.3514166483637162, + "grad_norm": 2.130311965942383, + "learning_rate": 1e-06, + "loss": 1.055, + "mean_token_accuracy": 0.6889743804931641, + "num_tokens": 81900379.0, + "step": 3200 + }, + { + "epoch": 0.35152646606632987, + "grad_norm": 2.523740291595459, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7165582776069641, + "num_tokens": 81921168.0, + "step": 3201 + }, + { + "epoch": 0.35163628376894357, + "grad_norm": 2.1756112575531006, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7103289365768433, + "num_tokens": 81948720.0, + "step": 3202 + }, + { + "epoch": 0.3517461014715572, + "grad_norm": 2.1308226585388184, + "learning_rate": 1e-06, + "loss": 1.0256, + "mean_token_accuracy": 0.6852025985717773, + "num_tokens": 81976943.0, + "step": 3203 + }, + { + "epoch": 0.35185591917417086, + "grad_norm": 2.251300811767578, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6927964687347412, + "num_tokens": 82002777.0, + "step": 3204 + }, + { + "epoch": 0.35196573687678456, + "grad_norm": 2.3434865474700928, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.710253894329071, + "num_tokens": 82027256.0, + "step": 3205 + }, + { + "epoch": 0.3520755545793982, + "grad_norm": 1.8966927528381348, + "learning_rate": 1e-06, + "loss": 1.0764, + "mean_token_accuracy": 0.6792932748794556, + "num_tokens": 82063214.0, + "step": 3206 + }, + { + "epoch": 0.35218537228201185, + "grad_norm": 2.242757797241211, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.7044092416763306, + "num_tokens": 82087446.0, + "step": 3207 + }, + { + "epoch": 0.3522951899846255, + "grad_norm": 2.1467018127441406, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.6895924210548401, + "num_tokens": 82114999.0, + "step": 3208 + }, + { + "epoch": 0.3524050076872392, + "grad_norm": 1.9916743040084839, + "learning_rate": 1e-06, + "loss": 1.0787, + "mean_token_accuracy": 0.6828920245170593, + "num_tokens": 82145729.0, + "step": 3209 + }, + { + "epoch": 0.35251482538985285, + "grad_norm": 2.333279848098755, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7247174382209778, + "num_tokens": 82166430.0, + "step": 3210 + }, + { + "epoch": 0.3526246430924665, + "grad_norm": 2.243734359741211, + "learning_rate": 1e-06, + "loss": 1.0244, + "mean_token_accuracy": 0.696147084236145, + "num_tokens": 82190939.0, + "step": 3211 + }, + { + "epoch": 0.35273446079508014, + "grad_norm": 2.1545000076293945, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7092684507369995, + "num_tokens": 82216416.0, + "step": 3212 + }, + { + "epoch": 0.35284427849769384, + "grad_norm": 1.9390335083007812, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.7018511295318604, + "num_tokens": 82246768.0, + "step": 3213 + }, + { + "epoch": 0.3529540962003075, + "grad_norm": 2.308682680130005, + "learning_rate": 1e-06, + "loss": 1.0803, + "mean_token_accuracy": 0.6824417114257812, + "num_tokens": 82272226.0, + "step": 3214 + }, + { + "epoch": 0.35306391390292113, + "grad_norm": 2.326179265975952, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6959085464477539, + "num_tokens": 82295776.0, + "step": 3215 + }, + { + "epoch": 0.35317373160553484, + "grad_norm": 1.9898208379745483, + "learning_rate": 1e-06, + "loss": 1.0556, + "mean_token_accuracy": 0.6809326410293579, + "num_tokens": 82327879.0, + "step": 3216 + }, + { + "epoch": 0.3532835493081485, + "grad_norm": 2.331792116165161, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.7037236094474792, + "num_tokens": 82353070.0, + "step": 3217 + }, + { + "epoch": 0.3533933670107621, + "grad_norm": 2.2739999294281006, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.699627161026001, + "num_tokens": 82377720.0, + "step": 3218 + }, + { + "epoch": 0.3535031847133758, + "grad_norm": 2.368697166442871, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.695507824420929, + "num_tokens": 82399938.0, + "step": 3219 + }, + { + "epoch": 0.3536130024159895, + "grad_norm": 2.060274362564087, + "learning_rate": 1e-06, + "loss": 1.0388, + "mean_token_accuracy": 0.6868168115615845, + "num_tokens": 82430072.0, + "step": 3220 + }, + { + "epoch": 0.3537228201186031, + "grad_norm": 2.3181333541870117, + "learning_rate": 1e-06, + "loss": 1.0616, + "mean_token_accuracy": 0.6876163482666016, + "num_tokens": 82455024.0, + "step": 3221 + }, + { + "epoch": 0.35383263782121677, + "grad_norm": 2.105058431625366, + "learning_rate": 1e-06, + "loss": 1.1082, + "mean_token_accuracy": 0.6686631441116333, + "num_tokens": 82485032.0, + "step": 3222 + }, + { + "epoch": 0.35394245552383047, + "grad_norm": 2.3757143020629883, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.6961811184883118, + "num_tokens": 82506801.0, + "step": 3223 + }, + { + "epoch": 0.3540522732264441, + "grad_norm": 2.3914523124694824, + "learning_rate": 1e-06, + "loss": 1.073, + "mean_token_accuracy": 0.6783617734909058, + "num_tokens": 82531624.0, + "step": 3224 + }, + { + "epoch": 0.35416209092905776, + "grad_norm": 2.4158546924591064, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7249177694320679, + "num_tokens": 82554139.0, + "step": 3225 + }, + { + "epoch": 0.3542719086316714, + "grad_norm": 2.424208879470825, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7224826812744141, + "num_tokens": 82575347.0, + "step": 3226 + }, + { + "epoch": 0.3543817263342851, + "grad_norm": 1.9953163862228394, + "learning_rate": 1e-06, + "loss": 1.0736, + "mean_token_accuracy": 0.6868549585342407, + "num_tokens": 82606892.0, + "step": 3227 + }, + { + "epoch": 0.35449154403689875, + "grad_norm": 2.307368278503418, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7156224846839905, + "num_tokens": 82629890.0, + "step": 3228 + }, + { + "epoch": 0.3546013617395124, + "grad_norm": 2.333588123321533, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7394202947616577, + "num_tokens": 82650823.0, + "step": 3229 + }, + { + "epoch": 0.35471117944212605, + "grad_norm": 2.2356743812561035, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.7070258855819702, + "num_tokens": 82674572.0, + "step": 3230 + }, + { + "epoch": 0.35482099714473975, + "grad_norm": 2.5828402042388916, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.7017815709114075, + "num_tokens": 82696058.0, + "step": 3231 + }, + { + "epoch": 0.3549308148473534, + "grad_norm": 2.1357433795928955, + "learning_rate": 1e-06, + "loss": 1.0377, + "mean_token_accuracy": 0.6855335235595703, + "num_tokens": 82725481.0, + "step": 3232 + }, + { + "epoch": 0.35504063254996704, + "grad_norm": 2.303938627243042, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7192510366439819, + "num_tokens": 82749309.0, + "step": 3233 + }, + { + "epoch": 0.35515045025258074, + "grad_norm": 2.8535120487213135, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6944433450698853, + "num_tokens": 82772690.0, + "step": 3234 + }, + { + "epoch": 0.3552602679551944, + "grad_norm": 2.2570762634277344, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7147142291069031, + "num_tokens": 82797150.0, + "step": 3235 + }, + { + "epoch": 0.35537008565780803, + "grad_norm": 2.127540111541748, + "learning_rate": 1e-06, + "loss": 1.0643, + "mean_token_accuracy": 0.6801465749740601, + "num_tokens": 82824900.0, + "step": 3236 + }, + { + "epoch": 0.3554799033604217, + "grad_norm": 2.2142601013183594, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6987094879150391, + "num_tokens": 82852714.0, + "step": 3237 + }, + { + "epoch": 0.3555897210630354, + "grad_norm": 2.100041627883911, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7210512161254883, + "num_tokens": 82879302.0, + "step": 3238 + }, + { + "epoch": 0.355699538765649, + "grad_norm": 2.3556036949157715, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.704338550567627, + "num_tokens": 82902096.0, + "step": 3239 + }, + { + "epoch": 0.35580935646826267, + "grad_norm": 1.9715523719787598, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7091137170791626, + "num_tokens": 82935120.0, + "step": 3240 + }, + { + "epoch": 0.3559191741708763, + "grad_norm": 1.8655264377593994, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7160215377807617, + "num_tokens": 82966630.0, + "step": 3241 + }, + { + "epoch": 0.35602899187349, + "grad_norm": 2.2779643535614014, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7095357179641724, + "num_tokens": 82989944.0, + "step": 3242 + }, + { + "epoch": 0.35613880957610367, + "grad_norm": 2.4856767654418945, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.705431342124939, + "num_tokens": 83011929.0, + "step": 3243 + }, + { + "epoch": 0.3562486272787173, + "grad_norm": 2.458235025405884, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7013006210327148, + "num_tokens": 83033900.0, + "step": 3244 + }, + { + "epoch": 0.356358444981331, + "grad_norm": 2.0152857303619385, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.6977901458740234, + "num_tokens": 83062773.0, + "step": 3245 + }, + { + "epoch": 0.35646826268394466, + "grad_norm": 2.6041877269744873, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.6971011757850647, + "num_tokens": 83083648.0, + "step": 3246 + }, + { + "epoch": 0.3565780803865583, + "grad_norm": 2.0413811206817627, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7128812074661255, + "num_tokens": 83113395.0, + "step": 3247 + }, + { + "epoch": 0.35668789808917195, + "grad_norm": 2.3796226978302, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.717101514339447, + "num_tokens": 83133901.0, + "step": 3248 + }, + { + "epoch": 0.35679771579178565, + "grad_norm": 2.0301971435546875, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.701897382736206, + "num_tokens": 83164115.0, + "step": 3249 + }, + { + "epoch": 0.3569075334943993, + "grad_norm": 2.263058662414551, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6966081857681274, + "num_tokens": 83190150.0, + "step": 3250 + }, + { + "epoch": 0.35701735119701294, + "grad_norm": 2.539735794067383, + "learning_rate": 1e-06, + "loss": 1.0365, + "mean_token_accuracy": 0.6927229762077332, + "num_tokens": 83210846.0, + "step": 3251 + }, + { + "epoch": 0.35712716889962665, + "grad_norm": 1.981563687324524, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7139789462089539, + "num_tokens": 83240680.0, + "step": 3252 + }, + { + "epoch": 0.3572369866022403, + "grad_norm": 2.1922824382781982, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6971220970153809, + "num_tokens": 83265262.0, + "step": 3253 + }, + { + "epoch": 0.35734680430485394, + "grad_norm": 2.2982234954833984, + "learning_rate": 1e-06, + "loss": 1.0433, + "mean_token_accuracy": 0.6905621886253357, + "num_tokens": 83288005.0, + "step": 3254 + }, + { + "epoch": 0.3574566220074676, + "grad_norm": 2.3323912620544434, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.7004444599151611, + "num_tokens": 83314460.0, + "step": 3255 + }, + { + "epoch": 0.3575664397100813, + "grad_norm": 2.196300506591797, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7052304744720459, + "num_tokens": 83340291.0, + "step": 3256 + }, + { + "epoch": 0.35767625741269493, + "grad_norm": 2.136385679244995, + "learning_rate": 1e-06, + "loss": 1.0594, + "mean_token_accuracy": 0.6916265487670898, + "num_tokens": 83369884.0, + "step": 3257 + }, + { + "epoch": 0.3577860751153086, + "grad_norm": 2.288114070892334, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6974983215332031, + "num_tokens": 83394118.0, + "step": 3258 + }, + { + "epoch": 0.3578958928179222, + "grad_norm": 2.5798985958099365, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7110146284103394, + "num_tokens": 83414447.0, + "step": 3259 + }, + { + "epoch": 0.3580057105205359, + "grad_norm": 1.9892578125, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7163892388343811, + "num_tokens": 83444827.0, + "step": 3260 + }, + { + "epoch": 0.35811552822314957, + "grad_norm": 2.3954153060913086, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7040709853172302, + "num_tokens": 83466534.0, + "step": 3261 + }, + { + "epoch": 0.3582253459257632, + "grad_norm": 2.250734567642212, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6974430084228516, + "num_tokens": 83491349.0, + "step": 3262 + }, + { + "epoch": 0.3583351636283769, + "grad_norm": 2.3766205310821533, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.7005251049995422, + "num_tokens": 83515798.0, + "step": 3263 + }, + { + "epoch": 0.35844498133099056, + "grad_norm": 2.5227270126342773, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7168931365013123, + "num_tokens": 83537146.0, + "step": 3264 + }, + { + "epoch": 0.3585547990336042, + "grad_norm": 2.174520254135132, + "learning_rate": 1e-06, + "loss": 1.0518, + "mean_token_accuracy": 0.6807395219802856, + "num_tokens": 83563942.0, + "step": 3265 + }, + { + "epoch": 0.35866461673621786, + "grad_norm": 2.028059959411621, + "learning_rate": 1e-06, + "loss": 1.1089, + "mean_token_accuracy": 0.6734554171562195, + "num_tokens": 83596415.0, + "step": 3266 + }, + { + "epoch": 0.35877443443883156, + "grad_norm": 2.464651584625244, + "learning_rate": 1e-06, + "loss": 1.0419, + "mean_token_accuracy": 0.679355800151825, + "num_tokens": 83618748.0, + "step": 3267 + }, + { + "epoch": 0.3588842521414452, + "grad_norm": 2.1732497215270996, + "learning_rate": 1e-06, + "loss": 1.065, + "mean_token_accuracy": 0.6785614490509033, + "num_tokens": 83647374.0, + "step": 3268 + }, + { + "epoch": 0.35899406984405885, + "grad_norm": 2.3116371631622314, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.7042810320854187, + "num_tokens": 83671631.0, + "step": 3269 + }, + { + "epoch": 0.35910388754667255, + "grad_norm": 2.0366709232330322, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6998647451400757, + "num_tokens": 83701205.0, + "step": 3270 + }, + { + "epoch": 0.3592137052492862, + "grad_norm": 2.47961688041687, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6960837244987488, + "num_tokens": 83724206.0, + "step": 3271 + }, + { + "epoch": 0.35932352295189984, + "grad_norm": 2.165158987045288, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6928710341453552, + "num_tokens": 83748994.0, + "step": 3272 + }, + { + "epoch": 0.3594333406545135, + "grad_norm": 2.5538411140441895, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7165200710296631, + "num_tokens": 83769132.0, + "step": 3273 + }, + { + "epoch": 0.3595431583571272, + "grad_norm": 2.1535489559173584, + "learning_rate": 1e-06, + "loss": 1.0545, + "mean_token_accuracy": 0.6852383613586426, + "num_tokens": 83796152.0, + "step": 3274 + }, + { + "epoch": 0.35965297605974084, + "grad_norm": 2.387779474258423, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.7023916244506836, + "num_tokens": 83819442.0, + "step": 3275 + }, + { + "epoch": 0.3597627937623545, + "grad_norm": 2.2180848121643066, + "learning_rate": 1e-06, + "loss": 1.0715, + "mean_token_accuracy": 0.6800030469894409, + "num_tokens": 83847946.0, + "step": 3276 + }, + { + "epoch": 0.35987261146496813, + "grad_norm": 2.4593255519866943, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6932355761528015, + "num_tokens": 83869544.0, + "step": 3277 + }, + { + "epoch": 0.35998242916758183, + "grad_norm": 2.0661301612854004, + "learning_rate": 1e-06, + "loss": 1.0592, + "mean_token_accuracy": 0.6882075071334839, + "num_tokens": 83898546.0, + "step": 3278 + }, + { + "epoch": 0.3600922468701955, + "grad_norm": 2.1828129291534424, + "learning_rate": 1e-06, + "loss": 1.0966, + "mean_token_accuracy": 0.6791329383850098, + "num_tokens": 83926812.0, + "step": 3279 + }, + { + "epoch": 0.3602020645728091, + "grad_norm": 2.537832260131836, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.709123969078064, + "num_tokens": 83947414.0, + "step": 3280 + }, + { + "epoch": 0.3603118822754228, + "grad_norm": 2.3821425437927246, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7076514959335327, + "num_tokens": 83969644.0, + "step": 3281 + }, + { + "epoch": 0.36042169997803647, + "grad_norm": 2.4140610694885254, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7104479074478149, + "num_tokens": 83991848.0, + "step": 3282 + }, + { + "epoch": 0.3605315176806501, + "grad_norm": 2.1799869537353516, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7197816371917725, + "num_tokens": 84016938.0, + "step": 3283 + }, + { + "epoch": 0.36064133538326376, + "grad_norm": 2.102332592010498, + "learning_rate": 1e-06, + "loss": 1.1179, + "mean_token_accuracy": 0.6657333374023438, + "num_tokens": 84046124.0, + "step": 3284 + }, + { + "epoch": 0.36075115308587746, + "grad_norm": 2.4456393718719482, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7126955389976501, + "num_tokens": 84068388.0, + "step": 3285 + }, + { + "epoch": 0.3608609707884911, + "grad_norm": 2.6027939319610596, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7083714008331299, + "num_tokens": 84087773.0, + "step": 3286 + }, + { + "epoch": 0.36097078849110475, + "grad_norm": 2.73903489112854, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7064259052276611, + "num_tokens": 84105064.0, + "step": 3287 + }, + { + "epoch": 0.3610806061937184, + "grad_norm": 1.932473063468933, + "learning_rate": 1e-06, + "loss": 1.0392, + "mean_token_accuracy": 0.6812301278114319, + "num_tokens": 84135947.0, + "step": 3288 + }, + { + "epoch": 0.3611904238963321, + "grad_norm": 2.142658233642578, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7106431722640991, + "num_tokens": 84160844.0, + "step": 3289 + }, + { + "epoch": 0.36130024159894575, + "grad_norm": 2.2562196254730225, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7283375859260559, + "num_tokens": 84183197.0, + "step": 3290 + }, + { + "epoch": 0.3614100593015594, + "grad_norm": 2.122911214828491, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7069064378738403, + "num_tokens": 84210648.0, + "step": 3291 + }, + { + "epoch": 0.3615198770041731, + "grad_norm": 2.256352424621582, + "learning_rate": 1e-06, + "loss": 1.0594, + "mean_token_accuracy": 0.680395245552063, + "num_tokens": 84237036.0, + "step": 3292 + }, + { + "epoch": 0.36162969470678674, + "grad_norm": 2.4464263916015625, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7103288173675537, + "num_tokens": 84262059.0, + "step": 3293 + }, + { + "epoch": 0.3617395124094004, + "grad_norm": 2.2013161182403564, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7013405561447144, + "num_tokens": 84289949.0, + "step": 3294 + }, + { + "epoch": 0.36184933011201403, + "grad_norm": 2.2597343921661377, + "learning_rate": 1e-06, + "loss": 1.11, + "mean_token_accuracy": 0.6708667278289795, + "num_tokens": 84319038.0, + "step": 3295 + }, + { + "epoch": 0.36195914781462774, + "grad_norm": 2.0560450553894043, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.7032757997512817, + "num_tokens": 84349313.0, + "step": 3296 + }, + { + "epoch": 0.3620689655172414, + "grad_norm": 1.9120277166366577, + "learning_rate": 1e-06, + "loss": 1.0604, + "mean_token_accuracy": 0.6815873384475708, + "num_tokens": 84385754.0, + "step": 3297 + }, + { + "epoch": 0.362178783219855, + "grad_norm": 2.1252248287200928, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7143900394439697, + "num_tokens": 84411793.0, + "step": 3298 + }, + { + "epoch": 0.36228860092246873, + "grad_norm": 1.8670659065246582, + "learning_rate": 1e-06, + "loss": 1.0288, + "mean_token_accuracy": 0.6902768611907959, + "num_tokens": 84445414.0, + "step": 3299 + }, + { + "epoch": 0.3623984186250824, + "grad_norm": 2.1826746463775635, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7007060647010803, + "num_tokens": 84471437.0, + "step": 3300 + }, + { + "epoch": 0.362508236327696, + "grad_norm": 2.206338405609131, + "learning_rate": 1e-06, + "loss": 1.0937, + "mean_token_accuracy": 0.6771517992019653, + "num_tokens": 84502277.0, + "step": 3301 + }, + { + "epoch": 0.36261805403030967, + "grad_norm": 2.594534158706665, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.725683331489563, + "num_tokens": 84522038.0, + "step": 3302 + }, + { + "epoch": 0.36272787173292337, + "grad_norm": 2.269648790359497, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7160672545433044, + "num_tokens": 84546748.0, + "step": 3303 + }, + { + "epoch": 0.362837689435537, + "grad_norm": 2.2878799438476562, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.7085007429122925, + "num_tokens": 84572405.0, + "step": 3304 + }, + { + "epoch": 0.36294750713815066, + "grad_norm": 2.233025074005127, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7179610729217529, + "num_tokens": 84595160.0, + "step": 3305 + }, + { + "epoch": 0.3630573248407643, + "grad_norm": 2.2850852012634277, + "learning_rate": 1e-06, + "loss": 1.0499, + "mean_token_accuracy": 0.6951881647109985, + "num_tokens": 84620236.0, + "step": 3306 + }, + { + "epoch": 0.363167142543378, + "grad_norm": 2.553567409515381, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.696932315826416, + "num_tokens": 84643845.0, + "step": 3307 + }, + { + "epoch": 0.36327696024599165, + "grad_norm": 2.464878559112549, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.7005306482315063, + "num_tokens": 84667408.0, + "step": 3308 + }, + { + "epoch": 0.3633867779486053, + "grad_norm": 2.5383553504943848, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6922010183334351, + "num_tokens": 84688069.0, + "step": 3309 + }, + { + "epoch": 0.363496595651219, + "grad_norm": 2.2880749702453613, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.6974530220031738, + "num_tokens": 84711700.0, + "step": 3310 + }, + { + "epoch": 0.36360641335383265, + "grad_norm": 2.2330844402313232, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.6989102363586426, + "num_tokens": 84736124.0, + "step": 3311 + }, + { + "epoch": 0.3637162310564463, + "grad_norm": 2.257110357284546, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7092333436012268, + "num_tokens": 84763488.0, + "step": 3312 + }, + { + "epoch": 0.36382604875905994, + "grad_norm": 2.16849946975708, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6999746561050415, + "num_tokens": 84791554.0, + "step": 3313 + }, + { + "epoch": 0.36393586646167364, + "grad_norm": 2.0175561904907227, + "learning_rate": 1e-06, + "loss": 1.0652, + "mean_token_accuracy": 0.689446210861206, + "num_tokens": 84823714.0, + "step": 3314 + }, + { + "epoch": 0.3640456841642873, + "grad_norm": 1.9633172750473022, + "learning_rate": 1e-06, + "loss": 1.0285, + "mean_token_accuracy": 0.6878620386123657, + "num_tokens": 84855214.0, + "step": 3315 + }, + { + "epoch": 0.36415550186690093, + "grad_norm": 2.3920953273773193, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7045351266860962, + "num_tokens": 84877556.0, + "step": 3316 + }, + { + "epoch": 0.3642653195695146, + "grad_norm": 2.318650722503662, + "learning_rate": 1e-06, + "loss": 1.0846, + "mean_token_accuracy": 0.6732564568519592, + "num_tokens": 84903337.0, + "step": 3317 + }, + { + "epoch": 0.3643751372721283, + "grad_norm": 2.0480923652648926, + "learning_rate": 1e-06, + "loss": 1.0736, + "mean_token_accuracy": 0.6838157176971436, + "num_tokens": 84932608.0, + "step": 3318 + }, + { + "epoch": 0.3644849549747419, + "grad_norm": 1.9554240703582764, + "learning_rate": 1e-06, + "loss": 1.056, + "mean_token_accuracy": 0.6811906099319458, + "num_tokens": 84963594.0, + "step": 3319 + }, + { + "epoch": 0.36459477267735557, + "grad_norm": 2.4328691959381104, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7264058589935303, + "num_tokens": 84987336.0, + "step": 3320 + }, + { + "epoch": 0.3647045903799693, + "grad_norm": 2.3520987033843994, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.7025536894798279, + "num_tokens": 85011841.0, + "step": 3321 + }, + { + "epoch": 0.3648144080825829, + "grad_norm": 1.955369472503662, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6969842314720154, + "num_tokens": 85044524.0, + "step": 3322 + }, + { + "epoch": 0.36492422578519657, + "grad_norm": 2.488463878631592, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7061217427253723, + "num_tokens": 85066208.0, + "step": 3323 + }, + { + "epoch": 0.3650340434878102, + "grad_norm": 2.2350168228149414, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7082033753395081, + "num_tokens": 85092127.0, + "step": 3324 + }, + { + "epoch": 0.3651438611904239, + "grad_norm": 2.127098321914673, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.710076093673706, + "num_tokens": 85119027.0, + "step": 3325 + }, + { + "epoch": 0.36525367889303756, + "grad_norm": 2.529238700866699, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.6954962611198425, + "num_tokens": 85140163.0, + "step": 3326 + }, + { + "epoch": 0.3653634965956512, + "grad_norm": 2.5025014877319336, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7060330510139465, + "num_tokens": 85161179.0, + "step": 3327 + }, + { + "epoch": 0.3654733142982649, + "grad_norm": 2.2261312007904053, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7105260491371155, + "num_tokens": 85188018.0, + "step": 3328 + }, + { + "epoch": 0.36558313200087855, + "grad_norm": 2.4014554023742676, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7075503468513489, + "num_tokens": 85211231.0, + "step": 3329 + }, + { + "epoch": 0.3656929497034922, + "grad_norm": 1.9729808568954468, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.707509458065033, + "num_tokens": 85242633.0, + "step": 3330 + }, + { + "epoch": 0.36580276740610584, + "grad_norm": 2.070152997970581, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6951221823692322, + "num_tokens": 85271248.0, + "step": 3331 + }, + { + "epoch": 0.36591258510871955, + "grad_norm": 2.534205198287964, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.709902286529541, + "num_tokens": 85290643.0, + "step": 3332 + }, + { + "epoch": 0.3660224028113332, + "grad_norm": 2.6887354850769043, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.6937719583511353, + "num_tokens": 85310464.0, + "step": 3333 + }, + { + "epoch": 0.36613222051394684, + "grad_norm": 1.9569882154464722, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.714169979095459, + "num_tokens": 85340187.0, + "step": 3334 + }, + { + "epoch": 0.3662420382165605, + "grad_norm": 2.081770896911621, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.718125581741333, + "num_tokens": 85367876.0, + "step": 3335 + }, + { + "epoch": 0.3663518559191742, + "grad_norm": 2.3489432334899902, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.699369490146637, + "num_tokens": 85392150.0, + "step": 3336 + }, + { + "epoch": 0.36646167362178783, + "grad_norm": 2.3057265281677246, + "learning_rate": 1e-06, + "loss": 1.0929, + "mean_token_accuracy": 0.6764715313911438, + "num_tokens": 85419722.0, + "step": 3337 + }, + { + "epoch": 0.3665714913244015, + "grad_norm": 2.069117784500122, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7026329040527344, + "num_tokens": 85447083.0, + "step": 3338 + }, + { + "epoch": 0.3666813090270152, + "grad_norm": 2.2216269969940186, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.694191575050354, + "num_tokens": 85471414.0, + "step": 3339 + }, + { + "epoch": 0.3667911267296288, + "grad_norm": 2.3789634704589844, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.702960729598999, + "num_tokens": 85494116.0, + "step": 3340 + }, + { + "epoch": 0.36690094443224247, + "grad_norm": 2.3047235012054443, + "learning_rate": 1e-06, + "loss": 1.0584, + "mean_token_accuracy": 0.6877678036689758, + "num_tokens": 85519358.0, + "step": 3341 + }, + { + "epoch": 0.3670107621348561, + "grad_norm": 2.0330734252929688, + "learning_rate": 1e-06, + "loss": 1.051, + "mean_token_accuracy": 0.6892344355583191, + "num_tokens": 85548837.0, + "step": 3342 + }, + { + "epoch": 0.3671205798374698, + "grad_norm": 2.8180863857269287, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7410826086997986, + "num_tokens": 85565192.0, + "step": 3343 + }, + { + "epoch": 0.36723039754008346, + "grad_norm": 2.4173498153686523, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.6934398412704468, + "num_tokens": 85587064.0, + "step": 3344 + }, + { + "epoch": 0.3673402152426971, + "grad_norm": 2.215773820877075, + "learning_rate": 1e-06, + "loss": 1.0965, + "mean_token_accuracy": 0.6702030897140503, + "num_tokens": 85615149.0, + "step": 3345 + }, + { + "epoch": 0.3674500329453108, + "grad_norm": 2.0422139167785645, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6978895664215088, + "num_tokens": 85643241.0, + "step": 3346 + }, + { + "epoch": 0.36755985064792446, + "grad_norm": 2.244422197341919, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7041366696357727, + "num_tokens": 85670051.0, + "step": 3347 + }, + { + "epoch": 0.3676696683505381, + "grad_norm": 2.4768760204315186, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7052780985832214, + "num_tokens": 85693089.0, + "step": 3348 + }, + { + "epoch": 0.36777948605315175, + "grad_norm": 2.5655598640441895, + "learning_rate": 1e-06, + "loss": 1.0223, + "mean_token_accuracy": 0.691206157207489, + "num_tokens": 85713801.0, + "step": 3349 + }, + { + "epoch": 0.36788930375576545, + "grad_norm": 2.4119527339935303, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7042464017868042, + "num_tokens": 85735563.0, + "step": 3350 + }, + { + "epoch": 0.3679991214583791, + "grad_norm": 2.3211591243743896, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.6974666714668274, + "num_tokens": 85759631.0, + "step": 3351 + }, + { + "epoch": 0.36810893916099274, + "grad_norm": 2.280738592147827, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7066053152084351, + "num_tokens": 85784471.0, + "step": 3352 + }, + { + "epoch": 0.3682187568636064, + "grad_norm": 2.321284055709839, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7100980281829834, + "num_tokens": 85808221.0, + "step": 3353 + }, + { + "epoch": 0.3683285745662201, + "grad_norm": 2.073505401611328, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6912756562232971, + "num_tokens": 85837624.0, + "step": 3354 + }, + { + "epoch": 0.36843839226883374, + "grad_norm": 2.311810255050659, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7020030617713928, + "num_tokens": 85860605.0, + "step": 3355 + }, + { + "epoch": 0.3685482099714474, + "grad_norm": 2.1161227226257324, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7080735564231873, + "num_tokens": 85890134.0, + "step": 3356 + }, + { + "epoch": 0.3686580276740611, + "grad_norm": 2.2024965286254883, + "learning_rate": 1e-06, + "loss": 1.1138, + "mean_token_accuracy": 0.6707265377044678, + "num_tokens": 85919495.0, + "step": 3357 + }, + { + "epoch": 0.36876784537667473, + "grad_norm": 2.100679397583008, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7119258642196655, + "num_tokens": 85946478.0, + "step": 3358 + }, + { + "epoch": 0.3688776630792884, + "grad_norm": 2.4561078548431396, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6855672001838684, + "num_tokens": 85967422.0, + "step": 3359 + }, + { + "epoch": 0.368987480781902, + "grad_norm": 2.3312618732452393, + "learning_rate": 1e-06, + "loss": 1.0238, + "mean_token_accuracy": 0.6865999102592468, + "num_tokens": 85991969.0, + "step": 3360 + }, + { + "epoch": 0.3690972984845157, + "grad_norm": 2.4606223106384277, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.7004603147506714, + "num_tokens": 86012679.0, + "step": 3361 + }, + { + "epoch": 0.36920711618712937, + "grad_norm": 2.0007994174957275, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.6998085975646973, + "num_tokens": 86041942.0, + "step": 3362 + }, + { + "epoch": 0.369316933889743, + "grad_norm": 2.257112979888916, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6859581470489502, + "num_tokens": 86069462.0, + "step": 3363 + }, + { + "epoch": 0.36942675159235666, + "grad_norm": 2.513887405395508, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6919094324111938, + "num_tokens": 86090415.0, + "step": 3364 + }, + { + "epoch": 0.36953656929497036, + "grad_norm": 2.4420855045318604, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7029476165771484, + "num_tokens": 86114280.0, + "step": 3365 + }, + { + "epoch": 0.369646386997584, + "grad_norm": 2.3252949714660645, + "learning_rate": 1e-06, + "loss": 1.0876, + "mean_token_accuracy": 0.6767888069152832, + "num_tokens": 86139320.0, + "step": 3366 + }, + { + "epoch": 0.36975620470019765, + "grad_norm": 2.146425724029541, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7194659113883972, + "num_tokens": 86166370.0, + "step": 3367 + }, + { + "epoch": 0.36986602240281136, + "grad_norm": 2.1090505123138428, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6944506764411926, + "num_tokens": 86197394.0, + "step": 3368 + }, + { + "epoch": 0.369975840105425, + "grad_norm": 2.3360209465026855, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7010655403137207, + "num_tokens": 86220799.0, + "step": 3369 + }, + { + "epoch": 0.37008565780803865, + "grad_norm": 2.1647861003875732, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7000632286071777, + "num_tokens": 86246975.0, + "step": 3370 + }, + { + "epoch": 0.3701954755106523, + "grad_norm": 2.259075880050659, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7173027396202087, + "num_tokens": 86274510.0, + "step": 3371 + }, + { + "epoch": 0.370305293213266, + "grad_norm": 2.213437557220459, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.7004991769790649, + "num_tokens": 86300955.0, + "step": 3372 + }, + { + "epoch": 0.37041511091587964, + "grad_norm": 2.126929998397827, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.6893806457519531, + "num_tokens": 86330605.0, + "step": 3373 + }, + { + "epoch": 0.3705249286184933, + "grad_norm": 2.1518545150756836, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.6992454528808594, + "num_tokens": 86358249.0, + "step": 3374 + }, + { + "epoch": 0.370634746321107, + "grad_norm": 2.221694231033325, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.7007765769958496, + "num_tokens": 86383661.0, + "step": 3375 + }, + { + "epoch": 0.37074456402372064, + "grad_norm": 2.471247434616089, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.7011731266975403, + "num_tokens": 86404417.0, + "step": 3376 + }, + { + "epoch": 0.3708543817263343, + "grad_norm": 2.0568461418151855, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7187901735305786, + "num_tokens": 86432281.0, + "step": 3377 + }, + { + "epoch": 0.3709641994289479, + "grad_norm": 2.376192092895508, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7256274223327637, + "num_tokens": 86452904.0, + "step": 3378 + }, + { + "epoch": 0.37107401713156163, + "grad_norm": 2.477069854736328, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7166978120803833, + "num_tokens": 86474085.0, + "step": 3379 + }, + { + "epoch": 0.3711838348341753, + "grad_norm": 1.9778320789337158, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6936007738113403, + "num_tokens": 86504556.0, + "step": 3380 + }, + { + "epoch": 0.3712936525367889, + "grad_norm": 2.242398977279663, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.7005918025970459, + "num_tokens": 86529939.0, + "step": 3381 + }, + { + "epoch": 0.37140347023940257, + "grad_norm": 2.336547613143921, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7073181867599487, + "num_tokens": 86553356.0, + "step": 3382 + }, + { + "epoch": 0.37151328794201627, + "grad_norm": 2.0643017292022705, + "learning_rate": 1e-06, + "loss": 0.7588, + "mean_token_accuracy": 0.7593939304351807, + "num_tokens": 86577250.0, + "step": 3383 + }, + { + "epoch": 0.3716231056446299, + "grad_norm": 2.1858198642730713, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6986223459243774, + "num_tokens": 86602916.0, + "step": 3384 + }, + { + "epoch": 0.37173292334724356, + "grad_norm": 2.4600369930267334, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6959330439567566, + "num_tokens": 86626951.0, + "step": 3385 + }, + { + "epoch": 0.37184274104985726, + "grad_norm": 1.9603826999664307, + "learning_rate": 1e-06, + "loss": 1.0975, + "mean_token_accuracy": 0.6786723136901855, + "num_tokens": 86663090.0, + "step": 3386 + }, + { + "epoch": 0.3719525587524709, + "grad_norm": 2.4769256114959717, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7071607112884521, + "num_tokens": 86687184.0, + "step": 3387 + }, + { + "epoch": 0.37206237645508455, + "grad_norm": 2.6178152561187744, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.7015594244003296, + "num_tokens": 86706308.0, + "step": 3388 + }, + { + "epoch": 0.3721721941576982, + "grad_norm": 2.494547128677368, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7033604979515076, + "num_tokens": 86726867.0, + "step": 3389 + }, + { + "epoch": 0.3722820118603119, + "grad_norm": 2.2039341926574707, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7066138982772827, + "num_tokens": 86751862.0, + "step": 3390 + }, + { + "epoch": 0.37239182956292555, + "grad_norm": 2.3886566162109375, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7103887796401978, + "num_tokens": 86774289.0, + "step": 3391 + }, + { + "epoch": 0.3725016472655392, + "grad_norm": 2.160444974899292, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6913098096847534, + "num_tokens": 86803739.0, + "step": 3392 + }, + { + "epoch": 0.37261146496815284, + "grad_norm": 2.3246123790740967, + "learning_rate": 1e-06, + "loss": 1.0948, + "mean_token_accuracy": 0.6784851551055908, + "num_tokens": 86827241.0, + "step": 3393 + }, + { + "epoch": 0.37272128267076654, + "grad_norm": 2.3865411281585693, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.703160285949707, + "num_tokens": 86847956.0, + "step": 3394 + }, + { + "epoch": 0.3728311003733802, + "grad_norm": 2.0383622646331787, + "learning_rate": 1e-06, + "loss": 1.0928, + "mean_token_accuracy": 0.6753938794136047, + "num_tokens": 86878214.0, + "step": 3395 + }, + { + "epoch": 0.37294091807599383, + "grad_norm": 2.2223565578460693, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6983576416969299, + "num_tokens": 86903782.0, + "step": 3396 + }, + { + "epoch": 0.37305073577860753, + "grad_norm": 2.001816749572754, + "learning_rate": 1e-06, + "loss": 1.0244, + "mean_token_accuracy": 0.6920322179794312, + "num_tokens": 86933286.0, + "step": 3397 + }, + { + "epoch": 0.3731605534812212, + "grad_norm": 2.1014444828033447, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7113136053085327, + "num_tokens": 86959202.0, + "step": 3398 + }, + { + "epoch": 0.3732703711838348, + "grad_norm": 2.2578370571136475, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.7076982855796814, + "num_tokens": 86983830.0, + "step": 3399 + }, + { + "epoch": 0.37338018888644847, + "grad_norm": 2.143278121948242, + "learning_rate": 1e-06, + "loss": 1.0372, + "mean_token_accuracy": 0.6888443827629089, + "num_tokens": 87013433.0, + "step": 3400 + }, + { + "epoch": 0.3734900065890622, + "grad_norm": 2.0151455402374268, + "learning_rate": 1e-06, + "loss": 1.032, + "mean_token_accuracy": 0.7014425992965698, + "num_tokens": 87045098.0, + "step": 3401 + }, + { + "epoch": 0.3735998242916758, + "grad_norm": 2.2361416816711426, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7181442379951477, + "num_tokens": 87070609.0, + "step": 3402 + }, + { + "epoch": 0.37370964199428947, + "grad_norm": 2.1480607986450195, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7217540144920349, + "num_tokens": 87094974.0, + "step": 3403 + }, + { + "epoch": 0.37381945969690317, + "grad_norm": 2.026329755783081, + "learning_rate": 1e-06, + "loss": 1.007, + "mean_token_accuracy": 0.697094738483429, + "num_tokens": 87125035.0, + "step": 3404 + }, + { + "epoch": 0.3739292773995168, + "grad_norm": 2.170217275619507, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7198174595832825, + "num_tokens": 87153011.0, + "step": 3405 + }, + { + "epoch": 0.37403909510213046, + "grad_norm": 2.429253578186035, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7051354050636292, + "num_tokens": 87175066.0, + "step": 3406 + }, + { + "epoch": 0.3741489128047441, + "grad_norm": 2.1822092533111572, + "learning_rate": 1e-06, + "loss": 1.0938, + "mean_token_accuracy": 0.6750043630599976, + "num_tokens": 87205669.0, + "step": 3407 + }, + { + "epoch": 0.3742587305073578, + "grad_norm": 2.029784679412842, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.7005590200424194, + "num_tokens": 87236036.0, + "step": 3408 + }, + { + "epoch": 0.37436854820997145, + "grad_norm": 2.081159830093384, + "learning_rate": 1e-06, + "loss": 1.0814, + "mean_token_accuracy": 0.6768223643302917, + "num_tokens": 87268334.0, + "step": 3409 + }, + { + "epoch": 0.3744783659125851, + "grad_norm": 1.9588316679000854, + "learning_rate": 1e-06, + "loss": 1.0217, + "mean_token_accuracy": 0.6925580501556396, + "num_tokens": 87302424.0, + "step": 3410 + }, + { + "epoch": 0.37458818361519874, + "grad_norm": 2.52958607673645, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.701055109500885, + "num_tokens": 87322989.0, + "step": 3411 + }, + { + "epoch": 0.37469800131781245, + "grad_norm": 2.3104212284088135, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6990007758140564, + "num_tokens": 87347973.0, + "step": 3412 + }, + { + "epoch": 0.3748078190204261, + "grad_norm": 2.0304126739501953, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7204382419586182, + "num_tokens": 87377179.0, + "step": 3413 + }, + { + "epoch": 0.37491763672303974, + "grad_norm": 1.9988762140274048, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6998769044876099, + "num_tokens": 87408187.0, + "step": 3414 + }, + { + "epoch": 0.37502745442565344, + "grad_norm": 2.3191332817077637, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7103760242462158, + "num_tokens": 87431756.0, + "step": 3415 + }, + { + "epoch": 0.3751372721282671, + "grad_norm": 2.7814128398895264, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.7008881568908691, + "num_tokens": 87452922.0, + "step": 3416 + }, + { + "epoch": 0.37524708983088073, + "grad_norm": 2.2637596130371094, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7179586887359619, + "num_tokens": 87477083.0, + "step": 3417 + }, + { + "epoch": 0.3753569075334944, + "grad_norm": 2.435710906982422, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6954156756401062, + "num_tokens": 87501831.0, + "step": 3418 + }, + { + "epoch": 0.3754667252361081, + "grad_norm": 2.4289422035217285, + "learning_rate": 1e-06, + "loss": 1.0645, + "mean_token_accuracy": 0.6883988976478577, + "num_tokens": 87526531.0, + "step": 3419 + }, + { + "epoch": 0.3755765429387217, + "grad_norm": 2.296262502670288, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.7014434337615967, + "num_tokens": 87550134.0, + "step": 3420 + }, + { + "epoch": 0.37568636064133537, + "grad_norm": 2.1783218383789062, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.6873223781585693, + "num_tokens": 87575827.0, + "step": 3421 + }, + { + "epoch": 0.37579617834394907, + "grad_norm": 2.361457586288452, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7151813507080078, + "num_tokens": 87600017.0, + "step": 3422 + }, + { + "epoch": 0.3759059960465627, + "grad_norm": 2.061671257019043, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7114100456237793, + "num_tokens": 87628754.0, + "step": 3423 + }, + { + "epoch": 0.37601581374917636, + "grad_norm": 1.953343391418457, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.7012108564376831, + "num_tokens": 87658210.0, + "step": 3424 + }, + { + "epoch": 0.37612563145179, + "grad_norm": 2.0183162689208984, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.6927720904350281, + "num_tokens": 87688672.0, + "step": 3425 + }, + { + "epoch": 0.3762354491544037, + "grad_norm": 2.2148261070251465, + "learning_rate": 1e-06, + "loss": 1.0459, + "mean_token_accuracy": 0.6874687671661377, + "num_tokens": 87715367.0, + "step": 3426 + }, + { + "epoch": 0.37634526685701736, + "grad_norm": 2.299649238586426, + "learning_rate": 1e-06, + "loss": 1.0286, + "mean_token_accuracy": 0.6960898637771606, + "num_tokens": 87741265.0, + "step": 3427 + }, + { + "epoch": 0.376455084559631, + "grad_norm": 2.164278507232666, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.7006861567497253, + "num_tokens": 87768505.0, + "step": 3428 + }, + { + "epoch": 0.37656490226224465, + "grad_norm": 2.4089674949645996, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7203987836837769, + "num_tokens": 87790068.0, + "step": 3429 + }, + { + "epoch": 0.37667471996485835, + "grad_norm": 2.0968382358551025, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.6843818426132202, + "num_tokens": 87820110.0, + "step": 3430 + }, + { + "epoch": 0.376784537667472, + "grad_norm": 2.011951446533203, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.7123485803604126, + "num_tokens": 87849967.0, + "step": 3431 + }, + { + "epoch": 0.37689435537008564, + "grad_norm": 2.585885524749756, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7084155082702637, + "num_tokens": 87869866.0, + "step": 3432 + }, + { + "epoch": 0.37700417307269934, + "grad_norm": 2.1131556034088135, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.6904738545417786, + "num_tokens": 87898408.0, + "step": 3433 + }, + { + "epoch": 0.377113990775313, + "grad_norm": 2.291736364364624, + "learning_rate": 1e-06, + "loss": 1.0217, + "mean_token_accuracy": 0.6940968036651611, + "num_tokens": 87922044.0, + "step": 3434 + }, + { + "epoch": 0.37722380847792664, + "grad_norm": 2.452139139175415, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7096028327941895, + "num_tokens": 87943514.0, + "step": 3435 + }, + { + "epoch": 0.3773336261805403, + "grad_norm": 2.3346142768859863, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.6938886642456055, + "num_tokens": 87966220.0, + "step": 3436 + }, + { + "epoch": 0.377443443883154, + "grad_norm": 2.297839641571045, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7148208022117615, + "num_tokens": 87990271.0, + "step": 3437 + }, + { + "epoch": 0.37755326158576763, + "grad_norm": 2.0365943908691406, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7085906267166138, + "num_tokens": 88020733.0, + "step": 3438 + }, + { + "epoch": 0.3776630792883813, + "grad_norm": 2.436033248901367, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.718590259552002, + "num_tokens": 88043529.0, + "step": 3439 + }, + { + "epoch": 0.3777728969909949, + "grad_norm": 1.880038857460022, + "learning_rate": 1e-06, + "loss": 1.0319, + "mean_token_accuracy": 0.6886873245239258, + "num_tokens": 88075992.0, + "step": 3440 + }, + { + "epoch": 0.3778827146936086, + "grad_norm": 2.086747884750366, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6941910982131958, + "num_tokens": 88104257.0, + "step": 3441 + }, + { + "epoch": 0.37799253239622227, + "grad_norm": 2.403947591781616, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6919853091239929, + "num_tokens": 88127935.0, + "step": 3442 + }, + { + "epoch": 0.3781023500988359, + "grad_norm": 2.2730371952056885, + "learning_rate": 1e-06, + "loss": 1.0654, + "mean_token_accuracy": 0.6895796060562134, + "num_tokens": 88153818.0, + "step": 3443 + }, + { + "epoch": 0.3782121678014496, + "grad_norm": 2.3619396686553955, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6962672472000122, + "num_tokens": 88176915.0, + "step": 3444 + }, + { + "epoch": 0.37832198550406326, + "grad_norm": 2.1047112941741943, + "learning_rate": 1e-06, + "loss": 1.012, + "mean_token_accuracy": 0.6921414136886597, + "num_tokens": 88203070.0, + "step": 3445 + }, + { + "epoch": 0.3784318032066769, + "grad_norm": 2.1845691204071045, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6797412633895874, + "num_tokens": 88230977.0, + "step": 3446 + }, + { + "epoch": 0.37854162090929055, + "grad_norm": 2.309771776199341, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.6968021392822266, + "num_tokens": 88254110.0, + "step": 3447 + }, + { + "epoch": 0.37865143861190426, + "grad_norm": 2.246907949447632, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6956825852394104, + "num_tokens": 88281056.0, + "step": 3448 + }, + { + "epoch": 0.3787612563145179, + "grad_norm": 2.4603631496429443, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.741458535194397, + "num_tokens": 88304083.0, + "step": 3449 + }, + { + "epoch": 0.37887107401713155, + "grad_norm": 2.341451406478882, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.702991247177124, + "num_tokens": 88327143.0, + "step": 3450 + }, + { + "epoch": 0.37898089171974525, + "grad_norm": 2.238539695739746, + "learning_rate": 1e-06, + "loss": 1.0752, + "mean_token_accuracy": 0.6807167530059814, + "num_tokens": 88351895.0, + "step": 3451 + }, + { + "epoch": 0.3790907094223589, + "grad_norm": 1.9991176128387451, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7045825719833374, + "num_tokens": 88382062.0, + "step": 3452 + }, + { + "epoch": 0.37920052712497254, + "grad_norm": 2.0140798091888428, + "learning_rate": 1e-06, + "loss": 1.0911, + "mean_token_accuracy": 0.6757621765136719, + "num_tokens": 88412696.0, + "step": 3453 + }, + { + "epoch": 0.3793103448275862, + "grad_norm": 2.645862579345703, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7109823226928711, + "num_tokens": 88434255.0, + "step": 3454 + }, + { + "epoch": 0.3794201625301999, + "grad_norm": 2.4253737926483154, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7133928537368774, + "num_tokens": 88456120.0, + "step": 3455 + }, + { + "epoch": 0.37952998023281354, + "grad_norm": 2.2605180740356445, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7248380184173584, + "num_tokens": 88480062.0, + "step": 3456 + }, + { + "epoch": 0.3796397979354272, + "grad_norm": 2.5435454845428467, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7112745046615601, + "num_tokens": 88501018.0, + "step": 3457 + }, + { + "epoch": 0.3797496156380408, + "grad_norm": 2.3400185108184814, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7021656036376953, + "num_tokens": 88524784.0, + "step": 3458 + }, + { + "epoch": 0.37985943334065453, + "grad_norm": 2.592353105545044, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.7063405513763428, + "num_tokens": 88546882.0, + "step": 3459 + }, + { + "epoch": 0.3799692510432682, + "grad_norm": 2.5097951889038086, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7131935954093933, + "num_tokens": 88569361.0, + "step": 3460 + }, + { + "epoch": 0.3800790687458818, + "grad_norm": 2.3237357139587402, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6921998262405396, + "num_tokens": 88597027.0, + "step": 3461 + }, + { + "epoch": 0.3801888864484955, + "grad_norm": 2.017336845397949, + "learning_rate": 1e-06, + "loss": 1.0654, + "mean_token_accuracy": 0.6802768111228943, + "num_tokens": 88629200.0, + "step": 3462 + }, + { + "epoch": 0.38029870415110917, + "grad_norm": 2.401096820831299, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7240603566169739, + "num_tokens": 88653437.0, + "step": 3463 + }, + { + "epoch": 0.3804085218537228, + "grad_norm": 2.6765213012695312, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7056003212928772, + "num_tokens": 88673841.0, + "step": 3464 + }, + { + "epoch": 0.38051833955633646, + "grad_norm": 2.2659072875976562, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.726227879524231, + "num_tokens": 88700466.0, + "step": 3465 + }, + { + "epoch": 0.38062815725895016, + "grad_norm": 2.3554465770721436, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7188282012939453, + "num_tokens": 88725044.0, + "step": 3466 + }, + { + "epoch": 0.3807379749615638, + "grad_norm": 2.0605390071868896, + "learning_rate": 1e-06, + "loss": 1.0252, + "mean_token_accuracy": 0.6842021942138672, + "num_tokens": 88755972.0, + "step": 3467 + }, + { + "epoch": 0.38084779266417745, + "grad_norm": 2.447361946105957, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.7049893140792847, + "num_tokens": 88779377.0, + "step": 3468 + }, + { + "epoch": 0.3809576103667911, + "grad_norm": 2.078284502029419, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7243909239768982, + "num_tokens": 88806164.0, + "step": 3469 + }, + { + "epoch": 0.3810674280694048, + "grad_norm": 1.9445465803146362, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7395931482315063, + "num_tokens": 88832955.0, + "step": 3470 + }, + { + "epoch": 0.38117724577201845, + "grad_norm": 2.2920820713043213, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7140834927558899, + "num_tokens": 88856552.0, + "step": 3471 + }, + { + "epoch": 0.3812870634746321, + "grad_norm": 2.315526008605957, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.6939349174499512, + "num_tokens": 88881021.0, + "step": 3472 + }, + { + "epoch": 0.3813968811772458, + "grad_norm": 2.2417726516723633, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7186139822006226, + "num_tokens": 88904010.0, + "step": 3473 + }, + { + "epoch": 0.38150669887985944, + "grad_norm": 2.39229416847229, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.6988753080368042, + "num_tokens": 88928516.0, + "step": 3474 + }, + { + "epoch": 0.3816165165824731, + "grad_norm": 2.0921759605407715, + "learning_rate": 1e-06, + "loss": 1.0659, + "mean_token_accuracy": 0.6860042810440063, + "num_tokens": 88957587.0, + "step": 3475 + }, + { + "epoch": 0.38172633428508673, + "grad_norm": 2.037487506866455, + "learning_rate": 1e-06, + "loss": 1.0538, + "mean_token_accuracy": 0.6887670159339905, + "num_tokens": 88986767.0, + "step": 3476 + }, + { + "epoch": 0.38183615198770043, + "grad_norm": 2.4350571632385254, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.7023879289627075, + "num_tokens": 89008828.0, + "step": 3477 + }, + { + "epoch": 0.3819459696903141, + "grad_norm": 2.148155689239502, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7221804857254028, + "num_tokens": 89035631.0, + "step": 3478 + }, + { + "epoch": 0.3820557873929277, + "grad_norm": 2.3605384826660156, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7238632440567017, + "num_tokens": 89057176.0, + "step": 3479 + }, + { + "epoch": 0.3821656050955414, + "grad_norm": 2.20352840423584, + "learning_rate": 1e-06, + "loss": 1.0637, + "mean_token_accuracy": 0.6905278563499451, + "num_tokens": 89083028.0, + "step": 3480 + }, + { + "epoch": 0.3822754227981551, + "grad_norm": 2.0967469215393066, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7043014764785767, + "num_tokens": 89111023.0, + "step": 3481 + }, + { + "epoch": 0.3823852405007687, + "grad_norm": 2.3745200634002686, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7051019668579102, + "num_tokens": 89132874.0, + "step": 3482 + }, + { + "epoch": 0.38249505820338237, + "grad_norm": 2.4169154167175293, + "learning_rate": 1e-06, + "loss": 1.0711, + "mean_token_accuracy": 0.680354118347168, + "num_tokens": 89157595.0, + "step": 3483 + }, + { + "epoch": 0.38260487590599607, + "grad_norm": 2.0611865520477295, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7050142884254456, + "num_tokens": 89185171.0, + "step": 3484 + }, + { + "epoch": 0.3827146936086097, + "grad_norm": 2.300441265106201, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7230792045593262, + "num_tokens": 89208050.0, + "step": 3485 + }, + { + "epoch": 0.38282451131122336, + "grad_norm": 2.0712599754333496, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7014822959899902, + "num_tokens": 89236592.0, + "step": 3486 + }, + { + "epoch": 0.382934329013837, + "grad_norm": 2.0661323070526123, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7155558466911316, + "num_tokens": 89264732.0, + "step": 3487 + }, + { + "epoch": 0.3830441467164507, + "grad_norm": 1.9915300607681274, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.6950030326843262, + "num_tokens": 89296843.0, + "step": 3488 + }, + { + "epoch": 0.38315396441906435, + "grad_norm": 2.2650413513183594, + "learning_rate": 1e-06, + "loss": 1.0029, + "mean_token_accuracy": 0.6991224884986877, + "num_tokens": 89323397.0, + "step": 3489 + }, + { + "epoch": 0.383263782121678, + "grad_norm": 2.1685850620269775, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7052390575408936, + "num_tokens": 89349147.0, + "step": 3490 + }, + { + "epoch": 0.3833735998242917, + "grad_norm": 2.3250656127929688, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7245489358901978, + "num_tokens": 89371292.0, + "step": 3491 + }, + { + "epoch": 0.38348341752690535, + "grad_norm": 2.067066192626953, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7285982370376587, + "num_tokens": 89400274.0, + "step": 3492 + }, + { + "epoch": 0.383593235229519, + "grad_norm": 2.250953197479248, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7191085815429688, + "num_tokens": 89424746.0, + "step": 3493 + }, + { + "epoch": 0.38370305293213264, + "grad_norm": 2.0857508182525635, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.7074625492095947, + "num_tokens": 89454445.0, + "step": 3494 + }, + { + "epoch": 0.38381287063474634, + "grad_norm": 2.036393404006958, + "learning_rate": 1e-06, + "loss": 1.0624, + "mean_token_accuracy": 0.6772199869155884, + "num_tokens": 89485141.0, + "step": 3495 + }, + { + "epoch": 0.38392268833736, + "grad_norm": 2.1552460193634033, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6930525302886963, + "num_tokens": 89510666.0, + "step": 3496 + }, + { + "epoch": 0.38403250603997363, + "grad_norm": 2.3500938415527344, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.7035219669342041, + "num_tokens": 89533463.0, + "step": 3497 + }, + { + "epoch": 0.38414232374258733, + "grad_norm": 2.42331862449646, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7293581962585449, + "num_tokens": 89555584.0, + "step": 3498 + }, + { + "epoch": 0.384252141445201, + "grad_norm": 2.6354169845581055, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.710674524307251, + "num_tokens": 89574501.0, + "step": 3499 + }, + { + "epoch": 0.3843619591478146, + "grad_norm": 2.244262456893921, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7024070024490356, + "num_tokens": 89600416.0, + "step": 3500 + }, + { + "epoch": 0.38447177685042827, + "grad_norm": 2.106200695037842, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.7130498886108398, + "num_tokens": 89628749.0, + "step": 3501 + }, + { + "epoch": 0.38458159455304197, + "grad_norm": 2.428870916366577, + "learning_rate": 1e-06, + "loss": 1.0636, + "mean_token_accuracy": 0.6813790798187256, + "num_tokens": 89652306.0, + "step": 3502 + }, + { + "epoch": 0.3846914122556556, + "grad_norm": 2.1484029293060303, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7176358699798584, + "num_tokens": 89678011.0, + "step": 3503 + }, + { + "epoch": 0.38480122995826926, + "grad_norm": 1.9306894540786743, + "learning_rate": 1e-06, + "loss": 1.0945, + "mean_token_accuracy": 0.6701663732528687, + "num_tokens": 89714411.0, + "step": 3504 + }, + { + "epoch": 0.3849110476608829, + "grad_norm": 2.2948617935180664, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7209081649780273, + "num_tokens": 89737593.0, + "step": 3505 + }, + { + "epoch": 0.3850208653634966, + "grad_norm": 2.2934556007385254, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7075262069702148, + "num_tokens": 89762760.0, + "step": 3506 + }, + { + "epoch": 0.38513068306611026, + "grad_norm": 2.552290678024292, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7200935482978821, + "num_tokens": 89784514.0, + "step": 3507 + }, + { + "epoch": 0.3852405007687239, + "grad_norm": 2.055102825164795, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7134333848953247, + "num_tokens": 89813124.0, + "step": 3508 + }, + { + "epoch": 0.3853503184713376, + "grad_norm": 2.6006407737731934, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7444247007369995, + "num_tokens": 89833638.0, + "step": 3509 + }, + { + "epoch": 0.38546013617395125, + "grad_norm": 2.5083956718444824, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7115962505340576, + "num_tokens": 89853060.0, + "step": 3510 + }, + { + "epoch": 0.3855699538765649, + "grad_norm": 2.367403507232666, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7043768167495728, + "num_tokens": 89875623.0, + "step": 3511 + }, + { + "epoch": 0.38567977157917854, + "grad_norm": 2.362814426422119, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7093349099159241, + "num_tokens": 89898029.0, + "step": 3512 + }, + { + "epoch": 0.38578958928179224, + "grad_norm": 2.3117713928222656, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7098644971847534, + "num_tokens": 89920128.0, + "step": 3513 + }, + { + "epoch": 0.3858994069844059, + "grad_norm": 2.132671594619751, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.695099949836731, + "num_tokens": 89946539.0, + "step": 3514 + }, + { + "epoch": 0.38600922468701954, + "grad_norm": 2.2543816566467285, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7073651552200317, + "num_tokens": 89974650.0, + "step": 3515 + }, + { + "epoch": 0.3861190423896332, + "grad_norm": 2.3273165225982666, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7106094360351562, + "num_tokens": 89997771.0, + "step": 3516 + }, + { + "epoch": 0.3862288600922469, + "grad_norm": 2.656499147415161, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7163621783256531, + "num_tokens": 90016799.0, + "step": 3517 + }, + { + "epoch": 0.38633867779486053, + "grad_norm": 2.0734753608703613, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6936517953872681, + "num_tokens": 90046512.0, + "step": 3518 + }, + { + "epoch": 0.3864484954974742, + "grad_norm": 2.5783498287200928, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7127504348754883, + "num_tokens": 90064716.0, + "step": 3519 + }, + { + "epoch": 0.3865583132000879, + "grad_norm": 2.4590847492218018, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7335934638977051, + "num_tokens": 90085518.0, + "step": 3520 + }, + { + "epoch": 0.3866681309027015, + "grad_norm": 2.198406219482422, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7203407287597656, + "num_tokens": 90119611.0, + "step": 3521 + }, + { + "epoch": 0.38677794860531517, + "grad_norm": 2.445981740951538, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.6972277164459229, + "num_tokens": 90142565.0, + "step": 3522 + }, + { + "epoch": 0.3868877663079288, + "grad_norm": 2.274430274963379, + "learning_rate": 1e-06, + "loss": 1.0689, + "mean_token_accuracy": 0.6789602041244507, + "num_tokens": 90168719.0, + "step": 3523 + }, + { + "epoch": 0.3869975840105425, + "grad_norm": 2.105550765991211, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7077186107635498, + "num_tokens": 90199393.0, + "step": 3524 + }, + { + "epoch": 0.38710740171315616, + "grad_norm": 2.2680506706237793, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7070595026016235, + "num_tokens": 90222466.0, + "step": 3525 + }, + { + "epoch": 0.3872172194157698, + "grad_norm": 1.9302706718444824, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.7006466388702393, + "num_tokens": 90253716.0, + "step": 3526 + }, + { + "epoch": 0.3873270371183835, + "grad_norm": 1.9963728189468384, + "learning_rate": 1e-06, + "loss": 1.1388, + "mean_token_accuracy": 0.6603199243545532, + "num_tokens": 90284727.0, + "step": 3527 + }, + { + "epoch": 0.38743685482099716, + "grad_norm": 2.5656824111938477, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6891986131668091, + "num_tokens": 90305858.0, + "step": 3528 + }, + { + "epoch": 0.3875466725236108, + "grad_norm": 2.4764153957366943, + "learning_rate": 1e-06, + "loss": 1.0433, + "mean_token_accuracy": 0.6859539747238159, + "num_tokens": 90328390.0, + "step": 3529 + }, + { + "epoch": 0.38765649022622445, + "grad_norm": 2.3799026012420654, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7112096548080444, + "num_tokens": 90352812.0, + "step": 3530 + }, + { + "epoch": 0.38776630792883815, + "grad_norm": 2.2229762077331543, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7219316363334656, + "num_tokens": 90378980.0, + "step": 3531 + }, + { + "epoch": 0.3878761256314518, + "grad_norm": 2.485604763031006, + "learning_rate": 1e-06, + "loss": 1.0367, + "mean_token_accuracy": 0.6878149509429932, + "num_tokens": 90402270.0, + "step": 3532 + }, + { + "epoch": 0.38798594333406544, + "grad_norm": 2.021167039871216, + "learning_rate": 1e-06, + "loss": 1.0652, + "mean_token_accuracy": 0.6852264404296875, + "num_tokens": 90433336.0, + "step": 3533 + }, + { + "epoch": 0.3880957610366791, + "grad_norm": 2.3849005699157715, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7247136831283569, + "num_tokens": 90455464.0, + "step": 3534 + }, + { + "epoch": 0.3882055787392928, + "grad_norm": 2.4539740085601807, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7269081473350525, + "num_tokens": 90476352.0, + "step": 3535 + }, + { + "epoch": 0.38831539644190644, + "grad_norm": 2.2399818897247314, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7019773125648499, + "num_tokens": 90503162.0, + "step": 3536 + }, + { + "epoch": 0.3884252141445201, + "grad_norm": 2.151144504547119, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.7004350423812866, + "num_tokens": 90529339.0, + "step": 3537 + }, + { + "epoch": 0.3885350318471338, + "grad_norm": 2.425339698791504, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7020261287689209, + "num_tokens": 90551612.0, + "step": 3538 + }, + { + "epoch": 0.38864484954974743, + "grad_norm": 2.262519598007202, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6971696615219116, + "num_tokens": 90576759.0, + "step": 3539 + }, + { + "epoch": 0.3887546672523611, + "grad_norm": 1.8118031024932861, + "learning_rate": 1e-06, + "loss": 1.1042, + "mean_token_accuracy": 0.6723151803016663, + "num_tokens": 90616502.0, + "step": 3540 + }, + { + "epoch": 0.3888644849549747, + "grad_norm": 2.1481118202209473, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.7099519968032837, + "num_tokens": 90643357.0, + "step": 3541 + }, + { + "epoch": 0.3889743026575884, + "grad_norm": 1.8311972618103027, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.6908115148544312, + "num_tokens": 90676194.0, + "step": 3542 + }, + { + "epoch": 0.38908412036020207, + "grad_norm": 2.4478771686553955, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7217611074447632, + "num_tokens": 90696220.0, + "step": 3543 + }, + { + "epoch": 0.3891939380628157, + "grad_norm": 2.0817034244537354, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6879197359085083, + "num_tokens": 90724027.0, + "step": 3544 + }, + { + "epoch": 0.38930375576542936, + "grad_norm": 2.0933899879455566, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6918670535087585, + "num_tokens": 90752539.0, + "step": 3545 + }, + { + "epoch": 0.38941357346804306, + "grad_norm": 2.244887590408325, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7133585214614868, + "num_tokens": 90777101.0, + "step": 3546 + }, + { + "epoch": 0.3895233911706567, + "grad_norm": 2.0851550102233887, + "learning_rate": 1e-06, + "loss": 1.0743, + "mean_token_accuracy": 0.6789240837097168, + "num_tokens": 90805985.0, + "step": 3547 + }, + { + "epoch": 0.38963320887327035, + "grad_norm": 2.2906832695007324, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7217573523521423, + "num_tokens": 90829337.0, + "step": 3548 + }, + { + "epoch": 0.38974302657588406, + "grad_norm": 2.0966851711273193, + "learning_rate": 1e-06, + "loss": 1.0403, + "mean_token_accuracy": 0.6876300573348999, + "num_tokens": 90858592.0, + "step": 3549 + }, + { + "epoch": 0.3898528442784977, + "grad_norm": 2.2097949981689453, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.7031075358390808, + "num_tokens": 90885223.0, + "step": 3550 + }, + { + "epoch": 0.38996266198111135, + "grad_norm": 1.9820624589920044, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6940532922744751, + "num_tokens": 90914866.0, + "step": 3551 + }, + { + "epoch": 0.390072479683725, + "grad_norm": 2.1165683269500732, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6990855932235718, + "num_tokens": 90942319.0, + "step": 3552 + }, + { + "epoch": 0.3901822973863387, + "grad_norm": 2.5759854316711426, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7208300828933716, + "num_tokens": 90961071.0, + "step": 3553 + }, + { + "epoch": 0.39029211508895234, + "grad_norm": 2.629051446914673, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.7388055324554443, + "num_tokens": 90979705.0, + "step": 3554 + }, + { + "epoch": 0.390401932791566, + "grad_norm": 2.241455554962158, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7180728912353516, + "num_tokens": 91004464.0, + "step": 3555 + }, + { + "epoch": 0.3905117504941797, + "grad_norm": 2.2830185890197754, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7191829681396484, + "num_tokens": 91028622.0, + "step": 3556 + }, + { + "epoch": 0.39062156819679333, + "grad_norm": 2.2922284603118896, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6925070285797119, + "num_tokens": 91052254.0, + "step": 3557 + }, + { + "epoch": 0.390731385899407, + "grad_norm": 2.358826160430908, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.7016542553901672, + "num_tokens": 91075036.0, + "step": 3558 + }, + { + "epoch": 0.3908412036020206, + "grad_norm": 2.040681838989258, + "learning_rate": 1e-06, + "loss": 1.0601, + "mean_token_accuracy": 0.6871283054351807, + "num_tokens": 91105149.0, + "step": 3559 + }, + { + "epoch": 0.3909510213046343, + "grad_norm": 2.1279773712158203, + "learning_rate": 1e-06, + "loss": 1.0665, + "mean_token_accuracy": 0.6789171099662781, + "num_tokens": 91133918.0, + "step": 3560 + }, + { + "epoch": 0.391060839007248, + "grad_norm": 2.6899447441101074, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7299720048904419, + "num_tokens": 91151466.0, + "step": 3561 + }, + { + "epoch": 0.3911706567098616, + "grad_norm": 2.304497718811035, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7046788930892944, + "num_tokens": 91174332.0, + "step": 3562 + }, + { + "epoch": 0.39128047441247527, + "grad_norm": 2.3035693168640137, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7102917432785034, + "num_tokens": 91198279.0, + "step": 3563 + }, + { + "epoch": 0.39139029211508897, + "grad_norm": 2.298903703689575, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7232614755630493, + "num_tokens": 91222463.0, + "step": 3564 + }, + { + "epoch": 0.3915001098177026, + "grad_norm": 2.451648473739624, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7072809934616089, + "num_tokens": 91244536.0, + "step": 3565 + }, + { + "epoch": 0.39160992752031626, + "grad_norm": 2.104262590408325, + "learning_rate": 1e-06, + "loss": 1.0137, + "mean_token_accuracy": 0.6939595341682434, + "num_tokens": 91273078.0, + "step": 3566 + }, + { + "epoch": 0.39171974522292996, + "grad_norm": 2.099933624267578, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.713469386100769, + "num_tokens": 91300700.0, + "step": 3567 + }, + { + "epoch": 0.3918295629255436, + "grad_norm": 1.9577466249465942, + "learning_rate": 1e-06, + "loss": 1.0294, + "mean_token_accuracy": 0.6935358047485352, + "num_tokens": 91334075.0, + "step": 3568 + }, + { + "epoch": 0.39193938062815725, + "grad_norm": 2.254505157470703, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7180188298225403, + "num_tokens": 91357836.0, + "step": 3569 + }, + { + "epoch": 0.3920491983307709, + "grad_norm": 2.3900249004364014, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6909999251365662, + "num_tokens": 91380222.0, + "step": 3570 + }, + { + "epoch": 0.3921590160333846, + "grad_norm": 2.5188779830932617, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7034169435501099, + "num_tokens": 91401256.0, + "step": 3571 + }, + { + "epoch": 0.39226883373599825, + "grad_norm": 2.255350351333618, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7036027908325195, + "num_tokens": 91426020.0, + "step": 3572 + }, + { + "epoch": 0.3923786514386119, + "grad_norm": 2.5249688625335693, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.704707145690918, + "num_tokens": 91448174.0, + "step": 3573 + }, + { + "epoch": 0.3924884691412256, + "grad_norm": 2.556288719177246, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7114292979240417, + "num_tokens": 91467238.0, + "step": 3574 + }, + { + "epoch": 0.39259828684383924, + "grad_norm": 2.313915252685547, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.703406810760498, + "num_tokens": 91491314.0, + "step": 3575 + }, + { + "epoch": 0.3927081045464529, + "grad_norm": 2.6607742309570312, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7293539047241211, + "num_tokens": 91510207.0, + "step": 3576 + }, + { + "epoch": 0.39281792224906653, + "grad_norm": 2.492152214050293, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.697867214679718, + "num_tokens": 91532357.0, + "step": 3577 + }, + { + "epoch": 0.39292773995168023, + "grad_norm": 2.3102502822875977, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6943241953849792, + "num_tokens": 91558915.0, + "step": 3578 + }, + { + "epoch": 0.3930375576542939, + "grad_norm": 1.9793609380722046, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.6961536407470703, + "num_tokens": 91589325.0, + "step": 3579 + }, + { + "epoch": 0.3931473753569075, + "grad_norm": 2.1266281604766846, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.6976813673973083, + "num_tokens": 91618414.0, + "step": 3580 + }, + { + "epoch": 0.39325719305952117, + "grad_norm": 2.2233448028564453, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7100845575332642, + "num_tokens": 91644170.0, + "step": 3581 + }, + { + "epoch": 0.39336701076213487, + "grad_norm": 2.3506195545196533, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.6979056596755981, + "num_tokens": 91667979.0, + "step": 3582 + }, + { + "epoch": 0.3934768284647485, + "grad_norm": 2.190213918685913, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.7029446959495544, + "num_tokens": 91692230.0, + "step": 3583 + }, + { + "epoch": 0.39358664616736216, + "grad_norm": 2.0900673866271973, + "learning_rate": 1e-06, + "loss": 1.046, + "mean_token_accuracy": 0.6919568777084351, + "num_tokens": 91720228.0, + "step": 3584 + }, + { + "epoch": 0.39369646386997587, + "grad_norm": 2.1964282989501953, + "learning_rate": 1e-06, + "loss": 1.0541, + "mean_token_accuracy": 0.6825114488601685, + "num_tokens": 91747304.0, + "step": 3585 + }, + { + "epoch": 0.3938062815725895, + "grad_norm": 2.322878837585449, + "learning_rate": 1e-06, + "loss": 1.0918, + "mean_token_accuracy": 0.6727104187011719, + "num_tokens": 91772219.0, + "step": 3586 + }, + { + "epoch": 0.39391609927520316, + "grad_norm": 2.1278154850006104, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.7031344175338745, + "num_tokens": 91800586.0, + "step": 3587 + }, + { + "epoch": 0.3940259169778168, + "grad_norm": 2.0548477172851562, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6931148767471313, + "num_tokens": 91832235.0, + "step": 3588 + }, + { + "epoch": 0.3941357346804305, + "grad_norm": 2.074415445327759, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7079921960830688, + "num_tokens": 91860058.0, + "step": 3589 + }, + { + "epoch": 0.39424555238304415, + "grad_norm": 2.0549070835113525, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6913554668426514, + "num_tokens": 91886896.0, + "step": 3590 + }, + { + "epoch": 0.3943553700856578, + "grad_norm": 2.2378783226013184, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.703745424747467, + "num_tokens": 91912278.0, + "step": 3591 + }, + { + "epoch": 0.39446518778827144, + "grad_norm": 2.022230386734009, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.705561101436615, + "num_tokens": 91942859.0, + "step": 3592 + }, + { + "epoch": 0.39457500549088514, + "grad_norm": 2.2786142826080322, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6940535306930542, + "num_tokens": 91965642.0, + "step": 3593 + }, + { + "epoch": 0.3946848231934988, + "grad_norm": 2.2564306259155273, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6973171234130859, + "num_tokens": 91992437.0, + "step": 3594 + }, + { + "epoch": 0.39479464089611244, + "grad_norm": 2.124624490737915, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6959116458892822, + "num_tokens": 92020965.0, + "step": 3595 + }, + { + "epoch": 0.39490445859872614, + "grad_norm": 2.0549962520599365, + "learning_rate": 1e-06, + "loss": 1.1094, + "mean_token_accuracy": 0.6664820909500122, + "num_tokens": 92053915.0, + "step": 3596 + }, + { + "epoch": 0.3950142763013398, + "grad_norm": 1.9989597797393799, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6956425309181213, + "num_tokens": 92084301.0, + "step": 3597 + }, + { + "epoch": 0.39512409400395343, + "grad_norm": 1.9399670362472534, + "learning_rate": 1e-06, + "loss": 1.0374, + "mean_token_accuracy": 0.6875508427619934, + "num_tokens": 92114807.0, + "step": 3598 + }, + { + "epoch": 0.3952339117065671, + "grad_norm": 2.2449769973754883, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.6996716260910034, + "num_tokens": 92139371.0, + "step": 3599 + }, + { + "epoch": 0.3953437294091808, + "grad_norm": 2.170100688934326, + "learning_rate": 1e-06, + "loss": 1.0693, + "mean_token_accuracy": 0.6809967160224915, + "num_tokens": 92167460.0, + "step": 3600 + }, + { + "epoch": 0.3954535471117944, + "grad_norm": 2.2839701175689697, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7107519507408142, + "num_tokens": 92193500.0, + "step": 3601 + }, + { + "epoch": 0.39556336481440807, + "grad_norm": 2.278444528579712, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6878378987312317, + "num_tokens": 92220982.0, + "step": 3602 + }, + { + "epoch": 0.39567318251702177, + "grad_norm": 2.1417267322540283, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6958197355270386, + "num_tokens": 92248190.0, + "step": 3603 + }, + { + "epoch": 0.3957830002196354, + "grad_norm": 2.4356276988983154, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.6884616017341614, + "num_tokens": 92271292.0, + "step": 3604 + }, + { + "epoch": 0.39589281792224906, + "grad_norm": 2.141559600830078, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6900213956832886, + "num_tokens": 92299771.0, + "step": 3605 + }, + { + "epoch": 0.3960026356248627, + "grad_norm": 2.2725775241851807, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.694006085395813, + "num_tokens": 92323964.0, + "step": 3606 + }, + { + "epoch": 0.3961124533274764, + "grad_norm": 2.435223340988159, + "learning_rate": 1e-06, + "loss": 1.0744, + "mean_token_accuracy": 0.6799386739730835, + "num_tokens": 92345361.0, + "step": 3607 + }, + { + "epoch": 0.39622227103009006, + "grad_norm": 2.3722879886627197, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.6961830258369446, + "num_tokens": 92367146.0, + "step": 3608 + }, + { + "epoch": 0.3963320887327037, + "grad_norm": 2.285576343536377, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7146804332733154, + "num_tokens": 92392087.0, + "step": 3609 + }, + { + "epoch": 0.39644190643531735, + "grad_norm": 2.1939258575439453, + "learning_rate": 1e-06, + "loss": 1.0414, + "mean_token_accuracy": 0.6848269701004028, + "num_tokens": 92417650.0, + "step": 3610 + }, + { + "epoch": 0.39655172413793105, + "grad_norm": 2.4587318897247314, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7018840312957764, + "num_tokens": 92438295.0, + "step": 3611 + }, + { + "epoch": 0.3966615418405447, + "grad_norm": 2.1288113594055176, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7079895734786987, + "num_tokens": 92464045.0, + "step": 3612 + }, + { + "epoch": 0.39677135954315834, + "grad_norm": 2.083510398864746, + "learning_rate": 1e-06, + "loss": 1.0532, + "mean_token_accuracy": 0.6928621530532837, + "num_tokens": 92492770.0, + "step": 3613 + }, + { + "epoch": 0.39688117724577204, + "grad_norm": 2.2607004642486572, + "learning_rate": 1e-06, + "loss": 1.0678, + "mean_token_accuracy": 0.685295581817627, + "num_tokens": 92521481.0, + "step": 3614 + }, + { + "epoch": 0.3969909949483857, + "grad_norm": 1.9573338031768799, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6814013719558716, + "num_tokens": 92554290.0, + "step": 3615 + }, + { + "epoch": 0.39710081265099934, + "grad_norm": 2.2619102001190186, + "learning_rate": 1e-06, + "loss": 1.052, + "mean_token_accuracy": 0.6851478815078735, + "num_tokens": 92580290.0, + "step": 3616 + }, + { + "epoch": 0.397210630353613, + "grad_norm": 2.0370898246765137, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7047337889671326, + "num_tokens": 92611304.0, + "step": 3617 + }, + { + "epoch": 0.3973204480562267, + "grad_norm": 2.220503330230713, + "learning_rate": 1e-06, + "loss": 1.0788, + "mean_token_accuracy": 0.6834436655044556, + "num_tokens": 92638329.0, + "step": 3618 + }, + { + "epoch": 0.39743026575884033, + "grad_norm": 2.1269638538360596, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7155846357345581, + "num_tokens": 92663951.0, + "step": 3619 + }, + { + "epoch": 0.397540083461454, + "grad_norm": 2.0778450965881348, + "learning_rate": 1e-06, + "loss": 1.0976, + "mean_token_accuracy": 0.6789697408676147, + "num_tokens": 92694308.0, + "step": 3620 + }, + { + "epoch": 0.3976499011640676, + "grad_norm": 2.2901649475097656, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.723480224609375, + "num_tokens": 92714745.0, + "step": 3621 + }, + { + "epoch": 0.3977597188666813, + "grad_norm": 2.3808510303497314, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7102040648460388, + "num_tokens": 92736469.0, + "step": 3622 + }, + { + "epoch": 0.39786953656929497, + "grad_norm": 2.2690389156341553, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7136436700820923, + "num_tokens": 92760678.0, + "step": 3623 + }, + { + "epoch": 0.3979793542719086, + "grad_norm": 2.390465259552002, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7046393752098083, + "num_tokens": 92783561.0, + "step": 3624 + }, + { + "epoch": 0.3980891719745223, + "grad_norm": 2.0349113941192627, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.69859778881073, + "num_tokens": 92815205.0, + "step": 3625 + }, + { + "epoch": 0.39819898967713596, + "grad_norm": 1.9622509479522705, + "learning_rate": 1e-06, + "loss": 1.0525, + "mean_token_accuracy": 0.6900523900985718, + "num_tokens": 92847346.0, + "step": 3626 + }, + { + "epoch": 0.3983088073797496, + "grad_norm": 2.097609043121338, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7161561250686646, + "num_tokens": 92874259.0, + "step": 3627 + }, + { + "epoch": 0.39841862508236325, + "grad_norm": 2.233215570449829, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.73661208152771, + "num_tokens": 92898131.0, + "step": 3628 + }, + { + "epoch": 0.39852844278497696, + "grad_norm": 2.1058168411254883, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.7022402286529541, + "num_tokens": 92924832.0, + "step": 3629 + }, + { + "epoch": 0.3986382604875906, + "grad_norm": 2.055889129638672, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7119411826133728, + "num_tokens": 92954383.0, + "step": 3630 + }, + { + "epoch": 0.39874807819020425, + "grad_norm": 2.22568416595459, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.6997801065444946, + "num_tokens": 92980880.0, + "step": 3631 + }, + { + "epoch": 0.39885789589281795, + "grad_norm": 2.169438600540161, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.6969584226608276, + "num_tokens": 93007438.0, + "step": 3632 + }, + { + "epoch": 0.3989677135954316, + "grad_norm": 2.6389145851135254, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.6967823505401611, + "num_tokens": 93026739.0, + "step": 3633 + }, + { + "epoch": 0.39907753129804524, + "grad_norm": 2.1480727195739746, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7078922986984253, + "num_tokens": 93051088.0, + "step": 3634 + }, + { + "epoch": 0.3991873490006589, + "grad_norm": 1.982283592224121, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.6998594999313354, + "num_tokens": 93080687.0, + "step": 3635 + }, + { + "epoch": 0.3992971667032726, + "grad_norm": 2.5527501106262207, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7236565351486206, + "num_tokens": 93100241.0, + "step": 3636 + }, + { + "epoch": 0.39940698440588623, + "grad_norm": 2.0281260013580322, + "learning_rate": 1e-06, + "loss": 1.0508, + "mean_token_accuracy": 0.6914819478988647, + "num_tokens": 93134389.0, + "step": 3637 + }, + { + "epoch": 0.3995168021084999, + "grad_norm": 2.1196744441986084, + "learning_rate": 1e-06, + "loss": 1.0339, + "mean_token_accuracy": 0.6904370784759521, + "num_tokens": 93164705.0, + "step": 3638 + }, + { + "epoch": 0.3996266198111135, + "grad_norm": 2.3117289543151855, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7078627347946167, + "num_tokens": 93187152.0, + "step": 3639 + }, + { + "epoch": 0.3997364375137272, + "grad_norm": 2.0075957775115967, + "learning_rate": 1e-06, + "loss": 1.0339, + "mean_token_accuracy": 0.6935909986495972, + "num_tokens": 93217805.0, + "step": 3640 + }, + { + "epoch": 0.3998462552163409, + "grad_norm": 2.0790176391601562, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7119364738464355, + "num_tokens": 93245672.0, + "step": 3641 + }, + { + "epoch": 0.3999560729189545, + "grad_norm": 2.2201125621795654, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.717933177947998, + "num_tokens": 93269677.0, + "step": 3642 + }, + { + "epoch": 0.4000658906215682, + "grad_norm": 1.9905229806900024, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7200535535812378, + "num_tokens": 93300334.0, + "step": 3643 + }, + { + "epoch": 0.40017570832418187, + "grad_norm": 2.1137452125549316, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.699725866317749, + "num_tokens": 93328955.0, + "step": 3644 + }, + { + "epoch": 0.4002855260267955, + "grad_norm": 1.9069064855575562, + "learning_rate": 1e-06, + "loss": 1.1287, + "mean_token_accuracy": 0.6704338192939758, + "num_tokens": 93363963.0, + "step": 3645 + }, + { + "epoch": 0.40039534372940916, + "grad_norm": 2.3682456016540527, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7105404138565063, + "num_tokens": 93385940.0, + "step": 3646 + }, + { + "epoch": 0.40050516143202286, + "grad_norm": 2.2306385040283203, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7113539576530457, + "num_tokens": 93411051.0, + "step": 3647 + }, + { + "epoch": 0.4006149791346365, + "grad_norm": 2.109253168106079, + "learning_rate": 1e-06, + "loss": 1.0399, + "mean_token_accuracy": 0.6861662268638611, + "num_tokens": 93439006.0, + "step": 3648 + }, + { + "epoch": 0.40072479683725015, + "grad_norm": 2.255762815475464, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.683204174041748, + "num_tokens": 93464714.0, + "step": 3649 + }, + { + "epoch": 0.4008346145398638, + "grad_norm": 2.5420804023742676, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6921520233154297, + "num_tokens": 93485117.0, + "step": 3650 + }, + { + "epoch": 0.4009444322424775, + "grad_norm": 2.165621280670166, + "learning_rate": 1e-06, + "loss": 1.0561, + "mean_token_accuracy": 0.6814645528793335, + "num_tokens": 93512215.0, + "step": 3651 + }, + { + "epoch": 0.40105424994509115, + "grad_norm": 2.106156587600708, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.7007479667663574, + "num_tokens": 93539580.0, + "step": 3652 + }, + { + "epoch": 0.4011640676477048, + "grad_norm": 2.136305093765259, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7086237668991089, + "num_tokens": 93567639.0, + "step": 3653 + }, + { + "epoch": 0.4012738853503185, + "grad_norm": 2.116589307785034, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7050139904022217, + "num_tokens": 93597726.0, + "step": 3654 + }, + { + "epoch": 0.40138370305293214, + "grad_norm": 2.283689498901367, + "learning_rate": 1e-06, + "loss": 1.0423, + "mean_token_accuracy": 0.6962496042251587, + "num_tokens": 93621836.0, + "step": 3655 + }, + { + "epoch": 0.4014935207555458, + "grad_norm": 2.238227128982544, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7090697884559631, + "num_tokens": 93645705.0, + "step": 3656 + }, + { + "epoch": 0.40160333845815943, + "grad_norm": 2.2213642597198486, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7148480415344238, + "num_tokens": 93670649.0, + "step": 3657 + }, + { + "epoch": 0.40171315616077313, + "grad_norm": 2.1157610416412354, + "learning_rate": 1e-06, + "loss": 1.0575, + "mean_token_accuracy": 0.6827446222305298, + "num_tokens": 93699013.0, + "step": 3658 + }, + { + "epoch": 0.4018229738633868, + "grad_norm": 2.283555746078491, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7165294885635376, + "num_tokens": 93724295.0, + "step": 3659 + }, + { + "epoch": 0.4019327915660004, + "grad_norm": 2.3480358123779297, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7090173959732056, + "num_tokens": 93749848.0, + "step": 3660 + }, + { + "epoch": 0.4020426092686141, + "grad_norm": 1.9489959478378296, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6987057328224182, + "num_tokens": 93781158.0, + "step": 3661 + }, + { + "epoch": 0.40215242697122777, + "grad_norm": 2.3502938747406006, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7159957885742188, + "num_tokens": 93804222.0, + "step": 3662 + }, + { + "epoch": 0.4022622446738414, + "grad_norm": 2.570225238800049, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7242265343666077, + "num_tokens": 93823278.0, + "step": 3663 + }, + { + "epoch": 0.40237206237645506, + "grad_norm": 2.4426255226135254, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.70079505443573, + "num_tokens": 93844947.0, + "step": 3664 + }, + { + "epoch": 0.40248188007906877, + "grad_norm": 2.55703067779541, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7148153185844421, + "num_tokens": 93864727.0, + "step": 3665 + }, + { + "epoch": 0.4025916977816824, + "grad_norm": 2.005678176879883, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.6923738121986389, + "num_tokens": 93900348.0, + "step": 3666 + }, + { + "epoch": 0.40270151548429606, + "grad_norm": 2.7569234371185303, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.6948643922805786, + "num_tokens": 93919161.0, + "step": 3667 + }, + { + "epoch": 0.4028113331869097, + "grad_norm": 2.0602786540985107, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6964748501777649, + "num_tokens": 93948843.0, + "step": 3668 + }, + { + "epoch": 0.4029211508895234, + "grad_norm": 2.148524522781372, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.688265323638916, + "num_tokens": 93974842.0, + "step": 3669 + }, + { + "epoch": 0.40303096859213705, + "grad_norm": 1.8945480585098267, + "learning_rate": 1e-06, + "loss": 1.061, + "mean_token_accuracy": 0.672288179397583, + "num_tokens": 94007866.0, + "step": 3670 + }, + { + "epoch": 0.4031407862947507, + "grad_norm": 2.2843973636627197, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7228631377220154, + "num_tokens": 94031259.0, + "step": 3671 + }, + { + "epoch": 0.4032506039973644, + "grad_norm": 2.0320732593536377, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7058780193328857, + "num_tokens": 94061400.0, + "step": 3672 + }, + { + "epoch": 0.40336042169997804, + "grad_norm": 2.2083609104156494, + "learning_rate": 1e-06, + "loss": 1.0482, + "mean_token_accuracy": 0.6855872273445129, + "num_tokens": 94089935.0, + "step": 3673 + }, + { + "epoch": 0.4034702394025917, + "grad_norm": 2.0886220932006836, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.709399938583374, + "num_tokens": 94116242.0, + "step": 3674 + }, + { + "epoch": 0.40358005710520534, + "grad_norm": 2.523200035095215, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6958240270614624, + "num_tokens": 94137629.0, + "step": 3675 + }, + { + "epoch": 0.40368987480781904, + "grad_norm": 2.4682724475860596, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7023060917854309, + "num_tokens": 94161479.0, + "step": 3676 + }, + { + "epoch": 0.4037996925104327, + "grad_norm": 2.327976703643799, + "learning_rate": 1e-06, + "loss": 1.1135, + "mean_token_accuracy": 0.6719962358474731, + "num_tokens": 94184924.0, + "step": 3677 + }, + { + "epoch": 0.40390951021304633, + "grad_norm": 2.520994186401367, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7261639833450317, + "num_tokens": 94206390.0, + "step": 3678 + }, + { + "epoch": 0.40401932791566003, + "grad_norm": 1.8544094562530518, + "learning_rate": 1e-06, + "loss": 1.0589, + "mean_token_accuracy": 0.6829895973205566, + "num_tokens": 94243460.0, + "step": 3679 + }, + { + "epoch": 0.4041291456182737, + "grad_norm": 2.02835750579834, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6905555725097656, + "num_tokens": 94276050.0, + "step": 3680 + }, + { + "epoch": 0.4042389633208873, + "grad_norm": 2.264735460281372, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.705345630645752, + "num_tokens": 94299208.0, + "step": 3681 + }, + { + "epoch": 0.40434878102350097, + "grad_norm": 2.188180446624756, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7087228298187256, + "num_tokens": 94323941.0, + "step": 3682 + }, + { + "epoch": 0.40445859872611467, + "grad_norm": 2.0520095825195312, + "learning_rate": 1e-06, + "loss": 1.051, + "mean_token_accuracy": 0.6852807998657227, + "num_tokens": 94353038.0, + "step": 3683 + }, + { + "epoch": 0.4045684164287283, + "grad_norm": 2.333170175552368, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7072620391845703, + "num_tokens": 94375502.0, + "step": 3684 + }, + { + "epoch": 0.40467823413134196, + "grad_norm": 1.9628233909606934, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.695281445980072, + "num_tokens": 94408110.0, + "step": 3685 + }, + { + "epoch": 0.4047880518339556, + "grad_norm": 2.362703561782837, + "learning_rate": 1e-06, + "loss": 1.0863, + "mean_token_accuracy": 0.6815794706344604, + "num_tokens": 94433702.0, + "step": 3686 + }, + { + "epoch": 0.4048978695365693, + "grad_norm": 2.4853591918945312, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6975885629653931, + "num_tokens": 94455191.0, + "step": 3687 + }, + { + "epoch": 0.40500768723918296, + "grad_norm": 2.163980722427368, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.6923305988311768, + "num_tokens": 94481549.0, + "step": 3688 + }, + { + "epoch": 0.4051175049417966, + "grad_norm": 2.1176655292510986, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7344552874565125, + "num_tokens": 94507673.0, + "step": 3689 + }, + { + "epoch": 0.4052273226444103, + "grad_norm": 2.324659585952759, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.7015622854232788, + "num_tokens": 94530654.0, + "step": 3690 + }, + { + "epoch": 0.40533714034702395, + "grad_norm": 2.2817583084106445, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7209635972976685, + "num_tokens": 94554812.0, + "step": 3691 + }, + { + "epoch": 0.4054469580496376, + "grad_norm": 2.8158650398254395, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7216942310333252, + "num_tokens": 94571135.0, + "step": 3692 + }, + { + "epoch": 0.40555677575225124, + "grad_norm": 2.050089120864868, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.6911924481391907, + "num_tokens": 94600994.0, + "step": 3693 + }, + { + "epoch": 0.40566659345486494, + "grad_norm": 2.4724628925323486, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7334496378898621, + "num_tokens": 94623308.0, + "step": 3694 + }, + { + "epoch": 0.4057764111574786, + "grad_norm": 2.1655688285827637, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6935686469078064, + "num_tokens": 94649269.0, + "step": 3695 + }, + { + "epoch": 0.40588622886009224, + "grad_norm": 2.213843584060669, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7059029340744019, + "num_tokens": 94672995.0, + "step": 3696 + }, + { + "epoch": 0.4059960465627059, + "grad_norm": 2.2615153789520264, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.6980783939361572, + "num_tokens": 94697078.0, + "step": 3697 + }, + { + "epoch": 0.4061058642653196, + "grad_norm": 2.319221019744873, + "learning_rate": 1e-06, + "loss": 1.0238, + "mean_token_accuracy": 0.6957824230194092, + "num_tokens": 94721910.0, + "step": 3698 + }, + { + "epoch": 0.40621568196793323, + "grad_norm": 2.3112096786499023, + "learning_rate": 1e-06, + "loss": 1.0913, + "mean_token_accuracy": 0.6827594041824341, + "num_tokens": 94749897.0, + "step": 3699 + }, + { + "epoch": 0.4063254996705469, + "grad_norm": 2.1870455741882324, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7112830877304077, + "num_tokens": 94775615.0, + "step": 3700 + }, + { + "epoch": 0.4064353173731606, + "grad_norm": 2.2671494483947754, + "learning_rate": 1e-06, + "loss": 1.0509, + "mean_token_accuracy": 0.680091381072998, + "num_tokens": 94801337.0, + "step": 3701 + }, + { + "epoch": 0.4065451350757742, + "grad_norm": 2.2636823654174805, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7313675284385681, + "num_tokens": 94824605.0, + "step": 3702 + }, + { + "epoch": 0.40665495277838787, + "grad_norm": 2.369987964630127, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.6985664367675781, + "num_tokens": 94848432.0, + "step": 3703 + }, + { + "epoch": 0.4067647704810015, + "grad_norm": 2.0696728229522705, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7106382250785828, + "num_tokens": 94877790.0, + "step": 3704 + }, + { + "epoch": 0.4068745881836152, + "grad_norm": 2.3888258934020996, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.6953226327896118, + "num_tokens": 94901309.0, + "step": 3705 + }, + { + "epoch": 0.40698440588622886, + "grad_norm": 2.567373514175415, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.731326699256897, + "num_tokens": 94921211.0, + "step": 3706 + }, + { + "epoch": 0.4070942235888425, + "grad_norm": 2.3304522037506104, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7116269469261169, + "num_tokens": 94944026.0, + "step": 3707 + }, + { + "epoch": 0.4072040412914562, + "grad_norm": 1.9537255764007568, + "learning_rate": 1e-06, + "loss": 1.0793, + "mean_token_accuracy": 0.6884146928787231, + "num_tokens": 94975306.0, + "step": 3708 + }, + { + "epoch": 0.40731385899406986, + "grad_norm": 2.3023157119750977, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.6985311508178711, + "num_tokens": 94999716.0, + "step": 3709 + }, + { + "epoch": 0.4074236766966835, + "grad_norm": 2.3379464149475098, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7078670263290405, + "num_tokens": 95022396.0, + "step": 3710 + }, + { + "epoch": 0.40753349439929715, + "grad_norm": 2.2891104221343994, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.7031830549240112, + "num_tokens": 95047380.0, + "step": 3711 + }, + { + "epoch": 0.40764331210191085, + "grad_norm": 2.392723798751831, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7145140767097473, + "num_tokens": 95069409.0, + "step": 3712 + }, + { + "epoch": 0.4077531298045245, + "grad_norm": 2.4655513763427734, + "learning_rate": 1e-06, + "loss": 1.0536, + "mean_token_accuracy": 0.6861239671707153, + "num_tokens": 95092084.0, + "step": 3713 + }, + { + "epoch": 0.40786294750713814, + "grad_norm": 2.53967547416687, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7079389095306396, + "num_tokens": 95111688.0, + "step": 3714 + }, + { + "epoch": 0.4079727652097518, + "grad_norm": 2.193138599395752, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7112051248550415, + "num_tokens": 95137773.0, + "step": 3715 + }, + { + "epoch": 0.4080825829123655, + "grad_norm": 2.3360390663146973, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.7005201578140259, + "num_tokens": 95159626.0, + "step": 3716 + }, + { + "epoch": 0.40819240061497913, + "grad_norm": 2.518690824508667, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.724552571773529, + "num_tokens": 95179508.0, + "step": 3717 + }, + { + "epoch": 0.4083022183175928, + "grad_norm": 2.098484992980957, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6992233395576477, + "num_tokens": 95207033.0, + "step": 3718 + }, + { + "epoch": 0.4084120360202065, + "grad_norm": 2.233407974243164, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6915715336799622, + "num_tokens": 95232202.0, + "step": 3719 + }, + { + "epoch": 0.4085218537228201, + "grad_norm": 2.181633710861206, + "learning_rate": 1e-06, + "loss": 1.0504, + "mean_token_accuracy": 0.6819219589233398, + "num_tokens": 95257413.0, + "step": 3720 + }, + { + "epoch": 0.4086316714254338, + "grad_norm": 2.1454575061798096, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7166188955307007, + "num_tokens": 95282806.0, + "step": 3721 + }, + { + "epoch": 0.4087414891280474, + "grad_norm": 2.0753657817840576, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.6996548771858215, + "num_tokens": 95313229.0, + "step": 3722 + }, + { + "epoch": 0.4088513068306611, + "grad_norm": 2.2758681774139404, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.6994757652282715, + "num_tokens": 95336610.0, + "step": 3723 + }, + { + "epoch": 0.40896112453327477, + "grad_norm": 2.035102128982544, + "learning_rate": 1e-06, + "loss": 1.0727, + "mean_token_accuracy": 0.681894063949585, + "num_tokens": 95367467.0, + "step": 3724 + }, + { + "epoch": 0.4090709422358884, + "grad_norm": 2.120455741882324, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7097181677818298, + "num_tokens": 95395039.0, + "step": 3725 + }, + { + "epoch": 0.40918075993850206, + "grad_norm": 2.2189667224884033, + "learning_rate": 1e-06, + "loss": 1.0289, + "mean_token_accuracy": 0.6869612336158752, + "num_tokens": 95421657.0, + "step": 3726 + }, + { + "epoch": 0.40929057764111576, + "grad_norm": 2.3551080226898193, + "learning_rate": 1e-06, + "loss": 1.058, + "mean_token_accuracy": 0.6849703788757324, + "num_tokens": 95447056.0, + "step": 3727 + }, + { + "epoch": 0.4094003953437294, + "grad_norm": 2.3334450721740723, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.7019751667976379, + "num_tokens": 95470029.0, + "step": 3728 + }, + { + "epoch": 0.40951021304634305, + "grad_norm": 2.0427405834198, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.6908149719238281, + "num_tokens": 95497861.0, + "step": 3729 + }, + { + "epoch": 0.40962003074895675, + "grad_norm": 2.207202672958374, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.6940386295318604, + "num_tokens": 95523741.0, + "step": 3730 + }, + { + "epoch": 0.4097298484515704, + "grad_norm": 2.292454481124878, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6944040656089783, + "num_tokens": 95548126.0, + "step": 3731 + }, + { + "epoch": 0.40983966615418405, + "grad_norm": 2.354403495788574, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7076663970947266, + "num_tokens": 95572475.0, + "step": 3732 + }, + { + "epoch": 0.4099494838567977, + "grad_norm": 2.31900691986084, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7330704927444458, + "num_tokens": 95593934.0, + "step": 3733 + }, + { + "epoch": 0.4100593015594114, + "grad_norm": 2.4994983673095703, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7324249148368835, + "num_tokens": 95613795.0, + "step": 3734 + }, + { + "epoch": 0.41016911926202504, + "grad_norm": 2.4737370014190674, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7212214469909668, + "num_tokens": 95635076.0, + "step": 3735 + }, + { + "epoch": 0.4102789369646387, + "grad_norm": 2.398763656616211, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6969612836837769, + "num_tokens": 95657531.0, + "step": 3736 + }, + { + "epoch": 0.4103887546672524, + "grad_norm": 2.3897628784179688, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7198857069015503, + "num_tokens": 95678618.0, + "step": 3737 + }, + { + "epoch": 0.41049857236986603, + "grad_norm": 2.137768268585205, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7305121421813965, + "num_tokens": 95704097.0, + "step": 3738 + }, + { + "epoch": 0.4106083900724797, + "grad_norm": 2.091641426086426, + "learning_rate": 1e-06, + "loss": 1.0425, + "mean_token_accuracy": 0.684443473815918, + "num_tokens": 95732544.0, + "step": 3739 + }, + { + "epoch": 0.4107182077750933, + "grad_norm": 2.2979965209960938, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7011914849281311, + "num_tokens": 95756855.0, + "step": 3740 + }, + { + "epoch": 0.410828025477707, + "grad_norm": 2.4283599853515625, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.6926351189613342, + "num_tokens": 95778048.0, + "step": 3741 + }, + { + "epoch": 0.41093784318032067, + "grad_norm": 2.1753439903259277, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.703901469707489, + "num_tokens": 95803771.0, + "step": 3742 + }, + { + "epoch": 0.4110476608829343, + "grad_norm": 2.116518974304199, + "learning_rate": 1e-06, + "loss": 1.0512, + "mean_token_accuracy": 0.6875308156013489, + "num_tokens": 95834551.0, + "step": 3743 + }, + { + "epoch": 0.41115747858554796, + "grad_norm": 2.186779737472534, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6870387196540833, + "num_tokens": 95862746.0, + "step": 3744 + }, + { + "epoch": 0.41126729628816167, + "grad_norm": 2.3579001426696777, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.7012476325035095, + "num_tokens": 95884771.0, + "step": 3745 + }, + { + "epoch": 0.4113771139907753, + "grad_norm": 2.091899871826172, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7092423439025879, + "num_tokens": 95910595.0, + "step": 3746 + }, + { + "epoch": 0.41148693169338896, + "grad_norm": 2.079346179962158, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6985747814178467, + "num_tokens": 95939250.0, + "step": 3747 + }, + { + "epoch": 0.41159674939600266, + "grad_norm": 2.0504322052001953, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.6950724124908447, + "num_tokens": 95967306.0, + "step": 3748 + }, + { + "epoch": 0.4117065670986163, + "grad_norm": 2.0062170028686523, + "learning_rate": 1e-06, + "loss": 1.0823, + "mean_token_accuracy": 0.6737916469573975, + "num_tokens": 95998515.0, + "step": 3749 + }, + { + "epoch": 0.41181638480122995, + "grad_norm": 2.349806547164917, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6994929313659668, + "num_tokens": 96022206.0, + "step": 3750 + }, + { + "epoch": 0.4119262025038436, + "grad_norm": 2.0817952156066895, + "learning_rate": 1e-06, + "loss": 1.0425, + "mean_token_accuracy": 0.6853510737419128, + "num_tokens": 96051324.0, + "step": 3751 + }, + { + "epoch": 0.4120360202064573, + "grad_norm": 1.8541135787963867, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7108752131462097, + "num_tokens": 96085214.0, + "step": 3752 + }, + { + "epoch": 0.41214583790907094, + "grad_norm": 2.1761672496795654, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7062975168228149, + "num_tokens": 96111349.0, + "step": 3753 + }, + { + "epoch": 0.4122556556116846, + "grad_norm": 2.2284538745880127, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.7032886743545532, + "num_tokens": 96136633.0, + "step": 3754 + }, + { + "epoch": 0.4123654733142983, + "grad_norm": 2.3052897453308105, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7067134380340576, + "num_tokens": 96160982.0, + "step": 3755 + }, + { + "epoch": 0.41247529101691194, + "grad_norm": 2.055396795272827, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7272588014602661, + "num_tokens": 96187053.0, + "step": 3756 + }, + { + "epoch": 0.4125851087195256, + "grad_norm": 2.4172794818878174, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7063193917274475, + "num_tokens": 96210475.0, + "step": 3757 + }, + { + "epoch": 0.41269492642213923, + "grad_norm": 2.253190040588379, + "learning_rate": 1e-06, + "loss": 1.0468, + "mean_token_accuracy": 0.6857136487960815, + "num_tokens": 96239109.0, + "step": 3758 + }, + { + "epoch": 0.41280474412475293, + "grad_norm": 2.0014708042144775, + "learning_rate": 1e-06, + "loss": 1.0454, + "mean_token_accuracy": 0.6860213875770569, + "num_tokens": 96270666.0, + "step": 3759 + }, + { + "epoch": 0.4129145618273666, + "grad_norm": 2.2651302814483643, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7056354284286499, + "num_tokens": 96296088.0, + "step": 3760 + }, + { + "epoch": 0.4130243795299802, + "grad_norm": 2.3233227729797363, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.7079188227653503, + "num_tokens": 96320693.0, + "step": 3761 + }, + { + "epoch": 0.41313419723259387, + "grad_norm": 2.217590093612671, + "learning_rate": 1e-06, + "loss": 1.0195, + "mean_token_accuracy": 0.6943342089653015, + "num_tokens": 96347895.0, + "step": 3762 + }, + { + "epoch": 0.41324401493520757, + "grad_norm": 2.081939935684204, + "learning_rate": 1e-06, + "loss": 1.0684, + "mean_token_accuracy": 0.6820917129516602, + "num_tokens": 96378157.0, + "step": 3763 + }, + { + "epoch": 0.4133538326378212, + "grad_norm": 2.1928255558013916, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7009080648422241, + "num_tokens": 96406333.0, + "step": 3764 + }, + { + "epoch": 0.41346365034043486, + "grad_norm": 2.1788437366485596, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7106058597564697, + "num_tokens": 96431791.0, + "step": 3765 + }, + { + "epoch": 0.41357346804304856, + "grad_norm": 2.308628797531128, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.6871365308761597, + "num_tokens": 96455812.0, + "step": 3766 + }, + { + "epoch": 0.4136832857456622, + "grad_norm": 2.1737935543060303, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.6949710845947266, + "num_tokens": 96482955.0, + "step": 3767 + }, + { + "epoch": 0.41379310344827586, + "grad_norm": 2.359809398651123, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.7079178094863892, + "num_tokens": 96508777.0, + "step": 3768 + }, + { + "epoch": 0.4139029211508895, + "grad_norm": 2.2574076652526855, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.6983808875083923, + "num_tokens": 96532789.0, + "step": 3769 + }, + { + "epoch": 0.4140127388535032, + "grad_norm": 2.314274787902832, + "learning_rate": 1e-06, + "loss": 1.043, + "mean_token_accuracy": 0.690388023853302, + "num_tokens": 96558217.0, + "step": 3770 + }, + { + "epoch": 0.41412255655611685, + "grad_norm": 2.0933921337127686, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.6968666315078735, + "num_tokens": 96585416.0, + "step": 3771 + }, + { + "epoch": 0.4142323742587305, + "grad_norm": 2.1320905685424805, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7204868197441101, + "num_tokens": 96613249.0, + "step": 3772 + }, + { + "epoch": 0.41434219196134414, + "grad_norm": 2.442413806915283, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6932626962661743, + "num_tokens": 96634768.0, + "step": 3773 + }, + { + "epoch": 0.41445200966395784, + "grad_norm": 2.0835304260253906, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.7059416770935059, + "num_tokens": 96664152.0, + "step": 3774 + }, + { + "epoch": 0.4145618273665715, + "grad_norm": 2.4447181224823, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.725898265838623, + "num_tokens": 96687624.0, + "step": 3775 + }, + { + "epoch": 0.41467164506918514, + "grad_norm": 2.2248690128326416, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.692884087562561, + "num_tokens": 96715903.0, + "step": 3776 + }, + { + "epoch": 0.41478146277179884, + "grad_norm": 2.4639079570770264, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7131465077400208, + "num_tokens": 96736306.0, + "step": 3777 + }, + { + "epoch": 0.4148912804744125, + "grad_norm": 2.050006628036499, + "learning_rate": 1e-06, + "loss": 1.1031, + "mean_token_accuracy": 0.6694597601890564, + "num_tokens": 96766914.0, + "step": 3778 + }, + { + "epoch": 0.41500109817702613, + "grad_norm": 2.15999698638916, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.6991491317749023, + "num_tokens": 96793904.0, + "step": 3779 + }, + { + "epoch": 0.4151109158796398, + "grad_norm": 2.621286153793335, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6991229057312012, + "num_tokens": 96814086.0, + "step": 3780 + }, + { + "epoch": 0.4152207335822535, + "grad_norm": 2.659123420715332, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7099881172180176, + "num_tokens": 96832992.0, + "step": 3781 + }, + { + "epoch": 0.4153305512848671, + "grad_norm": 2.1038386821746826, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6906166076660156, + "num_tokens": 96864995.0, + "step": 3782 + }, + { + "epoch": 0.41544036898748077, + "grad_norm": 2.4100615978240967, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.723164439201355, + "num_tokens": 96885724.0, + "step": 3783 + }, + { + "epoch": 0.41555018669009447, + "grad_norm": 2.202049970626831, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.6943386793136597, + "num_tokens": 96910846.0, + "step": 3784 + }, + { + "epoch": 0.4156600043927081, + "grad_norm": 2.4510324001312256, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6974217295646667, + "num_tokens": 96932764.0, + "step": 3785 + }, + { + "epoch": 0.41576982209532176, + "grad_norm": 2.13382887840271, + "learning_rate": 1e-06, + "loss": 1.0723, + "mean_token_accuracy": 0.6858022809028625, + "num_tokens": 96958243.0, + "step": 3786 + }, + { + "epoch": 0.4158796397979354, + "grad_norm": 1.9968085289001465, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.72455894947052, + "num_tokens": 96985630.0, + "step": 3787 + }, + { + "epoch": 0.4159894575005491, + "grad_norm": 2.102322578430176, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7003886103630066, + "num_tokens": 97013607.0, + "step": 3788 + }, + { + "epoch": 0.41609927520316276, + "grad_norm": 2.3989171981811523, + "learning_rate": 1e-06, + "loss": 1.0305, + "mean_token_accuracy": 0.6862235069274902, + "num_tokens": 97037637.0, + "step": 3789 + }, + { + "epoch": 0.4162090929057764, + "grad_norm": 2.3950424194335938, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7241554260253906, + "num_tokens": 97058394.0, + "step": 3790 + }, + { + "epoch": 0.41631891060839005, + "grad_norm": 2.338839292526245, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7186818718910217, + "num_tokens": 97079401.0, + "step": 3791 + }, + { + "epoch": 0.41642872831100375, + "grad_norm": 2.016833782196045, + "learning_rate": 1e-06, + "loss": 1.0829, + "mean_token_accuracy": 0.6730422973632812, + "num_tokens": 97110064.0, + "step": 3792 + }, + { + "epoch": 0.4165385460136174, + "grad_norm": 2.040013551712036, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7120879292488098, + "num_tokens": 97138541.0, + "step": 3793 + }, + { + "epoch": 0.41664836371623104, + "grad_norm": 1.9418988227844238, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7115949392318726, + "num_tokens": 97171320.0, + "step": 3794 + }, + { + "epoch": 0.41675818141884474, + "grad_norm": 2.1406383514404297, + "learning_rate": 1e-06, + "loss": 1.0308, + "mean_token_accuracy": 0.6864405870437622, + "num_tokens": 97198529.0, + "step": 3795 + }, + { + "epoch": 0.4168679991214584, + "grad_norm": 2.035041093826294, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.7011922001838684, + "num_tokens": 97227046.0, + "step": 3796 + }, + { + "epoch": 0.41697781682407203, + "grad_norm": 2.1765778064727783, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6949732899665833, + "num_tokens": 97253542.0, + "step": 3797 + }, + { + "epoch": 0.4170876345266857, + "grad_norm": 2.4554576873779297, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7090327739715576, + "num_tokens": 97274764.0, + "step": 3798 + }, + { + "epoch": 0.4171974522292994, + "grad_norm": 2.686098337173462, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.723556637763977, + "num_tokens": 97292988.0, + "step": 3799 + }, + { + "epoch": 0.417307269931913, + "grad_norm": 2.166264057159424, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7077717781066895, + "num_tokens": 97320735.0, + "step": 3800 + }, + { + "epoch": 0.4174170876345267, + "grad_norm": 2.428635597229004, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.6960564851760864, + "num_tokens": 97342870.0, + "step": 3801 + }, + { + "epoch": 0.4175269053371403, + "grad_norm": 2.1075527667999268, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7061476707458496, + "num_tokens": 97370547.0, + "step": 3802 + }, + { + "epoch": 0.417636723039754, + "grad_norm": 2.3301889896392822, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7121331691741943, + "num_tokens": 97394301.0, + "step": 3803 + }, + { + "epoch": 0.41774654074236767, + "grad_norm": 2.225137710571289, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7465615272521973, + "num_tokens": 97418841.0, + "step": 3804 + }, + { + "epoch": 0.4178563584449813, + "grad_norm": 2.203500270843506, + "learning_rate": 1e-06, + "loss": 1.0377, + "mean_token_accuracy": 0.7027652263641357, + "num_tokens": 97444750.0, + "step": 3805 + }, + { + "epoch": 0.417966176147595, + "grad_norm": 2.3723297119140625, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7059270143508911, + "num_tokens": 97467366.0, + "step": 3806 + }, + { + "epoch": 0.41807599385020866, + "grad_norm": 1.9859036207199097, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6973803043365479, + "num_tokens": 97497900.0, + "step": 3807 + }, + { + "epoch": 0.4181858115528223, + "grad_norm": 2.194401264190674, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7323644161224365, + "num_tokens": 97521940.0, + "step": 3808 + }, + { + "epoch": 0.41829562925543595, + "grad_norm": 2.4349069595336914, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.6955092549324036, + "num_tokens": 97543724.0, + "step": 3809 + }, + { + "epoch": 0.41840544695804965, + "grad_norm": 2.309347629547119, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6894010901451111, + "num_tokens": 97568458.0, + "step": 3810 + }, + { + "epoch": 0.4185152646606633, + "grad_norm": 1.8450982570648193, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7266793251037598, + "num_tokens": 97603579.0, + "step": 3811 + }, + { + "epoch": 0.41862508236327695, + "grad_norm": 2.2291336059570312, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.7020902633666992, + "num_tokens": 97628329.0, + "step": 3812 + }, + { + "epoch": 0.41873490006589065, + "grad_norm": 2.17704439163208, + "learning_rate": 1e-06, + "loss": 1.0367, + "mean_token_accuracy": 0.6827718019485474, + "num_tokens": 97655524.0, + "step": 3813 + }, + { + "epoch": 0.4188447177685043, + "grad_norm": 2.4184110164642334, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.6950052976608276, + "num_tokens": 97678283.0, + "step": 3814 + }, + { + "epoch": 0.41895453547111794, + "grad_norm": 1.8957537412643433, + "learning_rate": 1e-06, + "loss": 1.0294, + "mean_token_accuracy": 0.6874992847442627, + "num_tokens": 97711330.0, + "step": 3815 + }, + { + "epoch": 0.4190643531737316, + "grad_norm": 2.22017240524292, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6966373920440674, + "num_tokens": 97735387.0, + "step": 3816 + }, + { + "epoch": 0.4191741708763453, + "grad_norm": 2.1954097747802734, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7134337425231934, + "num_tokens": 97760954.0, + "step": 3817 + }, + { + "epoch": 0.41928398857895893, + "grad_norm": 2.135093927383423, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7227746248245239, + "num_tokens": 97788165.0, + "step": 3818 + }, + { + "epoch": 0.4193938062815726, + "grad_norm": 2.511624574661255, + "learning_rate": 1e-06, + "loss": 1.0246, + "mean_token_accuracy": 0.6920559406280518, + "num_tokens": 97809951.0, + "step": 3819 + }, + { + "epoch": 0.4195036239841862, + "grad_norm": 2.356942892074585, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.7038435339927673, + "num_tokens": 97835344.0, + "step": 3820 + }, + { + "epoch": 0.4196134416867999, + "grad_norm": 2.2219321727752686, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7254651188850403, + "num_tokens": 97859539.0, + "step": 3821 + }, + { + "epoch": 0.41972325938941357, + "grad_norm": 2.4358086585998535, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7269684076309204, + "num_tokens": 97879899.0, + "step": 3822 + }, + { + "epoch": 0.4198330770920272, + "grad_norm": 2.300492525100708, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.7063794732093811, + "num_tokens": 97903505.0, + "step": 3823 + }, + { + "epoch": 0.4199428947946409, + "grad_norm": 2.237395763397217, + "learning_rate": 1e-06, + "loss": 1.0936, + "mean_token_accuracy": 0.6749124526977539, + "num_tokens": 97929264.0, + "step": 3824 + }, + { + "epoch": 0.42005271249725457, + "grad_norm": 2.3898556232452393, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.728837788105011, + "num_tokens": 97950969.0, + "step": 3825 + }, + { + "epoch": 0.4201625301998682, + "grad_norm": 2.3746368885040283, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7294958829879761, + "num_tokens": 97972702.0, + "step": 3826 + }, + { + "epoch": 0.42027234790248186, + "grad_norm": 2.3122775554656982, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6940674185752869, + "num_tokens": 97998480.0, + "step": 3827 + }, + { + "epoch": 0.42038216560509556, + "grad_norm": 2.535627603530884, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7041248083114624, + "num_tokens": 98018890.0, + "step": 3828 + }, + { + "epoch": 0.4204919833077092, + "grad_norm": 2.1706976890563965, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.6975131034851074, + "num_tokens": 98046046.0, + "step": 3829 + }, + { + "epoch": 0.42060180101032285, + "grad_norm": 2.3042094707489014, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.682318389415741, + "num_tokens": 98069852.0, + "step": 3830 + }, + { + "epoch": 0.42071161871293655, + "grad_norm": 2.5199742317199707, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7172983884811401, + "num_tokens": 98090977.0, + "step": 3831 + }, + { + "epoch": 0.4208214364155502, + "grad_norm": 2.135251760482788, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6973243951797485, + "num_tokens": 98116129.0, + "step": 3832 + }, + { + "epoch": 0.42093125411816384, + "grad_norm": 2.277297019958496, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7091292142868042, + "num_tokens": 98140570.0, + "step": 3833 + }, + { + "epoch": 0.4210410718207775, + "grad_norm": 2.3767354488372803, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.6911053657531738, + "num_tokens": 98164434.0, + "step": 3834 + }, + { + "epoch": 0.4211508895233912, + "grad_norm": 2.132850408554077, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.702074408531189, + "num_tokens": 98192827.0, + "step": 3835 + }, + { + "epoch": 0.42126070722600484, + "grad_norm": 2.1570019721984863, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7154829502105713, + "num_tokens": 98219161.0, + "step": 3836 + }, + { + "epoch": 0.4213705249286185, + "grad_norm": 2.2040507793426514, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6940227746963501, + "num_tokens": 98246557.0, + "step": 3837 + }, + { + "epoch": 0.42148034263123213, + "grad_norm": 2.1397550106048584, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7156887054443359, + "num_tokens": 98274408.0, + "step": 3838 + }, + { + "epoch": 0.42159016033384583, + "grad_norm": 1.9310510158538818, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7109629511833191, + "num_tokens": 98303752.0, + "step": 3839 + }, + { + "epoch": 0.4216999780364595, + "grad_norm": 2.354940414428711, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7045211791992188, + "num_tokens": 98327369.0, + "step": 3840 + }, + { + "epoch": 0.4218097957390731, + "grad_norm": 2.2945752143859863, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7076801657676697, + "num_tokens": 98349819.0, + "step": 3841 + }, + { + "epoch": 0.4219196134416868, + "grad_norm": 1.9414116144180298, + "learning_rate": 1e-06, + "loss": 1.0751, + "mean_token_accuracy": 0.6830846071243286, + "num_tokens": 98384881.0, + "step": 3842 + }, + { + "epoch": 0.42202943114430047, + "grad_norm": 2.0406553745269775, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7202606797218323, + "num_tokens": 98414049.0, + "step": 3843 + }, + { + "epoch": 0.4221392488469141, + "grad_norm": 2.195946455001831, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.7014151811599731, + "num_tokens": 98438783.0, + "step": 3844 + }, + { + "epoch": 0.42224906654952776, + "grad_norm": 2.009624481201172, + "learning_rate": 1e-06, + "loss": 1.0657, + "mean_token_accuracy": 0.6818872690200806, + "num_tokens": 98468288.0, + "step": 3845 + }, + { + "epoch": 0.42235888425214146, + "grad_norm": 2.208824634552002, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7014026641845703, + "num_tokens": 98495510.0, + "step": 3846 + }, + { + "epoch": 0.4224687019547551, + "grad_norm": 2.2018043994903564, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.715190052986145, + "num_tokens": 98518959.0, + "step": 3847 + }, + { + "epoch": 0.42257851965736876, + "grad_norm": 2.172227621078491, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.7087382078170776, + "num_tokens": 98546070.0, + "step": 3848 + }, + { + "epoch": 0.4226883373599824, + "grad_norm": 2.2106449604034424, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6962527632713318, + "num_tokens": 98573111.0, + "step": 3849 + }, + { + "epoch": 0.4227981550625961, + "grad_norm": 2.104707717895508, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.6994740962982178, + "num_tokens": 98600827.0, + "step": 3850 + }, + { + "epoch": 0.42290797276520975, + "grad_norm": 2.338473320007324, + "learning_rate": 1e-06, + "loss": 1.0783, + "mean_token_accuracy": 0.6888720393180847, + "num_tokens": 98625388.0, + "step": 3851 + }, + { + "epoch": 0.4230177904678234, + "grad_norm": 2.5196986198425293, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7233381271362305, + "num_tokens": 98647599.0, + "step": 3852 + }, + { + "epoch": 0.4231276081704371, + "grad_norm": 2.250554323196411, + "learning_rate": 1e-06, + "loss": 1.0215, + "mean_token_accuracy": 0.6890132427215576, + "num_tokens": 98672674.0, + "step": 3853 + }, + { + "epoch": 0.42323742587305074, + "grad_norm": 2.219836473464966, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7060853838920593, + "num_tokens": 98698159.0, + "step": 3854 + }, + { + "epoch": 0.4233472435756644, + "grad_norm": 2.2850348949432373, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.709776759147644, + "num_tokens": 98721588.0, + "step": 3855 + }, + { + "epoch": 0.42345706127827804, + "grad_norm": 2.3805956840515137, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7046313285827637, + "num_tokens": 98744775.0, + "step": 3856 + }, + { + "epoch": 0.42356687898089174, + "grad_norm": 2.251952648162842, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7070013284683228, + "num_tokens": 98768480.0, + "step": 3857 + }, + { + "epoch": 0.4236766966835054, + "grad_norm": 2.252865791320801, + "learning_rate": 1e-06, + "loss": 1.0373, + "mean_token_accuracy": 0.6937950849533081, + "num_tokens": 98794544.0, + "step": 3858 + }, + { + "epoch": 0.42378651438611903, + "grad_norm": 2.249753475189209, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7194045782089233, + "num_tokens": 98819302.0, + "step": 3859 + }, + { + "epoch": 0.42389633208873273, + "grad_norm": 1.974853754043579, + "learning_rate": 1e-06, + "loss": 1.0239, + "mean_token_accuracy": 0.6944924592971802, + "num_tokens": 98850359.0, + "step": 3860 + }, + { + "epoch": 0.4240061497913464, + "grad_norm": 2.113518714904785, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7037311792373657, + "num_tokens": 98876329.0, + "step": 3861 + }, + { + "epoch": 0.42411596749396, + "grad_norm": 2.0707757472991943, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.7060620784759521, + "num_tokens": 98905952.0, + "step": 3862 + }, + { + "epoch": 0.42422578519657367, + "grad_norm": 2.6847054958343506, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7118833065032959, + "num_tokens": 98924924.0, + "step": 3863 + }, + { + "epoch": 0.42433560289918737, + "grad_norm": 2.0166561603546143, + "learning_rate": 1e-06, + "loss": 1.0542, + "mean_token_accuracy": 0.6787384748458862, + "num_tokens": 98955708.0, + "step": 3864 + }, + { + "epoch": 0.424445420601801, + "grad_norm": 2.353560447692871, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7039082050323486, + "num_tokens": 98978685.0, + "step": 3865 + }, + { + "epoch": 0.42455523830441466, + "grad_norm": 2.096263885498047, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6931819915771484, + "num_tokens": 99008603.0, + "step": 3866 + }, + { + "epoch": 0.4246650560070283, + "grad_norm": 2.078904390335083, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6933529376983643, + "num_tokens": 99038079.0, + "step": 3867 + }, + { + "epoch": 0.424774873709642, + "grad_norm": 2.061742067337036, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7251079082489014, + "num_tokens": 99065925.0, + "step": 3868 + }, + { + "epoch": 0.42488469141225566, + "grad_norm": 2.3937671184539795, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7285559177398682, + "num_tokens": 99086780.0, + "step": 3869 + }, + { + "epoch": 0.4249945091148693, + "grad_norm": 2.104968547821045, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.6970185041427612, + "num_tokens": 99114394.0, + "step": 3870 + }, + { + "epoch": 0.425104326817483, + "grad_norm": 2.4052135944366455, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7132461071014404, + "num_tokens": 99137927.0, + "step": 3871 + }, + { + "epoch": 0.42521414452009665, + "grad_norm": 2.1794111728668213, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.7001258730888367, + "num_tokens": 99165814.0, + "step": 3872 + }, + { + "epoch": 0.4253239622227103, + "grad_norm": 2.1809842586517334, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7055985927581787, + "num_tokens": 99191719.0, + "step": 3873 + }, + { + "epoch": 0.42543377992532394, + "grad_norm": 2.4887022972106934, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.7052029371261597, + "num_tokens": 99212947.0, + "step": 3874 + }, + { + "epoch": 0.42554359762793764, + "grad_norm": 2.155113697052002, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6922711133956909, + "num_tokens": 99242797.0, + "step": 3875 + }, + { + "epoch": 0.4256534153305513, + "grad_norm": 1.9480340480804443, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6941491365432739, + "num_tokens": 99277816.0, + "step": 3876 + }, + { + "epoch": 0.42576323303316493, + "grad_norm": 2.162501573562622, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.6990388631820679, + "num_tokens": 99302939.0, + "step": 3877 + }, + { + "epoch": 0.4258730507357786, + "grad_norm": 2.5222511291503906, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.718017041683197, + "num_tokens": 99322092.0, + "step": 3878 + }, + { + "epoch": 0.4259828684383923, + "grad_norm": 2.339360237121582, + "learning_rate": 1e-06, + "loss": 1.0279, + "mean_token_accuracy": 0.6864201426506042, + "num_tokens": 99346630.0, + "step": 3879 + }, + { + "epoch": 0.4260926861410059, + "grad_norm": 2.5056838989257812, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6888845562934875, + "num_tokens": 99369221.0, + "step": 3880 + }, + { + "epoch": 0.4262025038436196, + "grad_norm": 2.1316282749176025, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7144439220428467, + "num_tokens": 99395363.0, + "step": 3881 + }, + { + "epoch": 0.4263123215462333, + "grad_norm": 2.1017346382141113, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6958400011062622, + "num_tokens": 99421743.0, + "step": 3882 + }, + { + "epoch": 0.4264221392488469, + "grad_norm": 2.2241623401641846, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6925879120826721, + "num_tokens": 99449732.0, + "step": 3883 + }, + { + "epoch": 0.42653195695146057, + "grad_norm": 2.409569025039673, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.725056529045105, + "num_tokens": 99471520.0, + "step": 3884 + }, + { + "epoch": 0.4266417746540742, + "grad_norm": 2.3101696968078613, + "learning_rate": 1e-06, + "loss": 1.0412, + "mean_token_accuracy": 0.6892898082733154, + "num_tokens": 99498788.0, + "step": 3885 + }, + { + "epoch": 0.4267515923566879, + "grad_norm": 2.040825366973877, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7156553268432617, + "num_tokens": 99525264.0, + "step": 3886 + }, + { + "epoch": 0.42686141005930156, + "grad_norm": 2.1019139289855957, + "learning_rate": 1e-06, + "loss": 1.1046, + "mean_token_accuracy": 0.6732016801834106, + "num_tokens": 99553781.0, + "step": 3887 + }, + { + "epoch": 0.4269712277619152, + "grad_norm": 2.2588143348693848, + "learning_rate": 1e-06, + "loss": 1.0357, + "mean_token_accuracy": 0.6856173276901245, + "num_tokens": 99579224.0, + "step": 3888 + }, + { + "epoch": 0.4270810454645289, + "grad_norm": 2.2949209213256836, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6992138624191284, + "num_tokens": 99603232.0, + "step": 3889 + }, + { + "epoch": 0.42719086316714255, + "grad_norm": 2.248359441757202, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6903436183929443, + "num_tokens": 99629103.0, + "step": 3890 + }, + { + "epoch": 0.4273006808697562, + "grad_norm": 2.401566982269287, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7061843276023865, + "num_tokens": 99650970.0, + "step": 3891 + }, + { + "epoch": 0.42741049857236985, + "grad_norm": 2.3997552394866943, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7302408814430237, + "num_tokens": 99671901.0, + "step": 3892 + }, + { + "epoch": 0.42752031627498355, + "grad_norm": 2.4191367626190186, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7117196917533875, + "num_tokens": 99695154.0, + "step": 3893 + }, + { + "epoch": 0.4276301339775972, + "grad_norm": 1.9416677951812744, + "learning_rate": 1e-06, + "loss": 1.097, + "mean_token_accuracy": 0.6771408319473267, + "num_tokens": 99728970.0, + "step": 3894 + }, + { + "epoch": 0.42773995168021084, + "grad_norm": 2.0005972385406494, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.6967999339103699, + "num_tokens": 99761178.0, + "step": 3895 + }, + { + "epoch": 0.4278497693828245, + "grad_norm": 2.520127534866333, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7128916382789612, + "num_tokens": 99780944.0, + "step": 3896 + }, + { + "epoch": 0.4279595870854382, + "grad_norm": 2.564164638519287, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7051445245742798, + "num_tokens": 99801798.0, + "step": 3897 + }, + { + "epoch": 0.42806940478805183, + "grad_norm": 1.9024924039840698, + "learning_rate": 1e-06, + "loss": 1.0462, + "mean_token_accuracy": 0.683651328086853, + "num_tokens": 99836652.0, + "step": 3898 + }, + { + "epoch": 0.4281792224906655, + "grad_norm": 2.3817684650421143, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7062523365020752, + "num_tokens": 99859649.0, + "step": 3899 + }, + { + "epoch": 0.4282890401932792, + "grad_norm": 2.0464987754821777, + "learning_rate": 1e-06, + "loss": 1.1153, + "mean_token_accuracy": 0.6714529395103455, + "num_tokens": 99892321.0, + "step": 3900 + }, + { + "epoch": 0.4283988578958928, + "grad_norm": 2.720797061920166, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7233118414878845, + "num_tokens": 99909724.0, + "step": 3901 + }, + { + "epoch": 0.42850867559850647, + "grad_norm": 1.847222089767456, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7133539915084839, + "num_tokens": 99943448.0, + "step": 3902 + }, + { + "epoch": 0.4286184933011201, + "grad_norm": 2.3402345180511475, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6988190412521362, + "num_tokens": 99967635.0, + "step": 3903 + }, + { + "epoch": 0.4287283110037338, + "grad_norm": 2.1216623783111572, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7214303016662598, + "num_tokens": 99994039.0, + "step": 3904 + }, + { + "epoch": 0.42883812870634747, + "grad_norm": 2.204210042953491, + "learning_rate": 1e-06, + "loss": 1.0509, + "mean_token_accuracy": 0.6819009184837341, + "num_tokens": 100021293.0, + "step": 3905 + }, + { + "epoch": 0.4289479464089611, + "grad_norm": 2.076267957687378, + "learning_rate": 1e-06, + "loss": 1.0663, + "mean_token_accuracy": 0.6816278100013733, + "num_tokens": 100050439.0, + "step": 3906 + }, + { + "epoch": 0.4290577641115748, + "grad_norm": 2.0023627281188965, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6996927261352539, + "num_tokens": 100082042.0, + "step": 3907 + }, + { + "epoch": 0.42916758181418846, + "grad_norm": 2.446040391921997, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7090086936950684, + "num_tokens": 100104539.0, + "step": 3908 + }, + { + "epoch": 0.4292773995168021, + "grad_norm": 2.1375255584716797, + "learning_rate": 1e-06, + "loss": 1.0565, + "mean_token_accuracy": 0.6829383373260498, + "num_tokens": 100132129.0, + "step": 3909 + }, + { + "epoch": 0.42938721721941575, + "grad_norm": 2.247748374938965, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7306541204452515, + "num_tokens": 100155144.0, + "step": 3910 + }, + { + "epoch": 0.42949703492202945, + "grad_norm": 2.4408395290374756, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7221277356147766, + "num_tokens": 100177206.0, + "step": 3911 + }, + { + "epoch": 0.4296068526246431, + "grad_norm": 2.35752534866333, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7155782580375671, + "num_tokens": 100200301.0, + "step": 3912 + }, + { + "epoch": 0.42971667032725674, + "grad_norm": 2.029555320739746, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6926708817481995, + "num_tokens": 100227276.0, + "step": 3913 + }, + { + "epoch": 0.4298264880298704, + "grad_norm": 2.4975104331970215, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7105152010917664, + "num_tokens": 100248271.0, + "step": 3914 + }, + { + "epoch": 0.4299363057324841, + "grad_norm": 1.932680368423462, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6922420263290405, + "num_tokens": 100281261.0, + "step": 3915 + }, + { + "epoch": 0.43004612343509774, + "grad_norm": 2.2896804809570312, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7207993268966675, + "num_tokens": 100305239.0, + "step": 3916 + }, + { + "epoch": 0.4301559411377114, + "grad_norm": 2.0476107597351074, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7165505290031433, + "num_tokens": 100332330.0, + "step": 3917 + }, + { + "epoch": 0.4302657588403251, + "grad_norm": 1.9945698976516724, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.6963964700698853, + "num_tokens": 100363892.0, + "step": 3918 + }, + { + "epoch": 0.43037557654293873, + "grad_norm": 2.6283462047576904, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7020035982131958, + "num_tokens": 100383418.0, + "step": 3919 + }, + { + "epoch": 0.4304853942455524, + "grad_norm": 2.2919485569000244, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6995784044265747, + "num_tokens": 100408521.0, + "step": 3920 + }, + { + "epoch": 0.430595211948166, + "grad_norm": 2.456195592880249, + "learning_rate": 1e-06, + "loss": 1.0617, + "mean_token_accuracy": 0.6769440174102783, + "num_tokens": 100433270.0, + "step": 3921 + }, + { + "epoch": 0.4307050296507797, + "grad_norm": 2.1846535205841064, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.700225293636322, + "num_tokens": 100460738.0, + "step": 3922 + }, + { + "epoch": 0.43081484735339337, + "grad_norm": 2.0097906589508057, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6949737071990967, + "num_tokens": 100490483.0, + "step": 3923 + }, + { + "epoch": 0.430924665056007, + "grad_norm": 2.121515989303589, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6976827383041382, + "num_tokens": 100517794.0, + "step": 3924 + }, + { + "epoch": 0.43103448275862066, + "grad_norm": 2.194952964782715, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7153962850570679, + "num_tokens": 100542049.0, + "step": 3925 + }, + { + "epoch": 0.43114430046123436, + "grad_norm": 2.0141780376434326, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.7064210772514343, + "num_tokens": 100573137.0, + "step": 3926 + }, + { + "epoch": 0.431254118163848, + "grad_norm": 2.260106325149536, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.6967786550521851, + "num_tokens": 100598734.0, + "step": 3927 + }, + { + "epoch": 0.43136393586646166, + "grad_norm": 2.4910175800323486, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6979981064796448, + "num_tokens": 100621666.0, + "step": 3928 + }, + { + "epoch": 0.43147375356907536, + "grad_norm": 2.359036445617676, + "learning_rate": 1e-06, + "loss": 1.025, + "mean_token_accuracy": 0.6885892152786255, + "num_tokens": 100645246.0, + "step": 3929 + }, + { + "epoch": 0.431583571271689, + "grad_norm": 2.0347208976745605, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6997200846672058, + "num_tokens": 100675988.0, + "step": 3930 + }, + { + "epoch": 0.43169338897430265, + "grad_norm": 2.5851495265960693, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7178552746772766, + "num_tokens": 100695089.0, + "step": 3931 + }, + { + "epoch": 0.4318032066769163, + "grad_norm": 2.0302884578704834, + "learning_rate": 1e-06, + "loss": 1.0761, + "mean_token_accuracy": 0.6773134469985962, + "num_tokens": 100726607.0, + "step": 3932 + }, + { + "epoch": 0.43191302437953, + "grad_norm": 2.5766751766204834, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7026565074920654, + "num_tokens": 100746271.0, + "step": 3933 + }, + { + "epoch": 0.43202284208214364, + "grad_norm": 2.3263158798217773, + "learning_rate": 1e-06, + "loss": 1.0228, + "mean_token_accuracy": 0.7009091973304749, + "num_tokens": 100772019.0, + "step": 3934 + }, + { + "epoch": 0.4321326597847573, + "grad_norm": 2.1124048233032227, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7235205173492432, + "num_tokens": 100798686.0, + "step": 3935 + }, + { + "epoch": 0.432242477487371, + "grad_norm": 2.440500020980835, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.711582601070404, + "num_tokens": 100818114.0, + "step": 3936 + }, + { + "epoch": 0.43235229518998464, + "grad_norm": 2.5140492916107178, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.695960521697998, + "num_tokens": 100839766.0, + "step": 3937 + }, + { + "epoch": 0.4324621128925983, + "grad_norm": 2.344080686569214, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7132061123847961, + "num_tokens": 100862846.0, + "step": 3938 + }, + { + "epoch": 0.43257193059521193, + "grad_norm": 2.41251802444458, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.69632887840271, + "num_tokens": 100884620.0, + "step": 3939 + }, + { + "epoch": 0.43268174829782563, + "grad_norm": 2.0666213035583496, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7205780744552612, + "num_tokens": 100911626.0, + "step": 3940 + }, + { + "epoch": 0.4327915660004393, + "grad_norm": 2.472691297531128, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6952877044677734, + "num_tokens": 100933461.0, + "step": 3941 + }, + { + "epoch": 0.4329013837030529, + "grad_norm": 2.5370821952819824, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7080230712890625, + "num_tokens": 100952548.0, + "step": 3942 + }, + { + "epoch": 0.43301120140566657, + "grad_norm": 2.3831751346588135, + "learning_rate": 1e-06, + "loss": 1.0235, + "mean_token_accuracy": 0.6964505910873413, + "num_tokens": 100976029.0, + "step": 3943 + }, + { + "epoch": 0.43312101910828027, + "grad_norm": 2.3307604789733887, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7084876298904419, + "num_tokens": 100999044.0, + "step": 3944 + }, + { + "epoch": 0.4332308368108939, + "grad_norm": 2.400996446609497, + "learning_rate": 1e-06, + "loss": 1.0313, + "mean_token_accuracy": 0.6965744495391846, + "num_tokens": 101022795.0, + "step": 3945 + }, + { + "epoch": 0.43334065451350756, + "grad_norm": 2.583059787750244, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7172812223434448, + "num_tokens": 101041680.0, + "step": 3946 + }, + { + "epoch": 0.43345047221612126, + "grad_norm": 2.1819357872009277, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.6982999444007874, + "num_tokens": 101067335.0, + "step": 3947 + }, + { + "epoch": 0.4335602899187349, + "grad_norm": 2.2301084995269775, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7150629162788391, + "num_tokens": 101093182.0, + "step": 3948 + }, + { + "epoch": 0.43367010762134856, + "grad_norm": 2.421762704849243, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7112078666687012, + "num_tokens": 101116080.0, + "step": 3949 + }, + { + "epoch": 0.4337799253239622, + "grad_norm": 2.312317371368408, + "learning_rate": 1e-06, + "loss": 1.0196, + "mean_token_accuracy": 0.6911141276359558, + "num_tokens": 101139786.0, + "step": 3950 + }, + { + "epoch": 0.4338897430265759, + "grad_norm": 2.3481974601745605, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7208294868469238, + "num_tokens": 101161260.0, + "step": 3951 + }, + { + "epoch": 0.43399956072918955, + "grad_norm": 2.0813310146331787, + "learning_rate": 1e-06, + "loss": 1.1363, + "mean_token_accuracy": 0.6603164672851562, + "num_tokens": 101192789.0, + "step": 3952 + }, + { + "epoch": 0.4341093784318032, + "grad_norm": 2.7424778938293457, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7184107899665833, + "num_tokens": 101210735.0, + "step": 3953 + }, + { + "epoch": 0.43421919613441684, + "grad_norm": 2.2055234909057617, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6925380229949951, + "num_tokens": 101237005.0, + "step": 3954 + }, + { + "epoch": 0.43432901383703054, + "grad_norm": 2.154796838760376, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.723042368888855, + "num_tokens": 101261378.0, + "step": 3955 + }, + { + "epoch": 0.4344388315396442, + "grad_norm": 2.534477949142456, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.715394139289856, + "num_tokens": 101280442.0, + "step": 3956 + }, + { + "epoch": 0.43454864924225783, + "grad_norm": 2.203252077102661, + "learning_rate": 1e-06, + "loss": 1.0511, + "mean_token_accuracy": 0.6834288239479065, + "num_tokens": 101308668.0, + "step": 3957 + }, + { + "epoch": 0.43465846694487154, + "grad_norm": 2.344876527786255, + "learning_rate": 1e-06, + "loss": 1.045, + "mean_token_accuracy": 0.6873719692230225, + "num_tokens": 101332084.0, + "step": 3958 + }, + { + "epoch": 0.4347682846474852, + "grad_norm": 2.102813720703125, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7081581354141235, + "num_tokens": 101360507.0, + "step": 3959 + }, + { + "epoch": 0.4348781023500988, + "grad_norm": 2.1915171146392822, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7016825675964355, + "num_tokens": 101387807.0, + "step": 3960 + }, + { + "epoch": 0.4349879200527125, + "grad_norm": 2.263011932373047, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.6863179206848145, + "num_tokens": 101412152.0, + "step": 3961 + }, + { + "epoch": 0.4350977377553262, + "grad_norm": 2.580904006958008, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.703020453453064, + "num_tokens": 101432052.0, + "step": 3962 + }, + { + "epoch": 0.4352075554579398, + "grad_norm": 2.489880084991455, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7323125600814819, + "num_tokens": 101451892.0, + "step": 3963 + }, + { + "epoch": 0.43531737316055347, + "grad_norm": 2.1842844486236572, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6963754892349243, + "num_tokens": 101477289.0, + "step": 3964 + }, + { + "epoch": 0.43542719086316717, + "grad_norm": 2.2101244926452637, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.702400803565979, + "num_tokens": 101503106.0, + "step": 3965 + }, + { + "epoch": 0.4355370085657808, + "grad_norm": 2.192701578140259, + "learning_rate": 1e-06, + "loss": 1.0387, + "mean_token_accuracy": 0.686751127243042, + "num_tokens": 101530186.0, + "step": 3966 + }, + { + "epoch": 0.43564682626839446, + "grad_norm": 2.091287851333618, + "learning_rate": 1e-06, + "loss": 1.077, + "mean_token_accuracy": 0.6785337924957275, + "num_tokens": 101559762.0, + "step": 3967 + }, + { + "epoch": 0.4357566439710081, + "grad_norm": 2.2724428176879883, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7160180807113647, + "num_tokens": 101586554.0, + "step": 3968 + }, + { + "epoch": 0.4358664616736218, + "grad_norm": 2.1524407863616943, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6906631588935852, + "num_tokens": 101614382.0, + "step": 3969 + }, + { + "epoch": 0.43597627937623545, + "grad_norm": 2.3090362548828125, + "learning_rate": 1e-06, + "loss": 1.0395, + "mean_token_accuracy": 0.6883717179298401, + "num_tokens": 101640030.0, + "step": 3970 + }, + { + "epoch": 0.4360860970788491, + "grad_norm": 2.6386382579803467, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7226160764694214, + "num_tokens": 101658307.0, + "step": 3971 + }, + { + "epoch": 0.43619591478146275, + "grad_norm": 2.13224196434021, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7056834101676941, + "num_tokens": 101683271.0, + "step": 3972 + }, + { + "epoch": 0.43630573248407645, + "grad_norm": 1.9471004009246826, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.7014529705047607, + "num_tokens": 101714614.0, + "step": 3973 + }, + { + "epoch": 0.4364155501866901, + "grad_norm": 2.0983805656433105, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7054535746574402, + "num_tokens": 101741293.0, + "step": 3974 + }, + { + "epoch": 0.43652536788930374, + "grad_norm": 2.3659679889678955, + "learning_rate": 1e-06, + "loss": 1.1341, + "mean_token_accuracy": 0.670459508895874, + "num_tokens": 101766479.0, + "step": 3975 + }, + { + "epoch": 0.43663518559191744, + "grad_norm": 2.401094436645508, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7130228281021118, + "num_tokens": 101788517.0, + "step": 3976 + }, + { + "epoch": 0.4367450032945311, + "grad_norm": 2.483534336090088, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.722079336643219, + "num_tokens": 101810514.0, + "step": 3977 + }, + { + "epoch": 0.43685482099714473, + "grad_norm": 2.1445693969726562, + "learning_rate": 1e-06, + "loss": 1.0372, + "mean_token_accuracy": 0.6870694756507874, + "num_tokens": 101839995.0, + "step": 3978 + }, + { + "epoch": 0.4369646386997584, + "grad_norm": 2.4597885608673096, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7184410095214844, + "num_tokens": 101860103.0, + "step": 3979 + }, + { + "epoch": 0.4370744564023721, + "grad_norm": 2.303295850753784, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7286785840988159, + "num_tokens": 101881515.0, + "step": 3980 + }, + { + "epoch": 0.4371842741049857, + "grad_norm": 2.1789493560791016, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.707115650177002, + "num_tokens": 101906495.0, + "step": 3981 + }, + { + "epoch": 0.43729409180759937, + "grad_norm": 2.13749098777771, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7110475301742554, + "num_tokens": 101934253.0, + "step": 3982 + }, + { + "epoch": 0.4374039095102131, + "grad_norm": 2.2770049571990967, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7022979259490967, + "num_tokens": 101957999.0, + "step": 3983 + }, + { + "epoch": 0.4375137272128267, + "grad_norm": 2.178330421447754, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.707451343536377, + "num_tokens": 101983591.0, + "step": 3984 + }, + { + "epoch": 0.43762354491544037, + "grad_norm": 2.4676856994628906, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7043091058731079, + "num_tokens": 102005859.0, + "step": 3985 + }, + { + "epoch": 0.437733362618054, + "grad_norm": 2.160230875015259, + "learning_rate": 1e-06, + "loss": 1.0586, + "mean_token_accuracy": 0.6882517337799072, + "num_tokens": 102032475.0, + "step": 3986 + }, + { + "epoch": 0.4378431803206677, + "grad_norm": 2.2880024909973145, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6873149275779724, + "num_tokens": 102056438.0, + "step": 3987 + }, + { + "epoch": 0.43795299802328136, + "grad_norm": 2.464076280593872, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6983082294464111, + "num_tokens": 102078626.0, + "step": 3988 + }, + { + "epoch": 0.438062815725895, + "grad_norm": 2.2300190925598145, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7290127277374268, + "num_tokens": 102104078.0, + "step": 3989 + }, + { + "epoch": 0.43817263342850865, + "grad_norm": 2.255232095718384, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.688936710357666, + "num_tokens": 102129013.0, + "step": 3990 + }, + { + "epoch": 0.43828245113112235, + "grad_norm": 2.1939291954040527, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.718177080154419, + "num_tokens": 102154493.0, + "step": 3991 + }, + { + "epoch": 0.438392268833736, + "grad_norm": 2.1541357040405273, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7075307965278625, + "num_tokens": 102181402.0, + "step": 3992 + }, + { + "epoch": 0.43850208653634964, + "grad_norm": 2.233534812927246, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6939427852630615, + "num_tokens": 102207189.0, + "step": 3993 + }, + { + "epoch": 0.43861190423896335, + "grad_norm": 1.8042258024215698, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7062901258468628, + "num_tokens": 102242656.0, + "step": 3994 + }, + { + "epoch": 0.438721721941577, + "grad_norm": 2.575326919555664, + "learning_rate": 1e-06, + "loss": 1.0531, + "mean_token_accuracy": 0.7041333913803101, + "num_tokens": 102263471.0, + "step": 3995 + }, + { + "epoch": 0.43883153964419064, + "grad_norm": 2.327669382095337, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7156054377555847, + "num_tokens": 102284802.0, + "step": 3996 + }, + { + "epoch": 0.4389413573468043, + "grad_norm": 2.4928932189941406, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7080784440040588, + "num_tokens": 102304644.0, + "step": 3997 + }, + { + "epoch": 0.439051175049418, + "grad_norm": 2.3832035064697266, + "learning_rate": 1e-06, + "loss": 1.047, + "mean_token_accuracy": 0.6812077760696411, + "num_tokens": 102328634.0, + "step": 3998 + }, + { + "epoch": 0.43916099275203163, + "grad_norm": 2.4415125846862793, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.7064037322998047, + "num_tokens": 102349957.0, + "step": 3999 + }, + { + "epoch": 0.4392708104546453, + "grad_norm": 2.193856716156006, + "learning_rate": 1e-06, + "loss": 1.135, + "mean_token_accuracy": 0.6711710691452026, + "num_tokens": 102378739.0, + "step": 4000 + }, + { + "epoch": 0.4393806281572589, + "grad_norm": 2.2405929565429688, + "learning_rate": 1e-06, + "loss": 1.0866, + "mean_token_accuracy": 0.6741837859153748, + "num_tokens": 102405567.0, + "step": 4001 + }, + { + "epoch": 0.4394904458598726, + "grad_norm": 2.3521220684051514, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6970819234848022, + "num_tokens": 102427901.0, + "step": 4002 + }, + { + "epoch": 0.43960026356248627, + "grad_norm": 2.1715681552886963, + "learning_rate": 1e-06, + "loss": 1.0593, + "mean_token_accuracy": 0.6858404874801636, + "num_tokens": 102456854.0, + "step": 4003 + }, + { + "epoch": 0.4397100812650999, + "grad_norm": 2.4344303607940674, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.6990203857421875, + "num_tokens": 102477638.0, + "step": 4004 + }, + { + "epoch": 0.4398198989677136, + "grad_norm": 2.2040975093841553, + "learning_rate": 1e-06, + "loss": 1.0195, + "mean_token_accuracy": 0.6913173198699951, + "num_tokens": 102503005.0, + "step": 4005 + }, + { + "epoch": 0.43992971667032726, + "grad_norm": 2.291268825531006, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.703965961933136, + "num_tokens": 102527829.0, + "step": 4006 + }, + { + "epoch": 0.4400395343729409, + "grad_norm": 2.0174942016601562, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7106987237930298, + "num_tokens": 102556883.0, + "step": 4007 + }, + { + "epoch": 0.44014935207555456, + "grad_norm": 2.597874641418457, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.6991869211196899, + "num_tokens": 102577452.0, + "step": 4008 + }, + { + "epoch": 0.44025916977816826, + "grad_norm": 2.2454159259796143, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.7022345066070557, + "num_tokens": 102603662.0, + "step": 4009 + }, + { + "epoch": 0.4403689874807819, + "grad_norm": 2.0112380981445312, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.6945136785507202, + "num_tokens": 102634735.0, + "step": 4010 + }, + { + "epoch": 0.44047880518339555, + "grad_norm": 2.1676547527313232, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6982309818267822, + "num_tokens": 102663554.0, + "step": 4011 + }, + { + "epoch": 0.44058862288600925, + "grad_norm": 2.130587100982666, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.6999205350875854, + "num_tokens": 102688547.0, + "step": 4012 + }, + { + "epoch": 0.4406984405886229, + "grad_norm": 2.475724697113037, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.7030031681060791, + "num_tokens": 102708268.0, + "step": 4013 + }, + { + "epoch": 0.44080825829123654, + "grad_norm": 2.187812566757202, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6967290639877319, + "num_tokens": 102733039.0, + "step": 4014 + }, + { + "epoch": 0.4409180759938502, + "grad_norm": 2.050180196762085, + "learning_rate": 1e-06, + "loss": 1.0317, + "mean_token_accuracy": 0.6953500509262085, + "num_tokens": 102763130.0, + "step": 4015 + }, + { + "epoch": 0.4410278936964639, + "grad_norm": 2.304525852203369, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7209820747375488, + "num_tokens": 102786503.0, + "step": 4016 + }, + { + "epoch": 0.44113771139907754, + "grad_norm": 2.370222806930542, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7475333213806152, + "num_tokens": 102807973.0, + "step": 4017 + }, + { + "epoch": 0.4412475291016912, + "grad_norm": 1.9132320880889893, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6960305571556091, + "num_tokens": 102839755.0, + "step": 4018 + }, + { + "epoch": 0.44135734680430483, + "grad_norm": 2.2507643699645996, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6961678266525269, + "num_tokens": 102865717.0, + "step": 4019 + }, + { + "epoch": 0.44146716450691853, + "grad_norm": 2.2327089309692383, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.746909499168396, + "num_tokens": 102891734.0, + "step": 4020 + }, + { + "epoch": 0.4415769822095322, + "grad_norm": 2.7256269454956055, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7214275598526001, + "num_tokens": 102908477.0, + "step": 4021 + }, + { + "epoch": 0.4416867999121458, + "grad_norm": 2.1574044227600098, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.7013901472091675, + "num_tokens": 102934029.0, + "step": 4022 + }, + { + "epoch": 0.4417966176147595, + "grad_norm": 2.0967600345611572, + "learning_rate": 1e-06, + "loss": 1.0479, + "mean_token_accuracy": 0.6807947158813477, + "num_tokens": 102963145.0, + "step": 4023 + }, + { + "epoch": 0.44190643531737317, + "grad_norm": 2.5056772232055664, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.6961912512779236, + "num_tokens": 102984174.0, + "step": 4024 + }, + { + "epoch": 0.4420162530199868, + "grad_norm": 2.248877763748169, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6952719688415527, + "num_tokens": 103010655.0, + "step": 4025 + }, + { + "epoch": 0.44212607072260046, + "grad_norm": 2.3328537940979004, + "learning_rate": 1e-06, + "loss": 1.0533, + "mean_token_accuracy": 0.6811414361000061, + "num_tokens": 103035713.0, + "step": 4026 + }, + { + "epoch": 0.44223588842521416, + "grad_norm": 2.625998020172119, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.707953929901123, + "num_tokens": 103055460.0, + "step": 4027 + }, + { + "epoch": 0.4423457061278278, + "grad_norm": 2.356940507888794, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7135047316551208, + "num_tokens": 103077566.0, + "step": 4028 + }, + { + "epoch": 0.44245552383044146, + "grad_norm": 2.343384265899658, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.7071115374565125, + "num_tokens": 103100604.0, + "step": 4029 + }, + { + "epoch": 0.4425653415330551, + "grad_norm": 1.959157109260559, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.715433657169342, + "num_tokens": 103132223.0, + "step": 4030 + }, + { + "epoch": 0.4426751592356688, + "grad_norm": 2.095362663269043, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6905267238616943, + "num_tokens": 103162231.0, + "step": 4031 + }, + { + "epoch": 0.44278497693828245, + "grad_norm": 2.0242438316345215, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6989785432815552, + "num_tokens": 103191566.0, + "step": 4032 + }, + { + "epoch": 0.4428947946408961, + "grad_norm": 2.6189727783203125, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.6929826736450195, + "num_tokens": 103210023.0, + "step": 4033 + }, + { + "epoch": 0.4430046123435098, + "grad_norm": 2.304415702819824, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.7056474089622498, + "num_tokens": 103233840.0, + "step": 4034 + }, + { + "epoch": 0.44311443004612344, + "grad_norm": 2.1971561908721924, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7006008625030518, + "num_tokens": 103259748.0, + "step": 4035 + }, + { + "epoch": 0.4432242477487371, + "grad_norm": 2.2750117778778076, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7093414664268494, + "num_tokens": 103284359.0, + "step": 4036 + }, + { + "epoch": 0.44333406545135073, + "grad_norm": 2.155282497406006, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6953498125076294, + "num_tokens": 103311713.0, + "step": 4037 + }, + { + "epoch": 0.44344388315396444, + "grad_norm": 2.333686113357544, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6985125541687012, + "num_tokens": 103336697.0, + "step": 4038 + }, + { + "epoch": 0.4435537008565781, + "grad_norm": 2.3956172466278076, + "learning_rate": 1e-06, + "loss": 1.0738, + "mean_token_accuracy": 0.687961220741272, + "num_tokens": 103358792.0, + "step": 4039 + }, + { + "epoch": 0.4436635185591917, + "grad_norm": 1.7846094369888306, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7036382555961609, + "num_tokens": 103394413.0, + "step": 4040 + }, + { + "epoch": 0.44377333626180543, + "grad_norm": 2.4136478900909424, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7193092107772827, + "num_tokens": 103418081.0, + "step": 4041 + }, + { + "epoch": 0.4438831539644191, + "grad_norm": 2.3639440536499023, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.7004008293151855, + "num_tokens": 103441713.0, + "step": 4042 + }, + { + "epoch": 0.4439929716670327, + "grad_norm": 2.176814317703247, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6916204690933228, + "num_tokens": 103467960.0, + "step": 4043 + }, + { + "epoch": 0.44410278936964637, + "grad_norm": 2.3470773696899414, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7292490601539612, + "num_tokens": 103488494.0, + "step": 4044 + }, + { + "epoch": 0.44421260707226007, + "grad_norm": 2.341179370880127, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7101912498474121, + "num_tokens": 103512009.0, + "step": 4045 + }, + { + "epoch": 0.4443224247748737, + "grad_norm": 2.4527876377105713, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.6926063299179077, + "num_tokens": 103534682.0, + "step": 4046 + }, + { + "epoch": 0.44443224247748736, + "grad_norm": 2.2307825088500977, + "learning_rate": 1e-06, + "loss": 1.0454, + "mean_token_accuracy": 0.6927740573883057, + "num_tokens": 103561908.0, + "step": 4047 + }, + { + "epoch": 0.444542060180101, + "grad_norm": 2.319509983062744, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7041622400283813, + "num_tokens": 103583732.0, + "step": 4048 + }, + { + "epoch": 0.4446518778827147, + "grad_norm": 2.044299364089966, + "learning_rate": 1e-06, + "loss": 1.0216, + "mean_token_accuracy": 0.7054954767227173, + "num_tokens": 103613483.0, + "step": 4049 + }, + { + "epoch": 0.44476169558532835, + "grad_norm": 2.2714080810546875, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.6994599103927612, + "num_tokens": 103638017.0, + "step": 4050 + }, + { + "epoch": 0.444871513287942, + "grad_norm": 2.30259108543396, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6905925869941711, + "num_tokens": 103662294.0, + "step": 4051 + }, + { + "epoch": 0.4449813309905557, + "grad_norm": 2.274263620376587, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7033201456069946, + "num_tokens": 103686620.0, + "step": 4052 + }, + { + "epoch": 0.44509114869316935, + "grad_norm": 2.0745575428009033, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6948579549789429, + "num_tokens": 103715494.0, + "step": 4053 + }, + { + "epoch": 0.445200966395783, + "grad_norm": 2.1413917541503906, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7203166484832764, + "num_tokens": 103743459.0, + "step": 4054 + }, + { + "epoch": 0.44531078409839664, + "grad_norm": 2.29015851020813, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7087686061859131, + "num_tokens": 103768328.0, + "step": 4055 + }, + { + "epoch": 0.44542060180101034, + "grad_norm": 2.1154255867004395, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7148492932319641, + "num_tokens": 103795881.0, + "step": 4056 + }, + { + "epoch": 0.445530419503624, + "grad_norm": 2.2558512687683105, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6874275207519531, + "num_tokens": 103820289.0, + "step": 4057 + }, + { + "epoch": 0.44564023720623763, + "grad_norm": 2.370457172393799, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7100390195846558, + "num_tokens": 103843584.0, + "step": 4058 + }, + { + "epoch": 0.44575005490885133, + "grad_norm": 2.513549566268921, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.6830927729606628, + "num_tokens": 103866126.0, + "step": 4059 + }, + { + "epoch": 0.445859872611465, + "grad_norm": 2.1714046001434326, + "learning_rate": 1e-06, + "loss": 1.0336, + "mean_token_accuracy": 0.6957855820655823, + "num_tokens": 103894284.0, + "step": 4060 + }, + { + "epoch": 0.4459696903140786, + "grad_norm": 2.2812860012054443, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7019413709640503, + "num_tokens": 103918692.0, + "step": 4061 + }, + { + "epoch": 0.44607950801669227, + "grad_norm": 2.1978790760040283, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7095475196838379, + "num_tokens": 103944678.0, + "step": 4062 + }, + { + "epoch": 0.446189325719306, + "grad_norm": 2.2910845279693604, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.711060643196106, + "num_tokens": 103968832.0, + "step": 4063 + }, + { + "epoch": 0.4462991434219196, + "grad_norm": 2.386845111846924, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.6935584545135498, + "num_tokens": 103993087.0, + "step": 4064 + }, + { + "epoch": 0.44640896112453327, + "grad_norm": 2.1211581230163574, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7043286561965942, + "num_tokens": 104020252.0, + "step": 4065 + }, + { + "epoch": 0.4465187788271469, + "grad_norm": 2.087350845336914, + "learning_rate": 1e-06, + "loss": 1.0443, + "mean_token_accuracy": 0.6821994185447693, + "num_tokens": 104048447.0, + "step": 4066 + }, + { + "epoch": 0.4466285965297606, + "grad_norm": 2.0845022201538086, + "learning_rate": 1e-06, + "loss": 1.0272, + "mean_token_accuracy": 0.6841462254524231, + "num_tokens": 104077170.0, + "step": 4067 + }, + { + "epoch": 0.44673841423237426, + "grad_norm": 2.3904213905334473, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7048431634902954, + "num_tokens": 104099746.0, + "step": 4068 + }, + { + "epoch": 0.4468482319349879, + "grad_norm": 2.7038021087646484, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7372509241104126, + "num_tokens": 104119235.0, + "step": 4069 + }, + { + "epoch": 0.4469580496376016, + "grad_norm": 2.1616926193237305, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7012907266616821, + "num_tokens": 104148130.0, + "step": 4070 + }, + { + "epoch": 0.44706786734021525, + "grad_norm": 2.345566511154175, + "learning_rate": 1e-06, + "loss": 1.0717, + "mean_token_accuracy": 0.6803196668624878, + "num_tokens": 104172993.0, + "step": 4071 + }, + { + "epoch": 0.4471776850428289, + "grad_norm": 2.1962573528289795, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7028326392173767, + "num_tokens": 104198187.0, + "step": 4072 + }, + { + "epoch": 0.44728750274544254, + "grad_norm": 2.350041151046753, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7042083740234375, + "num_tokens": 104221984.0, + "step": 4073 + }, + { + "epoch": 0.44739732044805625, + "grad_norm": 2.4701359272003174, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.7113458514213562, + "num_tokens": 104245253.0, + "step": 4074 + }, + { + "epoch": 0.4475071381506699, + "grad_norm": 2.4364542961120605, + "learning_rate": 1e-06, + "loss": 1.0852, + "mean_token_accuracy": 0.6743633151054382, + "num_tokens": 104269801.0, + "step": 4075 + }, + { + "epoch": 0.44761695585328354, + "grad_norm": 2.395345449447632, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7170647382736206, + "num_tokens": 104293101.0, + "step": 4076 + }, + { + "epoch": 0.4477267735558972, + "grad_norm": 2.537252426147461, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7176424264907837, + "num_tokens": 104313822.0, + "step": 4077 + }, + { + "epoch": 0.4478365912585109, + "grad_norm": 2.3004343509674072, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7063226699829102, + "num_tokens": 104338757.0, + "step": 4078 + }, + { + "epoch": 0.44794640896112453, + "grad_norm": 2.104797840118408, + "learning_rate": 1e-06, + "loss": 1.0827, + "mean_token_accuracy": 0.685836136341095, + "num_tokens": 104369330.0, + "step": 4079 + }, + { + "epoch": 0.4480562266637382, + "grad_norm": 2.35495924949646, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7017388939857483, + "num_tokens": 104393887.0, + "step": 4080 + }, + { + "epoch": 0.4481660443663519, + "grad_norm": 2.4810855388641357, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7105637788772583, + "num_tokens": 104414517.0, + "step": 4081 + }, + { + "epoch": 0.4482758620689655, + "grad_norm": 2.1952574253082275, + "learning_rate": 1e-06, + "loss": 1.0398, + "mean_token_accuracy": 0.6896750926971436, + "num_tokens": 104439271.0, + "step": 4082 + }, + { + "epoch": 0.44838567977157917, + "grad_norm": 2.2706692218780518, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7058209776878357, + "num_tokens": 104465100.0, + "step": 4083 + }, + { + "epoch": 0.4484954974741928, + "grad_norm": 2.1907124519348145, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7104365825653076, + "num_tokens": 104490370.0, + "step": 4084 + }, + { + "epoch": 0.4486053151768065, + "grad_norm": 2.1303365230560303, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6997249126434326, + "num_tokens": 104518037.0, + "step": 4085 + }, + { + "epoch": 0.44871513287942016, + "grad_norm": 2.1699953079223633, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7006559371948242, + "num_tokens": 104544464.0, + "step": 4086 + }, + { + "epoch": 0.4488249505820338, + "grad_norm": 2.254538059234619, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7160173058509827, + "num_tokens": 104569367.0, + "step": 4087 + }, + { + "epoch": 0.4489347682846475, + "grad_norm": 2.2316572666168213, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7035185098648071, + "num_tokens": 104592243.0, + "step": 4088 + }, + { + "epoch": 0.44904458598726116, + "grad_norm": 2.489755392074585, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7207822203636169, + "num_tokens": 104612305.0, + "step": 4089 + }, + { + "epoch": 0.4491544036898748, + "grad_norm": 2.479572296142578, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.7013394236564636, + "num_tokens": 104634131.0, + "step": 4090 + }, + { + "epoch": 0.44926422139248845, + "grad_norm": 2.134939193725586, + "learning_rate": 1e-06, + "loss": 1.0438, + "mean_token_accuracy": 0.6841717958450317, + "num_tokens": 104662742.0, + "step": 4091 + }, + { + "epoch": 0.44937403909510215, + "grad_norm": 1.8172063827514648, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6995153427124023, + "num_tokens": 104700257.0, + "step": 4092 + }, + { + "epoch": 0.4494838567977158, + "grad_norm": 2.1563398838043213, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7165347337722778, + "num_tokens": 104727956.0, + "step": 4093 + }, + { + "epoch": 0.44959367450032944, + "grad_norm": 2.230828285217285, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7074549198150635, + "num_tokens": 104753134.0, + "step": 4094 + }, + { + "epoch": 0.4497034922029431, + "grad_norm": 2.024421453475952, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7091729640960693, + "num_tokens": 104782824.0, + "step": 4095 + }, + { + "epoch": 0.4498133099055568, + "grad_norm": 2.1863718032836914, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7341672778129578, + "num_tokens": 104807316.0, + "step": 4096 + }, + { + "epoch": 0.44992312760817044, + "grad_norm": 2.285306453704834, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7071800827980042, + "num_tokens": 104833364.0, + "step": 4097 + }, + { + "epoch": 0.4500329453107841, + "grad_norm": 2.1738362312316895, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7211058735847473, + "num_tokens": 104857079.0, + "step": 4098 + }, + { + "epoch": 0.4501427630133978, + "grad_norm": 2.011350393295288, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6946940422058105, + "num_tokens": 104886845.0, + "step": 4099 + }, + { + "epoch": 0.45025258071601143, + "grad_norm": 2.270749568939209, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7059420943260193, + "num_tokens": 104913703.0, + "step": 4100 + }, + { + "epoch": 0.4503623984186251, + "grad_norm": 2.114295721054077, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7116731405258179, + "num_tokens": 104942047.0, + "step": 4101 + }, + { + "epoch": 0.4504722161212387, + "grad_norm": 1.9458644390106201, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7244307994842529, + "num_tokens": 104969319.0, + "step": 4102 + }, + { + "epoch": 0.4505820338238524, + "grad_norm": 2.138903856277466, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.6930498480796814, + "num_tokens": 104997634.0, + "step": 4103 + }, + { + "epoch": 0.45069185152646607, + "grad_norm": 2.1524698734283447, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.690150260925293, + "num_tokens": 105025222.0, + "step": 4104 + }, + { + "epoch": 0.4508016692290797, + "grad_norm": 2.4194388389587402, + "learning_rate": 1e-06, + "loss": 1.0405, + "mean_token_accuracy": 0.6894251108169556, + "num_tokens": 105048221.0, + "step": 4105 + }, + { + "epoch": 0.45091148693169336, + "grad_norm": 2.283259391784668, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7127031087875366, + "num_tokens": 105070362.0, + "step": 4106 + }, + { + "epoch": 0.45102130463430706, + "grad_norm": 2.038115978240967, + "learning_rate": 1e-06, + "loss": 1.0472, + "mean_token_accuracy": 0.6994394659996033, + "num_tokens": 105098669.0, + "step": 4107 + }, + { + "epoch": 0.4511311223369207, + "grad_norm": 2.3752048015594482, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7215021848678589, + "num_tokens": 105121803.0, + "step": 4108 + }, + { + "epoch": 0.45124094003953436, + "grad_norm": 2.229393482208252, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7331669330596924, + "num_tokens": 105146445.0, + "step": 4109 + }, + { + "epoch": 0.45135075774214806, + "grad_norm": 2.3223047256469727, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7057374119758606, + "num_tokens": 105169099.0, + "step": 4110 + }, + { + "epoch": 0.4514605754447617, + "grad_norm": 2.2336270809173584, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6923656463623047, + "num_tokens": 105195437.0, + "step": 4111 + }, + { + "epoch": 0.45157039314737535, + "grad_norm": 2.0943994522094727, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.7013183832168579, + "num_tokens": 105224163.0, + "step": 4112 + }, + { + "epoch": 0.451680210849989, + "grad_norm": 2.5161986351013184, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.6975580453872681, + "num_tokens": 105246104.0, + "step": 4113 + }, + { + "epoch": 0.4517900285526027, + "grad_norm": 1.9906893968582153, + "learning_rate": 1e-06, + "loss": 1.0347, + "mean_token_accuracy": 0.6868783235549927, + "num_tokens": 105278972.0, + "step": 4114 + }, + { + "epoch": 0.45189984625521634, + "grad_norm": 1.9628506898880005, + "learning_rate": 1e-06, + "loss": 1.0839, + "mean_token_accuracy": 0.6757383346557617, + "num_tokens": 105314939.0, + "step": 4115 + }, + { + "epoch": 0.45200966395783, + "grad_norm": 2.319368600845337, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6992476582527161, + "num_tokens": 105338208.0, + "step": 4116 + }, + { + "epoch": 0.4521194816604437, + "grad_norm": 2.0612993240356445, + "learning_rate": 1e-06, + "loss": 1.0376, + "mean_token_accuracy": 0.6935687065124512, + "num_tokens": 105367944.0, + "step": 4117 + }, + { + "epoch": 0.45222929936305734, + "grad_norm": 2.2374091148376465, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7073566913604736, + "num_tokens": 105391827.0, + "step": 4118 + }, + { + "epoch": 0.452339117065671, + "grad_norm": 2.0512750148773193, + "learning_rate": 1e-06, + "loss": 1.0425, + "mean_token_accuracy": 0.6831969022750854, + "num_tokens": 105422817.0, + "step": 4119 + }, + { + "epoch": 0.4524489347682846, + "grad_norm": 2.090785503387451, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.691540002822876, + "num_tokens": 105451334.0, + "step": 4120 + }, + { + "epoch": 0.45255875247089833, + "grad_norm": 2.066094160079956, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7240575551986694, + "num_tokens": 105477403.0, + "step": 4121 + }, + { + "epoch": 0.452668570173512, + "grad_norm": 2.074374198913574, + "learning_rate": 1e-06, + "loss": 1.1592, + "mean_token_accuracy": 0.6504712104797363, + "num_tokens": 105510464.0, + "step": 4122 + }, + { + "epoch": 0.4527783878761256, + "grad_norm": 2.0686793327331543, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7070127725601196, + "num_tokens": 105538226.0, + "step": 4123 + }, + { + "epoch": 0.45288820557873927, + "grad_norm": 2.502307653427124, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7157144546508789, + "num_tokens": 105557906.0, + "step": 4124 + }, + { + "epoch": 0.45299802328135297, + "grad_norm": 2.0699708461761475, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.7051939964294434, + "num_tokens": 105585828.0, + "step": 4125 + }, + { + "epoch": 0.4531078409839666, + "grad_norm": 2.3150196075439453, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7048420906066895, + "num_tokens": 105610139.0, + "step": 4126 + }, + { + "epoch": 0.45321765868658026, + "grad_norm": 2.428258180618286, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.705454409122467, + "num_tokens": 105632435.0, + "step": 4127 + }, + { + "epoch": 0.45332747638919396, + "grad_norm": 2.2422611713409424, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.704902708530426, + "num_tokens": 105657296.0, + "step": 4128 + }, + { + "epoch": 0.4534372940918076, + "grad_norm": 2.2168478965759277, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.688123345375061, + "num_tokens": 105682760.0, + "step": 4129 + }, + { + "epoch": 0.45354711179442125, + "grad_norm": 2.3691556453704834, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7179855108261108, + "num_tokens": 105705194.0, + "step": 4130 + }, + { + "epoch": 0.4536569294970349, + "grad_norm": 2.257096767425537, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.6966084241867065, + "num_tokens": 105728713.0, + "step": 4131 + }, + { + "epoch": 0.4537667471996486, + "grad_norm": 2.187000036239624, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7040757536888123, + "num_tokens": 105755178.0, + "step": 4132 + }, + { + "epoch": 0.45387656490226225, + "grad_norm": 2.582200288772583, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.699940025806427, + "num_tokens": 105775372.0, + "step": 4133 + }, + { + "epoch": 0.4539863826048759, + "grad_norm": 2.0997421741485596, + "learning_rate": 1e-06, + "loss": 1.0351, + "mean_token_accuracy": 0.6870182752609253, + "num_tokens": 105803408.0, + "step": 4134 + }, + { + "epoch": 0.4540962003074896, + "grad_norm": 2.2434678077697754, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7078009843826294, + "num_tokens": 105829244.0, + "step": 4135 + }, + { + "epoch": 0.45420601801010324, + "grad_norm": 2.236696243286133, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7035631537437439, + "num_tokens": 105853366.0, + "step": 4136 + }, + { + "epoch": 0.4543158357127169, + "grad_norm": 2.332153081893921, + "learning_rate": 1e-06, + "loss": 1.0931, + "mean_token_accuracy": 0.6796749830245972, + "num_tokens": 105878671.0, + "step": 4137 + }, + { + "epoch": 0.45442565341533053, + "grad_norm": 2.436049461364746, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7248734831809998, + "num_tokens": 105897836.0, + "step": 4138 + }, + { + "epoch": 0.45453547111794423, + "grad_norm": 2.540175676345825, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7117438316345215, + "num_tokens": 105919012.0, + "step": 4139 + }, + { + "epoch": 0.4546452888205579, + "grad_norm": 2.2014169692993164, + "learning_rate": 1e-06, + "loss": 1.0222, + "mean_token_accuracy": 0.691947877407074, + "num_tokens": 105945720.0, + "step": 4140 + }, + { + "epoch": 0.4547551065231715, + "grad_norm": 2.0354089736938477, + "learning_rate": 1e-06, + "loss": 1.0167, + "mean_token_accuracy": 0.6911974549293518, + "num_tokens": 105974626.0, + "step": 4141 + }, + { + "epoch": 0.45486492422578517, + "grad_norm": 2.2405855655670166, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6960852146148682, + "num_tokens": 105998053.0, + "step": 4142 + }, + { + "epoch": 0.4549747419283989, + "grad_norm": 2.2762234210968018, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7263344526290894, + "num_tokens": 106020847.0, + "step": 4143 + }, + { + "epoch": 0.4550845596310125, + "grad_norm": 2.474893569946289, + "learning_rate": 1e-06, + "loss": 1.0448, + "mean_token_accuracy": 0.6888607740402222, + "num_tokens": 106043645.0, + "step": 4144 + }, + { + "epoch": 0.45519437733362617, + "grad_norm": 2.1498682498931885, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7094621658325195, + "num_tokens": 106069617.0, + "step": 4145 + }, + { + "epoch": 0.45530419503623987, + "grad_norm": 2.4456725120544434, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7108122706413269, + "num_tokens": 106090018.0, + "step": 4146 + }, + { + "epoch": 0.4554140127388535, + "grad_norm": 2.04423451423645, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6935321688652039, + "num_tokens": 106120975.0, + "step": 4147 + }, + { + "epoch": 0.45552383044146716, + "grad_norm": 2.0402071475982666, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.7020441889762878, + "num_tokens": 106150312.0, + "step": 4148 + }, + { + "epoch": 0.4556336481440808, + "grad_norm": 2.509882926940918, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7134418487548828, + "num_tokens": 106169454.0, + "step": 4149 + }, + { + "epoch": 0.4557434658466945, + "grad_norm": 2.101531505584717, + "learning_rate": 1e-06, + "loss": 1.0234, + "mean_token_accuracy": 0.7012091875076294, + "num_tokens": 106197141.0, + "step": 4150 + }, + { + "epoch": 0.45585328354930815, + "grad_norm": 2.4569664001464844, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7169573903083801, + "num_tokens": 106218407.0, + "step": 4151 + }, + { + "epoch": 0.4559631012519218, + "grad_norm": 2.2668652534484863, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7030999064445496, + "num_tokens": 106243480.0, + "step": 4152 + }, + { + "epoch": 0.45607291895453544, + "grad_norm": 2.118112087249756, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.7168259620666504, + "num_tokens": 106271132.0, + "step": 4153 + }, + { + "epoch": 0.45618273665714915, + "grad_norm": 2.294631004333496, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7084298133850098, + "num_tokens": 106294961.0, + "step": 4154 + }, + { + "epoch": 0.4562925543597628, + "grad_norm": 2.055591583251953, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7123171091079712, + "num_tokens": 106322704.0, + "step": 4155 + }, + { + "epoch": 0.45640237206237644, + "grad_norm": 2.060481548309326, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.718294620513916, + "num_tokens": 106352207.0, + "step": 4156 + }, + { + "epoch": 0.45651218976499014, + "grad_norm": 2.2658116817474365, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7350938320159912, + "num_tokens": 106374947.0, + "step": 4157 + }, + { + "epoch": 0.4566220074676038, + "grad_norm": 2.009779214859009, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.7052122354507446, + "num_tokens": 106404005.0, + "step": 4158 + }, + { + "epoch": 0.45673182517021743, + "grad_norm": 2.4533369541168213, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7124430537223816, + "num_tokens": 106423214.0, + "step": 4159 + }, + { + "epoch": 0.4568416428728311, + "grad_norm": 2.114928960800171, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7077456712722778, + "num_tokens": 106449398.0, + "step": 4160 + }, + { + "epoch": 0.4569514605754448, + "grad_norm": 1.7597267627716064, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7207967638969421, + "num_tokens": 106484108.0, + "step": 4161 + }, + { + "epoch": 0.4570612782780584, + "grad_norm": 2.1095004081726074, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.6987708806991577, + "num_tokens": 106510293.0, + "step": 4162 + }, + { + "epoch": 0.45717109598067207, + "grad_norm": 2.092700481414795, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7134636640548706, + "num_tokens": 106537327.0, + "step": 4163 + }, + { + "epoch": 0.4572809136832858, + "grad_norm": 2.4035449028015137, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7271673679351807, + "num_tokens": 106559472.0, + "step": 4164 + }, + { + "epoch": 0.4573907313858994, + "grad_norm": 2.4694690704345703, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7072595953941345, + "num_tokens": 106580256.0, + "step": 4165 + }, + { + "epoch": 0.45750054908851306, + "grad_norm": 2.0212647914886475, + "learning_rate": 1e-06, + "loss": 1.0826, + "mean_token_accuracy": 0.6796745657920837, + "num_tokens": 106610781.0, + "step": 4166 + }, + { + "epoch": 0.4576103667911267, + "grad_norm": 2.155447006225586, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.7067053318023682, + "num_tokens": 106636373.0, + "step": 4167 + }, + { + "epoch": 0.4577201844937404, + "grad_norm": 2.0530812740325928, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.7015323638916016, + "num_tokens": 106665644.0, + "step": 4168 + }, + { + "epoch": 0.45783000219635406, + "grad_norm": 2.226398229598999, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.702376127243042, + "num_tokens": 106690679.0, + "step": 4169 + }, + { + "epoch": 0.4579398198989677, + "grad_norm": 2.457679510116577, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.737843930721283, + "num_tokens": 106710610.0, + "step": 4170 + }, + { + "epoch": 0.45804963760158135, + "grad_norm": 2.37896990776062, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7195813059806824, + "num_tokens": 106731119.0, + "step": 4171 + }, + { + "epoch": 0.45815945530419505, + "grad_norm": 2.3740108013153076, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.719738245010376, + "num_tokens": 106752138.0, + "step": 4172 + }, + { + "epoch": 0.4582692730068087, + "grad_norm": 2.1477315425872803, + "learning_rate": 1e-06, + "loss": 1.0335, + "mean_token_accuracy": 0.6903647184371948, + "num_tokens": 106779728.0, + "step": 4173 + }, + { + "epoch": 0.45837909070942234, + "grad_norm": 2.029390335083008, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7057225704193115, + "num_tokens": 106808669.0, + "step": 4174 + }, + { + "epoch": 0.45848890841203604, + "grad_norm": 2.144488573074341, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.7012670636177063, + "num_tokens": 106835710.0, + "step": 4175 + }, + { + "epoch": 0.4585987261146497, + "grad_norm": 2.25903582572937, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.6979238390922546, + "num_tokens": 106862402.0, + "step": 4176 + }, + { + "epoch": 0.45870854381726334, + "grad_norm": 2.3578805923461914, + "learning_rate": 1e-06, + "loss": 1.0479, + "mean_token_accuracy": 0.6935194730758667, + "num_tokens": 106885111.0, + "step": 4177 + }, + { + "epoch": 0.458818361519877, + "grad_norm": 2.225188970565796, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7099449634552002, + "num_tokens": 106908761.0, + "step": 4178 + }, + { + "epoch": 0.4589281792224907, + "grad_norm": 2.185828685760498, + "learning_rate": 1e-06, + "loss": 1.0195, + "mean_token_accuracy": 0.6943936944007874, + "num_tokens": 106935019.0, + "step": 4179 + }, + { + "epoch": 0.45903799692510433, + "grad_norm": 2.2932770252227783, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7262101173400879, + "num_tokens": 106957130.0, + "step": 4180 + }, + { + "epoch": 0.459147814627718, + "grad_norm": 2.382974147796631, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7094475030899048, + "num_tokens": 106979381.0, + "step": 4181 + }, + { + "epoch": 0.4592576323303316, + "grad_norm": 2.0789949893951416, + "learning_rate": 1e-06, + "loss": 1.0369, + "mean_token_accuracy": 0.6944200396537781, + "num_tokens": 107007657.0, + "step": 4182 + }, + { + "epoch": 0.4593674500329453, + "grad_norm": 2.3374810218811035, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6996400952339172, + "num_tokens": 107031322.0, + "step": 4183 + }, + { + "epoch": 0.45947726773555897, + "grad_norm": 2.3282597064971924, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7031205296516418, + "num_tokens": 107056210.0, + "step": 4184 + }, + { + "epoch": 0.4595870854381726, + "grad_norm": 2.1195473670959473, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7055848836898804, + "num_tokens": 107085521.0, + "step": 4185 + }, + { + "epoch": 0.4596969031407863, + "grad_norm": 2.3742289543151855, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7018375992774963, + "num_tokens": 107109102.0, + "step": 4186 + }, + { + "epoch": 0.45980672084339996, + "grad_norm": 2.196894645690918, + "learning_rate": 1e-06, + "loss": 1.0376, + "mean_token_accuracy": 0.6877830028533936, + "num_tokens": 107135106.0, + "step": 4187 + }, + { + "epoch": 0.4599165385460136, + "grad_norm": 2.142587661743164, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7061911821365356, + "num_tokens": 107165608.0, + "step": 4188 + }, + { + "epoch": 0.46002635624862726, + "grad_norm": 2.459735631942749, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7073757648468018, + "num_tokens": 107187741.0, + "step": 4189 + }, + { + "epoch": 0.46013617395124096, + "grad_norm": 2.332200527191162, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7044637203216553, + "num_tokens": 107212028.0, + "step": 4190 + }, + { + "epoch": 0.4602459916538546, + "grad_norm": 2.0894997119903564, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6981541514396667, + "num_tokens": 107240499.0, + "step": 4191 + }, + { + "epoch": 0.46035580935646825, + "grad_norm": 2.1978888511657715, + "learning_rate": 1e-06, + "loss": 1.0196, + "mean_token_accuracy": 0.6921594738960266, + "num_tokens": 107268267.0, + "step": 4192 + }, + { + "epoch": 0.46046562705908195, + "grad_norm": 2.338386297225952, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7157226800918579, + "num_tokens": 107291942.0, + "step": 4193 + }, + { + "epoch": 0.4605754447616956, + "grad_norm": 2.271117687225342, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7342309355735779, + "num_tokens": 107314901.0, + "step": 4194 + }, + { + "epoch": 0.46068526246430924, + "grad_norm": 1.9516606330871582, + "learning_rate": 1e-06, + "loss": 1.0471, + "mean_token_accuracy": 0.6776635050773621, + "num_tokens": 107346199.0, + "step": 4195 + }, + { + "epoch": 0.4607950801669229, + "grad_norm": 2.2220945358276367, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6942030191421509, + "num_tokens": 107371601.0, + "step": 4196 + }, + { + "epoch": 0.4609048978695366, + "grad_norm": 2.4468681812286377, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7134039402008057, + "num_tokens": 107393708.0, + "step": 4197 + }, + { + "epoch": 0.46101471557215024, + "grad_norm": 2.240405559539795, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7067989110946655, + "num_tokens": 107418672.0, + "step": 4198 + }, + { + "epoch": 0.4611245332747639, + "grad_norm": 2.3310084342956543, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.7026062607765198, + "num_tokens": 107442526.0, + "step": 4199 + }, + { + "epoch": 0.4612343509773775, + "grad_norm": 2.293503761291504, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7091259956359863, + "num_tokens": 107466203.0, + "step": 4200 + }, + { + "epoch": 0.46134416867999123, + "grad_norm": 1.9643945693969727, + "learning_rate": 1e-06, + "loss": 1.041, + "mean_token_accuracy": 0.6844279766082764, + "num_tokens": 107499939.0, + "step": 4201 + }, + { + "epoch": 0.4614539863826049, + "grad_norm": 1.8175599575042725, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6878639459609985, + "num_tokens": 107535966.0, + "step": 4202 + }, + { + "epoch": 0.4615638040852185, + "grad_norm": 2.387413263320923, + "learning_rate": 1e-06, + "loss": 1.1307, + "mean_token_accuracy": 0.6590060591697693, + "num_tokens": 107558569.0, + "step": 4203 + }, + { + "epoch": 0.4616736217878322, + "grad_norm": 1.8546714782714844, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7153215408325195, + "num_tokens": 107590827.0, + "step": 4204 + }, + { + "epoch": 0.46178343949044587, + "grad_norm": 2.0790855884552, + "learning_rate": 1e-06, + "loss": 1.0489, + "mean_token_accuracy": 0.6842883825302124, + "num_tokens": 107621990.0, + "step": 4205 + }, + { + "epoch": 0.4618932571930595, + "grad_norm": 2.4185526371002197, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.7012288570404053, + "num_tokens": 107645371.0, + "step": 4206 + }, + { + "epoch": 0.46200307489567316, + "grad_norm": 2.241609811782837, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6934289932250977, + "num_tokens": 107670026.0, + "step": 4207 + }, + { + "epoch": 0.46211289259828686, + "grad_norm": 2.067526340484619, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7119095325469971, + "num_tokens": 107697909.0, + "step": 4208 + }, + { + "epoch": 0.4622227103009005, + "grad_norm": 2.0505807399749756, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.701433539390564, + "num_tokens": 107726575.0, + "step": 4209 + }, + { + "epoch": 0.46233252800351415, + "grad_norm": 2.1471195220947266, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7007067203521729, + "num_tokens": 107751739.0, + "step": 4210 + }, + { + "epoch": 0.46244234570612786, + "grad_norm": 2.376220464706421, + "learning_rate": 1e-06, + "loss": 1.0994, + "mean_token_accuracy": 0.6755008697509766, + "num_tokens": 107774861.0, + "step": 4211 + }, + { + "epoch": 0.4625521634087415, + "grad_norm": 2.1210734844207764, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6954836845397949, + "num_tokens": 107801220.0, + "step": 4212 + }, + { + "epoch": 0.46266198111135515, + "grad_norm": 2.2606258392333984, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7108745574951172, + "num_tokens": 107825923.0, + "step": 4213 + }, + { + "epoch": 0.4627717988139688, + "grad_norm": 2.155374765396118, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7260830402374268, + "num_tokens": 107850416.0, + "step": 4214 + }, + { + "epoch": 0.4628816165165825, + "grad_norm": 2.2927398681640625, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.717294454574585, + "num_tokens": 107875606.0, + "step": 4215 + }, + { + "epoch": 0.46299143421919614, + "grad_norm": 1.9761884212493896, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.724985659122467, + "num_tokens": 107903652.0, + "step": 4216 + }, + { + "epoch": 0.4631012519218098, + "grad_norm": 2.2068679332733154, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7139937281608582, + "num_tokens": 107928012.0, + "step": 4217 + }, + { + "epoch": 0.46321106962442343, + "grad_norm": 2.4067957401275635, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.721366286277771, + "num_tokens": 107951618.0, + "step": 4218 + }, + { + "epoch": 0.46332088732703713, + "grad_norm": 2.07498836517334, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7024606466293335, + "num_tokens": 107981053.0, + "step": 4219 + }, + { + "epoch": 0.4634307050296508, + "grad_norm": 1.9589953422546387, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.7017526626586914, + "num_tokens": 108013968.0, + "step": 4220 + }, + { + "epoch": 0.4635405227322644, + "grad_norm": 2.3384318351745605, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7068694829940796, + "num_tokens": 108037194.0, + "step": 4221 + }, + { + "epoch": 0.4636503404348781, + "grad_norm": 2.290745973587036, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7169687151908875, + "num_tokens": 108062204.0, + "step": 4222 + }, + { + "epoch": 0.4637601581374918, + "grad_norm": 2.049345016479492, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.695522129535675, + "num_tokens": 108092099.0, + "step": 4223 + }, + { + "epoch": 0.4638699758401054, + "grad_norm": 2.0758209228515625, + "learning_rate": 1e-06, + "loss": 1.1246, + "mean_token_accuracy": 0.6694404482841492, + "num_tokens": 108121947.0, + "step": 4224 + }, + { + "epoch": 0.46397979354271907, + "grad_norm": 2.2247118949890137, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.7005917429924011, + "num_tokens": 108147806.0, + "step": 4225 + }, + { + "epoch": 0.46408961124533277, + "grad_norm": 2.1356616020202637, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7021445035934448, + "num_tokens": 108177479.0, + "step": 4226 + }, + { + "epoch": 0.4641994289479464, + "grad_norm": 2.1449902057647705, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7316963076591492, + "num_tokens": 108202708.0, + "step": 4227 + }, + { + "epoch": 0.46430924665056006, + "grad_norm": 2.4092187881469727, + "learning_rate": 1e-06, + "loss": 1.0511, + "mean_token_accuracy": 0.6876975893974304, + "num_tokens": 108224244.0, + "step": 4228 + }, + { + "epoch": 0.4644190643531737, + "grad_norm": 2.241331100463867, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6976096630096436, + "num_tokens": 108249070.0, + "step": 4229 + }, + { + "epoch": 0.4645288820557874, + "grad_norm": 2.0005037784576416, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6980884075164795, + "num_tokens": 108278403.0, + "step": 4230 + }, + { + "epoch": 0.46463869975840105, + "grad_norm": 2.3966917991638184, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7273659706115723, + "num_tokens": 108298991.0, + "step": 4231 + }, + { + "epoch": 0.4647485174610147, + "grad_norm": 2.3643569946289062, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7081904411315918, + "num_tokens": 108322612.0, + "step": 4232 + }, + { + "epoch": 0.4648583351636284, + "grad_norm": 2.079789876937866, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7227933406829834, + "num_tokens": 108349876.0, + "step": 4233 + }, + { + "epoch": 0.46496815286624205, + "grad_norm": 2.0340096950531006, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.7016434669494629, + "num_tokens": 108377539.0, + "step": 4234 + }, + { + "epoch": 0.4650779705688557, + "grad_norm": 2.064131259918213, + "learning_rate": 1e-06, + "loss": 1.1264, + "mean_token_accuracy": 0.6730815172195435, + "num_tokens": 108407968.0, + "step": 4235 + }, + { + "epoch": 0.46518778827146934, + "grad_norm": 1.9731369018554688, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7232786417007446, + "num_tokens": 108436771.0, + "step": 4236 + }, + { + "epoch": 0.46529760597408304, + "grad_norm": 2.308306932449341, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.6979857683181763, + "num_tokens": 108460728.0, + "step": 4237 + }, + { + "epoch": 0.4654074236766967, + "grad_norm": 2.0938684940338135, + "learning_rate": 1e-06, + "loss": 1.064, + "mean_token_accuracy": 0.6784940361976624, + "num_tokens": 108488344.0, + "step": 4238 + }, + { + "epoch": 0.46551724137931033, + "grad_norm": 2.1318118572235107, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.7080646753311157, + "num_tokens": 108516022.0, + "step": 4239 + }, + { + "epoch": 0.46562705908192403, + "grad_norm": 2.307664394378662, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.7029047012329102, + "num_tokens": 108541277.0, + "step": 4240 + }, + { + "epoch": 0.4657368767845377, + "grad_norm": 2.1852123737335205, + "learning_rate": 1e-06, + "loss": 1.025, + "mean_token_accuracy": 0.6983118057250977, + "num_tokens": 108566168.0, + "step": 4241 + }, + { + "epoch": 0.4658466944871513, + "grad_norm": 2.2980828285217285, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.717572033405304, + "num_tokens": 108587562.0, + "step": 4242 + }, + { + "epoch": 0.46595651218976497, + "grad_norm": 2.337258815765381, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.7013269662857056, + "num_tokens": 108610945.0, + "step": 4243 + }, + { + "epoch": 0.4660663298923787, + "grad_norm": 2.1908748149871826, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.697306752204895, + "num_tokens": 108637617.0, + "step": 4244 + }, + { + "epoch": 0.4661761475949923, + "grad_norm": 1.9765249490737915, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.707133948802948, + "num_tokens": 108666987.0, + "step": 4245 + }, + { + "epoch": 0.46628596529760596, + "grad_norm": 2.610048294067383, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7051624059677124, + "num_tokens": 108687007.0, + "step": 4246 + }, + { + "epoch": 0.4663957830002196, + "grad_norm": 2.003109931945801, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7032173871994019, + "num_tokens": 108717083.0, + "step": 4247 + }, + { + "epoch": 0.4665056007028333, + "grad_norm": 2.596588134765625, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7039085626602173, + "num_tokens": 108738817.0, + "step": 4248 + }, + { + "epoch": 0.46661541840544696, + "grad_norm": 2.17289400100708, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6905331611633301, + "num_tokens": 108767302.0, + "step": 4249 + }, + { + "epoch": 0.4667252361080606, + "grad_norm": 2.3691577911376953, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7031769752502441, + "num_tokens": 108789460.0, + "step": 4250 + }, + { + "epoch": 0.4668350538106743, + "grad_norm": 2.459221839904785, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6928257942199707, + "num_tokens": 108811960.0, + "step": 4251 + }, + { + "epoch": 0.46694487151328795, + "grad_norm": 2.4188485145568848, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7222230434417725, + "num_tokens": 108831821.0, + "step": 4252 + }, + { + "epoch": 0.4670546892159016, + "grad_norm": 2.0904276371002197, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7097024321556091, + "num_tokens": 108857966.0, + "step": 4253 + }, + { + "epoch": 0.46716450691851524, + "grad_norm": 2.2548396587371826, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7143201231956482, + "num_tokens": 108880385.0, + "step": 4254 + }, + { + "epoch": 0.46727432462112894, + "grad_norm": 2.3596603870391846, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7115959525108337, + "num_tokens": 108901962.0, + "step": 4255 + }, + { + "epoch": 0.4673841423237426, + "grad_norm": 2.2732715606689453, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7248905897140503, + "num_tokens": 108926532.0, + "step": 4256 + }, + { + "epoch": 0.46749396002635624, + "grad_norm": 2.0840632915496826, + "learning_rate": 1e-06, + "loss": 1.0295, + "mean_token_accuracy": 0.686255931854248, + "num_tokens": 108955061.0, + "step": 4257 + }, + { + "epoch": 0.4676037777289699, + "grad_norm": 2.3679964542388916, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7129335403442383, + "num_tokens": 108977351.0, + "step": 4258 + }, + { + "epoch": 0.4677135954315836, + "grad_norm": 2.2758500576019287, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6992847919464111, + "num_tokens": 109002169.0, + "step": 4259 + }, + { + "epoch": 0.46782341313419723, + "grad_norm": 2.2791569232940674, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7082428336143494, + "num_tokens": 109025249.0, + "step": 4260 + }, + { + "epoch": 0.4679332308368109, + "grad_norm": 2.1067614555358887, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.6929517984390259, + "num_tokens": 109052550.0, + "step": 4261 + }, + { + "epoch": 0.4680430485394246, + "grad_norm": 2.4342050552368164, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.703659176826477, + "num_tokens": 109073140.0, + "step": 4262 + }, + { + "epoch": 0.4681528662420382, + "grad_norm": 2.5327095985412598, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7037512063980103, + "num_tokens": 109092929.0, + "step": 4263 + }, + { + "epoch": 0.46826268394465187, + "grad_norm": 2.2771029472351074, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7038842439651489, + "num_tokens": 109117053.0, + "step": 4264 + }, + { + "epoch": 0.4683725016472655, + "grad_norm": 2.2715351581573486, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7032994031906128, + "num_tokens": 109142880.0, + "step": 4265 + }, + { + "epoch": 0.4684823193498792, + "grad_norm": 2.2145798206329346, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.714962363243103, + "num_tokens": 109170399.0, + "step": 4266 + }, + { + "epoch": 0.46859213705249286, + "grad_norm": 2.5670461654663086, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7271907925605774, + "num_tokens": 109190064.0, + "step": 4267 + }, + { + "epoch": 0.4687019547551065, + "grad_norm": 2.4944326877593994, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7147495746612549, + "num_tokens": 109210213.0, + "step": 4268 + }, + { + "epoch": 0.4688117724577202, + "grad_norm": 2.2941224575042725, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7067373991012573, + "num_tokens": 109234057.0, + "step": 4269 + }, + { + "epoch": 0.46892159016033386, + "grad_norm": 2.2925403118133545, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7104620933532715, + "num_tokens": 109257113.0, + "step": 4270 + }, + { + "epoch": 0.4690314078629475, + "grad_norm": 2.233753204345703, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.7057167291641235, + "num_tokens": 109281813.0, + "step": 4271 + }, + { + "epoch": 0.46914122556556115, + "grad_norm": 2.211235284805298, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6907375454902649, + "num_tokens": 109306919.0, + "step": 4272 + }, + { + "epoch": 0.46925104326817485, + "grad_norm": 2.1085267066955566, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7137829661369324, + "num_tokens": 109334323.0, + "step": 4273 + }, + { + "epoch": 0.4693608609707885, + "grad_norm": 2.2854182720184326, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7051592469215393, + "num_tokens": 109358064.0, + "step": 4274 + }, + { + "epoch": 0.46947067867340214, + "grad_norm": 2.0063743591308594, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7100420594215393, + "num_tokens": 109384896.0, + "step": 4275 + }, + { + "epoch": 0.4695804963760158, + "grad_norm": 2.2388081550598145, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.6979767680168152, + "num_tokens": 109410082.0, + "step": 4276 + }, + { + "epoch": 0.4696903140786295, + "grad_norm": 2.500561237335205, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7007437944412231, + "num_tokens": 109430818.0, + "step": 4277 + }, + { + "epoch": 0.46980013178124314, + "grad_norm": 2.0457067489624023, + "learning_rate": 1e-06, + "loss": 1.0759, + "mean_token_accuracy": 0.676936149597168, + "num_tokens": 109460272.0, + "step": 4278 + }, + { + "epoch": 0.4699099494838568, + "grad_norm": 2.2544589042663574, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7073667645454407, + "num_tokens": 109484765.0, + "step": 4279 + }, + { + "epoch": 0.4700197671864705, + "grad_norm": 1.985700249671936, + "learning_rate": 1e-06, + "loss": 1.0528, + "mean_token_accuracy": 0.6838052272796631, + "num_tokens": 109513594.0, + "step": 4280 + }, + { + "epoch": 0.47012958488908413, + "grad_norm": 2.428316354751587, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7001481652259827, + "num_tokens": 109535246.0, + "step": 4281 + }, + { + "epoch": 0.4702394025916978, + "grad_norm": 2.0033986568450928, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.6922858953475952, + "num_tokens": 109564584.0, + "step": 4282 + }, + { + "epoch": 0.4703492202943114, + "grad_norm": 2.4248690605163574, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7251863479614258, + "num_tokens": 109587923.0, + "step": 4283 + }, + { + "epoch": 0.4704590379969251, + "grad_norm": 2.202207326889038, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.709357738494873, + "num_tokens": 109613772.0, + "step": 4284 + }, + { + "epoch": 0.47056885569953877, + "grad_norm": 2.289762496948242, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6882781982421875, + "num_tokens": 109637542.0, + "step": 4285 + }, + { + "epoch": 0.4706786734021524, + "grad_norm": 2.261237382888794, + "learning_rate": 1e-06, + "loss": 1.0282, + "mean_token_accuracy": 0.6908250451087952, + "num_tokens": 109664074.0, + "step": 4286 + }, + { + "epoch": 0.4707884911047661, + "grad_norm": 2.3246939182281494, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.6908538937568665, + "num_tokens": 109689165.0, + "step": 4287 + }, + { + "epoch": 0.47089830880737976, + "grad_norm": 2.1793723106384277, + "learning_rate": 1e-06, + "loss": 1.0606, + "mean_token_accuracy": 0.682280421257019, + "num_tokens": 109717921.0, + "step": 4288 + }, + { + "epoch": 0.4710081265099934, + "grad_norm": 1.8537676334381104, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.698997974395752, + "num_tokens": 109751557.0, + "step": 4289 + }, + { + "epoch": 0.47111794421260705, + "grad_norm": 2.3310022354125977, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.6969625949859619, + "num_tokens": 109777986.0, + "step": 4290 + }, + { + "epoch": 0.47122776191522076, + "grad_norm": 2.3826041221618652, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7124050855636597, + "num_tokens": 109798943.0, + "step": 4291 + }, + { + "epoch": 0.4713375796178344, + "grad_norm": 1.9585602283477783, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.695205807685852, + "num_tokens": 109831140.0, + "step": 4292 + }, + { + "epoch": 0.47144739732044805, + "grad_norm": 2.022048234939575, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7012592554092407, + "num_tokens": 109863667.0, + "step": 4293 + }, + { + "epoch": 0.4715572150230617, + "grad_norm": 2.1241769790649414, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6987857222557068, + "num_tokens": 109889648.0, + "step": 4294 + }, + { + "epoch": 0.4716670327256754, + "grad_norm": 2.1134095191955566, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7245267629623413, + "num_tokens": 109913924.0, + "step": 4295 + }, + { + "epoch": 0.47177685042828904, + "grad_norm": 2.125619411468506, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.7002537250518799, + "num_tokens": 109943936.0, + "step": 4296 + }, + { + "epoch": 0.4718866681309027, + "grad_norm": 2.152682304382324, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7296788692474365, + "num_tokens": 109970736.0, + "step": 4297 + }, + { + "epoch": 0.4719964858335164, + "grad_norm": 2.1944737434387207, + "learning_rate": 1e-06, + "loss": 1.0295, + "mean_token_accuracy": 0.6925919055938721, + "num_tokens": 109996218.0, + "step": 4298 + }, + { + "epoch": 0.47210630353613003, + "grad_norm": 2.2879767417907715, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6897621154785156, + "num_tokens": 110022901.0, + "step": 4299 + }, + { + "epoch": 0.4722161212387437, + "grad_norm": 2.0426831245422363, + "learning_rate": 1e-06, + "loss": 1.0383, + "mean_token_accuracy": 0.6901953816413879, + "num_tokens": 110052377.0, + "step": 4300 + }, + { + "epoch": 0.4723259389413573, + "grad_norm": 2.0535502433776855, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7107534408569336, + "num_tokens": 110080315.0, + "step": 4301 + }, + { + "epoch": 0.472435756643971, + "grad_norm": 2.354304552078247, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7153483629226685, + "num_tokens": 110102956.0, + "step": 4302 + }, + { + "epoch": 0.4725455743465847, + "grad_norm": 2.24051570892334, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7172024250030518, + "num_tokens": 110127357.0, + "step": 4303 + }, + { + "epoch": 0.4726553920491983, + "grad_norm": 2.0886735916137695, + "learning_rate": 1e-06, + "loss": 1.1093, + "mean_token_accuracy": 0.6727563142776489, + "num_tokens": 110159123.0, + "step": 4304 + }, + { + "epoch": 0.47276520975181197, + "grad_norm": 2.240865707397461, + "learning_rate": 1e-06, + "loss": 1.0519, + "mean_token_accuracy": 0.6812422871589661, + "num_tokens": 110187355.0, + "step": 4305 + }, + { + "epoch": 0.47287502745442567, + "grad_norm": 2.241020441055298, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7124927043914795, + "num_tokens": 110214544.0, + "step": 4306 + }, + { + "epoch": 0.4729848451570393, + "grad_norm": 2.2703640460968018, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.6959836483001709, + "num_tokens": 110240941.0, + "step": 4307 + }, + { + "epoch": 0.47309466285965296, + "grad_norm": 2.4946558475494385, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7233014106750488, + "num_tokens": 110261987.0, + "step": 4308 + }, + { + "epoch": 0.47320448056226666, + "grad_norm": 2.646803617477417, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7052826881408691, + "num_tokens": 110281300.0, + "step": 4309 + }, + { + "epoch": 0.4733142982648803, + "grad_norm": 2.0023627281188965, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7094178199768066, + "num_tokens": 110310452.0, + "step": 4310 + }, + { + "epoch": 0.47342411596749395, + "grad_norm": 2.130255937576294, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.697638750076294, + "num_tokens": 110336549.0, + "step": 4311 + }, + { + "epoch": 0.4735339336701076, + "grad_norm": 2.11040997505188, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7064282894134521, + "num_tokens": 110362457.0, + "step": 4312 + }, + { + "epoch": 0.4736437513727213, + "grad_norm": 2.2002859115600586, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7228351831436157, + "num_tokens": 110388213.0, + "step": 4313 + }, + { + "epoch": 0.47375356907533495, + "grad_norm": 1.8933892250061035, + "learning_rate": 1e-06, + "loss": 1.0824, + "mean_token_accuracy": 0.6836364269256592, + "num_tokens": 110420265.0, + "step": 4314 + }, + { + "epoch": 0.4738633867779486, + "grad_norm": 2.3338067531585693, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7123228311538696, + "num_tokens": 110444383.0, + "step": 4315 + }, + { + "epoch": 0.4739732044805623, + "grad_norm": 2.4244918823242188, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7155413627624512, + "num_tokens": 110466353.0, + "step": 4316 + }, + { + "epoch": 0.47408302218317594, + "grad_norm": 2.092576026916504, + "learning_rate": 1e-06, + "loss": 1.0514, + "mean_token_accuracy": 0.6869093179702759, + "num_tokens": 110495378.0, + "step": 4317 + }, + { + "epoch": 0.4741928398857896, + "grad_norm": 2.0435707569122314, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.6978676319122314, + "num_tokens": 110522180.0, + "step": 4318 + }, + { + "epoch": 0.47430265758840323, + "grad_norm": 2.028136968612671, + "learning_rate": 1e-06, + "loss": 1.058, + "mean_token_accuracy": 0.6830133199691772, + "num_tokens": 110553107.0, + "step": 4319 + }, + { + "epoch": 0.47441247529101693, + "grad_norm": 2.0285606384277344, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7129311561584473, + "num_tokens": 110580799.0, + "step": 4320 + }, + { + "epoch": 0.4745222929936306, + "grad_norm": 2.0610570907592773, + "learning_rate": 1e-06, + "loss": 1.0876, + "mean_token_accuracy": 0.6770316362380981, + "num_tokens": 110611789.0, + "step": 4321 + }, + { + "epoch": 0.4746321106962442, + "grad_norm": 2.118778944015503, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7190953493118286, + "num_tokens": 110637637.0, + "step": 4322 + }, + { + "epoch": 0.47474192839885787, + "grad_norm": 2.3317506313323975, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.6992834806442261, + "num_tokens": 110662399.0, + "step": 4323 + }, + { + "epoch": 0.4748517461014716, + "grad_norm": 2.2293694019317627, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6919174194335938, + "num_tokens": 110687444.0, + "step": 4324 + }, + { + "epoch": 0.4749615638040852, + "grad_norm": 2.2483954429626465, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.7033113241195679, + "num_tokens": 110712680.0, + "step": 4325 + }, + { + "epoch": 0.47507138150669886, + "grad_norm": 2.5682077407836914, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7153806090354919, + "num_tokens": 110731857.0, + "step": 4326 + }, + { + "epoch": 0.47518119920931257, + "grad_norm": 2.053391695022583, + "learning_rate": 1e-06, + "loss": 1.0316, + "mean_token_accuracy": 0.6830240488052368, + "num_tokens": 110760727.0, + "step": 4327 + }, + { + "epoch": 0.4752910169119262, + "grad_norm": 1.980432152748108, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.6954958438873291, + "num_tokens": 110794101.0, + "step": 4328 + }, + { + "epoch": 0.47540083461453986, + "grad_norm": 2.427528142929077, + "learning_rate": 1e-06, + "loss": 1.0223, + "mean_token_accuracy": 0.6929697394371033, + "num_tokens": 110817437.0, + "step": 4329 + }, + { + "epoch": 0.4755106523171535, + "grad_norm": 2.2421374320983887, + "learning_rate": 1e-06, + "loss": 1.0278, + "mean_token_accuracy": 0.6868349313735962, + "num_tokens": 110842381.0, + "step": 4330 + }, + { + "epoch": 0.4756204700197672, + "grad_norm": 2.180345296859741, + "learning_rate": 1e-06, + "loss": 1.0816, + "mean_token_accuracy": 0.6789601445198059, + "num_tokens": 110868517.0, + "step": 4331 + }, + { + "epoch": 0.47573028772238085, + "grad_norm": 2.0728578567504883, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7157312631607056, + "num_tokens": 110895466.0, + "step": 4332 + }, + { + "epoch": 0.4758401054249945, + "grad_norm": 2.426060914993286, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.706506609916687, + "num_tokens": 110918640.0, + "step": 4333 + }, + { + "epoch": 0.47594992312760814, + "grad_norm": 2.15860652923584, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.71871018409729, + "num_tokens": 110943216.0, + "step": 4334 + }, + { + "epoch": 0.47605974083022184, + "grad_norm": 2.320458173751831, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7104172110557556, + "num_tokens": 110964740.0, + "step": 4335 + }, + { + "epoch": 0.4761695585328355, + "grad_norm": 2.310377597808838, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6864089965820312, + "num_tokens": 110990206.0, + "step": 4336 + }, + { + "epoch": 0.47627937623544914, + "grad_norm": 2.2302744388580322, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7190155982971191, + "num_tokens": 111013919.0, + "step": 4337 + }, + { + "epoch": 0.47638919393806284, + "grad_norm": 2.065425157546997, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6923565864562988, + "num_tokens": 111041624.0, + "step": 4338 + }, + { + "epoch": 0.4764990116406765, + "grad_norm": 2.6715750694274902, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7164522409439087, + "num_tokens": 111061679.0, + "step": 4339 + }, + { + "epoch": 0.47660882934329013, + "grad_norm": 2.506479501724243, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7060925960540771, + "num_tokens": 111083242.0, + "step": 4340 + }, + { + "epoch": 0.4767186470459038, + "grad_norm": 2.6880295276641846, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7111334800720215, + "num_tokens": 111101082.0, + "step": 4341 + }, + { + "epoch": 0.4768284647485175, + "grad_norm": 2.0503599643707275, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7163025140762329, + "num_tokens": 111128759.0, + "step": 4342 + }, + { + "epoch": 0.4769382824511311, + "grad_norm": 2.494828701019287, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7243648767471313, + "num_tokens": 111149982.0, + "step": 4343 + }, + { + "epoch": 0.47704810015374477, + "grad_norm": 2.4261302947998047, + "learning_rate": 1e-06, + "loss": 1.0435, + "mean_token_accuracy": 0.6901232004165649, + "num_tokens": 111171817.0, + "step": 4344 + }, + { + "epoch": 0.47715791785635847, + "grad_norm": 1.9847426414489746, + "learning_rate": 1e-06, + "loss": 1.0484, + "mean_token_accuracy": 0.6854397058486938, + "num_tokens": 111203920.0, + "step": 4345 + }, + { + "epoch": 0.4772677355589721, + "grad_norm": 2.1101267337799072, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7212895154953003, + "num_tokens": 111229418.0, + "step": 4346 + }, + { + "epoch": 0.47737755326158576, + "grad_norm": 2.2740893363952637, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7074750065803528, + "num_tokens": 111254413.0, + "step": 4347 + }, + { + "epoch": 0.4774873709641994, + "grad_norm": 2.0785117149353027, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.7057467699050903, + "num_tokens": 111281184.0, + "step": 4348 + }, + { + "epoch": 0.4775971886668131, + "grad_norm": 2.418884515762329, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7171947956085205, + "num_tokens": 111300791.0, + "step": 4349 + }, + { + "epoch": 0.47770700636942676, + "grad_norm": 1.882348895072937, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.695643424987793, + "num_tokens": 111333729.0, + "step": 4350 + }, + { + "epoch": 0.4778168240720404, + "grad_norm": 2.365121603012085, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7040603160858154, + "num_tokens": 111358609.0, + "step": 4351 + }, + { + "epoch": 0.47792664177465405, + "grad_norm": 2.140028476715088, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6900351643562317, + "num_tokens": 111384578.0, + "step": 4352 + }, + { + "epoch": 0.47803645947726775, + "grad_norm": 2.016340970993042, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.6962827444076538, + "num_tokens": 111417833.0, + "step": 4353 + }, + { + "epoch": 0.4781462771798814, + "grad_norm": 2.4905550479888916, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7048115730285645, + "num_tokens": 111439152.0, + "step": 4354 + }, + { + "epoch": 0.47825609488249504, + "grad_norm": 2.4717602729797363, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7159886360168457, + "num_tokens": 111458933.0, + "step": 4355 + }, + { + "epoch": 0.47836591258510874, + "grad_norm": 2.1875052452087402, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7218919992446899, + "num_tokens": 111482734.0, + "step": 4356 + }, + { + "epoch": 0.4784757302877224, + "grad_norm": 2.095733880996704, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.7015600800514221, + "num_tokens": 111509715.0, + "step": 4357 + }, + { + "epoch": 0.47858554799033604, + "grad_norm": 2.0604007244110107, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6968433856964111, + "num_tokens": 111540516.0, + "step": 4358 + }, + { + "epoch": 0.4786953656929497, + "grad_norm": 2.312509298324585, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.7032040357589722, + "num_tokens": 111564876.0, + "step": 4359 + }, + { + "epoch": 0.4788051833955634, + "grad_norm": 2.1336829662323, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.699674129486084, + "num_tokens": 111593259.0, + "step": 4360 + }, + { + "epoch": 0.47891500109817703, + "grad_norm": 2.610750675201416, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7173196077346802, + "num_tokens": 111611734.0, + "step": 4361 + }, + { + "epoch": 0.4790248188007907, + "grad_norm": 2.448272228240967, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7124295234680176, + "num_tokens": 111631843.0, + "step": 4362 + }, + { + "epoch": 0.4791346365034044, + "grad_norm": 2.577024459838867, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7005695104598999, + "num_tokens": 111652638.0, + "step": 4363 + }, + { + "epoch": 0.479244454206018, + "grad_norm": 2.288379669189453, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7238821983337402, + "num_tokens": 111676623.0, + "step": 4364 + }, + { + "epoch": 0.47935427190863167, + "grad_norm": 2.1146347522735596, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7068370580673218, + "num_tokens": 111704298.0, + "step": 4365 + }, + { + "epoch": 0.4794640896112453, + "grad_norm": 2.137917995452881, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.690136730670929, + "num_tokens": 111732255.0, + "step": 4366 + }, + { + "epoch": 0.479573907313859, + "grad_norm": 2.051575183868408, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.685014009475708, + "num_tokens": 111761322.0, + "step": 4367 + }, + { + "epoch": 0.47968372501647266, + "grad_norm": 2.2541885375976562, + "learning_rate": 1e-06, + "loss": 1.0264, + "mean_token_accuracy": 0.6834064722061157, + "num_tokens": 111785558.0, + "step": 4368 + }, + { + "epoch": 0.4797935427190863, + "grad_norm": 2.072539806365967, + "learning_rate": 1e-06, + "loss": 1.0481, + "mean_token_accuracy": 0.683116614818573, + "num_tokens": 111815112.0, + "step": 4369 + }, + { + "epoch": 0.47990336042169995, + "grad_norm": 2.158294916152954, + "learning_rate": 1e-06, + "loss": 1.053, + "mean_token_accuracy": 0.6901442408561707, + "num_tokens": 111842645.0, + "step": 4370 + }, + { + "epoch": 0.48001317812431366, + "grad_norm": 2.2052600383758545, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7263305187225342, + "num_tokens": 111866549.0, + "step": 4371 + }, + { + "epoch": 0.4801229958269273, + "grad_norm": 2.2408618927001953, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.703157365322113, + "num_tokens": 111892752.0, + "step": 4372 + }, + { + "epoch": 0.48023281352954095, + "grad_norm": 2.4884769916534424, + "learning_rate": 1e-06, + "loss": 1.0218, + "mean_token_accuracy": 0.6945458054542542, + "num_tokens": 111915369.0, + "step": 4373 + }, + { + "epoch": 0.48034263123215465, + "grad_norm": 2.1147801876068115, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6811343431472778, + "num_tokens": 111944553.0, + "step": 4374 + }, + { + "epoch": 0.4804524489347683, + "grad_norm": 2.0823206901550293, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7077237963676453, + "num_tokens": 111971708.0, + "step": 4375 + }, + { + "epoch": 0.48056226663738194, + "grad_norm": 2.0887937545776367, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.6983165740966797, + "num_tokens": 111999184.0, + "step": 4376 + }, + { + "epoch": 0.4806720843399956, + "grad_norm": 2.3940341472625732, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7139605283737183, + "num_tokens": 112020873.0, + "step": 4377 + }, + { + "epoch": 0.4807819020426093, + "grad_norm": 2.1385269165039062, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7238121032714844, + "num_tokens": 112046372.0, + "step": 4378 + }, + { + "epoch": 0.48089171974522293, + "grad_norm": 2.445265293121338, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7111751437187195, + "num_tokens": 112068895.0, + "step": 4379 + }, + { + "epoch": 0.4810015374478366, + "grad_norm": 2.5025815963745117, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6919793486595154, + "num_tokens": 112089763.0, + "step": 4380 + }, + { + "epoch": 0.4811113551504502, + "grad_norm": 2.179272413253784, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.705157995223999, + "num_tokens": 112115305.0, + "step": 4381 + }, + { + "epoch": 0.4812211728530639, + "grad_norm": 2.067436456680298, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7019503712654114, + "num_tokens": 112141772.0, + "step": 4382 + }, + { + "epoch": 0.4813309905556776, + "grad_norm": 2.274022102355957, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7139735221862793, + "num_tokens": 112164620.0, + "step": 4383 + }, + { + "epoch": 0.4814408082582912, + "grad_norm": 2.284330129623413, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.6963783502578735, + "num_tokens": 112190297.0, + "step": 4384 + }, + { + "epoch": 0.4815506259609049, + "grad_norm": 2.3954691886901855, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.6986526250839233, + "num_tokens": 112211961.0, + "step": 4385 + }, + { + "epoch": 0.48166044366351857, + "grad_norm": 2.16037654876709, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7338846921920776, + "num_tokens": 112237170.0, + "step": 4386 + }, + { + "epoch": 0.4817702613661322, + "grad_norm": 2.6278767585754395, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7079516649246216, + "num_tokens": 112256822.0, + "step": 4387 + }, + { + "epoch": 0.48188007906874586, + "grad_norm": 2.212897539138794, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.716063916683197, + "num_tokens": 112279432.0, + "step": 4388 + }, + { + "epoch": 0.48198989677135956, + "grad_norm": 2.2233612537384033, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7123464345932007, + "num_tokens": 112302915.0, + "step": 4389 + }, + { + "epoch": 0.4820997144739732, + "grad_norm": 2.4543933868408203, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.7011710405349731, + "num_tokens": 112325147.0, + "step": 4390 + }, + { + "epoch": 0.48220953217658685, + "grad_norm": 2.1841628551483154, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.7134841084480286, + "num_tokens": 112351354.0, + "step": 4391 + }, + { + "epoch": 0.48231934987920055, + "grad_norm": 2.0774173736572266, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6947351098060608, + "num_tokens": 112379660.0, + "step": 4392 + }, + { + "epoch": 0.4824291675818142, + "grad_norm": 2.47735333442688, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7111291885375977, + "num_tokens": 112399816.0, + "step": 4393 + }, + { + "epoch": 0.48253898528442785, + "grad_norm": 2.7133736610412598, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7108571529388428, + "num_tokens": 112417929.0, + "step": 4394 + }, + { + "epoch": 0.4826488029870415, + "grad_norm": 2.0202865600585938, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7036042213439941, + "num_tokens": 112446540.0, + "step": 4395 + }, + { + "epoch": 0.4827586206896552, + "grad_norm": 2.038635015487671, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7281476855278015, + "num_tokens": 112472082.0, + "step": 4396 + }, + { + "epoch": 0.48286843839226884, + "grad_norm": 2.4954421520233154, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7089654803276062, + "num_tokens": 112493903.0, + "step": 4397 + }, + { + "epoch": 0.4829782560948825, + "grad_norm": 2.5362071990966797, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.6954169273376465, + "num_tokens": 112514162.0, + "step": 4398 + }, + { + "epoch": 0.48308807379749613, + "grad_norm": 2.5116231441497803, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.7048628330230713, + "num_tokens": 112534839.0, + "step": 4399 + }, + { + "epoch": 0.48319789150010983, + "grad_norm": 2.2697010040283203, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7293351292610168, + "num_tokens": 112557228.0, + "step": 4400 + }, + { + "epoch": 0.4833077092027235, + "grad_norm": 2.100407600402832, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7121983766555786, + "num_tokens": 112583078.0, + "step": 4401 + }, + { + "epoch": 0.4834175269053371, + "grad_norm": 2.314918041229248, + "learning_rate": 1e-06, + "loss": 1.0436, + "mean_token_accuracy": 0.6851553916931152, + "num_tokens": 112605991.0, + "step": 4402 + }, + { + "epoch": 0.4835273446079508, + "grad_norm": 2.1901464462280273, + "learning_rate": 1e-06, + "loss": 1.0664, + "mean_token_accuracy": 0.6780426502227783, + "num_tokens": 112631141.0, + "step": 4403 + }, + { + "epoch": 0.4836371623105645, + "grad_norm": 2.029040575027466, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7039751410484314, + "num_tokens": 112658874.0, + "step": 4404 + }, + { + "epoch": 0.4837469800131781, + "grad_norm": 2.0336785316467285, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7171602249145508, + "num_tokens": 112686945.0, + "step": 4405 + }, + { + "epoch": 0.48385679771579176, + "grad_norm": 2.079827070236206, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.7121039628982544, + "num_tokens": 112716043.0, + "step": 4406 + }, + { + "epoch": 0.48396661541840547, + "grad_norm": 2.248613119125366, + "learning_rate": 1e-06, + "loss": 1.0693, + "mean_token_accuracy": 0.688892126083374, + "num_tokens": 112743888.0, + "step": 4407 + }, + { + "epoch": 0.4840764331210191, + "grad_norm": 2.221877098083496, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.7006437182426453, + "num_tokens": 112768730.0, + "step": 4408 + }, + { + "epoch": 0.48418625082363276, + "grad_norm": 2.4513726234436035, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7035189270973206, + "num_tokens": 112789660.0, + "step": 4409 + }, + { + "epoch": 0.4842960685262464, + "grad_norm": 2.1444993019104004, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.7007120251655579, + "num_tokens": 112818700.0, + "step": 4410 + }, + { + "epoch": 0.4844058862288601, + "grad_norm": 2.5054829120635986, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.730307936668396, + "num_tokens": 112838908.0, + "step": 4411 + }, + { + "epoch": 0.48451570393147375, + "grad_norm": 2.2949559688568115, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7164590954780579, + "num_tokens": 112862816.0, + "step": 4412 + }, + { + "epoch": 0.4846255216340874, + "grad_norm": 2.3607723712921143, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7094133496284485, + "num_tokens": 112884221.0, + "step": 4413 + }, + { + "epoch": 0.4847353393367011, + "grad_norm": 2.34932279586792, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7111608982086182, + "num_tokens": 112907602.0, + "step": 4414 + }, + { + "epoch": 0.48484515703931474, + "grad_norm": 2.2536563873291016, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6979184746742249, + "num_tokens": 112933400.0, + "step": 4415 + }, + { + "epoch": 0.4849549747419284, + "grad_norm": 2.6954727172851562, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.718565821647644, + "num_tokens": 112951189.0, + "step": 4416 + }, + { + "epoch": 0.48506479244454204, + "grad_norm": 2.383788585662842, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7128359079360962, + "num_tokens": 112973485.0, + "step": 4417 + }, + { + "epoch": 0.48517461014715574, + "grad_norm": 2.2117152214050293, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.6995605230331421, + "num_tokens": 112998738.0, + "step": 4418 + }, + { + "epoch": 0.4852844278497694, + "grad_norm": 2.1871964931488037, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7264478206634521, + "num_tokens": 113023718.0, + "step": 4419 + }, + { + "epoch": 0.48539424555238303, + "grad_norm": 2.1362462043762207, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.6929901838302612, + "num_tokens": 113049952.0, + "step": 4420 + }, + { + "epoch": 0.48550406325499673, + "grad_norm": 2.466987133026123, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.706178605556488, + "num_tokens": 113071085.0, + "step": 4421 + }, + { + "epoch": 0.4856138809576104, + "grad_norm": 2.496887445449829, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.6970528364181519, + "num_tokens": 113093603.0, + "step": 4422 + }, + { + "epoch": 0.485723698660224, + "grad_norm": 2.1034858226776123, + "learning_rate": 1e-06, + "loss": 1.0713, + "mean_token_accuracy": 0.6848019361495972, + "num_tokens": 113122101.0, + "step": 4423 + }, + { + "epoch": 0.48583351636283767, + "grad_norm": 2.4879486560821533, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7357471585273743, + "num_tokens": 113143439.0, + "step": 4424 + }, + { + "epoch": 0.48594333406545137, + "grad_norm": 2.4780609607696533, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7068595290184021, + "num_tokens": 113162892.0, + "step": 4425 + }, + { + "epoch": 0.486053151768065, + "grad_norm": 2.5576069355010986, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.710342288017273, + "num_tokens": 113182307.0, + "step": 4426 + }, + { + "epoch": 0.48616296947067866, + "grad_norm": 1.899011254310608, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6847996711730957, + "num_tokens": 113213665.0, + "step": 4427 + }, + { + "epoch": 0.4862727871732923, + "grad_norm": 2.1022446155548096, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.701521635055542, + "num_tokens": 113242180.0, + "step": 4428 + }, + { + "epoch": 0.486382604875906, + "grad_norm": 2.17594051361084, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.7023854851722717, + "num_tokens": 113267623.0, + "step": 4429 + }, + { + "epoch": 0.48649242257851966, + "grad_norm": 2.2558510303497314, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6963715553283691, + "num_tokens": 113293762.0, + "step": 4430 + }, + { + "epoch": 0.4866022402811333, + "grad_norm": 2.035310983657837, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7052887678146362, + "num_tokens": 113323194.0, + "step": 4431 + }, + { + "epoch": 0.486712057983747, + "grad_norm": 1.9835362434387207, + "learning_rate": 1e-06, + "loss": 1.1003, + "mean_token_accuracy": 0.6761519908905029, + "num_tokens": 113357289.0, + "step": 4432 + }, + { + "epoch": 0.48682187568636065, + "grad_norm": 2.2406246662139893, + "learning_rate": 1e-06, + "loss": 1.0278, + "mean_token_accuracy": 0.6823233366012573, + "num_tokens": 113381782.0, + "step": 4433 + }, + { + "epoch": 0.4869316933889743, + "grad_norm": 2.054971933364868, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7064962387084961, + "num_tokens": 113410295.0, + "step": 4434 + }, + { + "epoch": 0.48704151109158794, + "grad_norm": 2.4213807582855225, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7062355279922485, + "num_tokens": 113434768.0, + "step": 4435 + }, + { + "epoch": 0.48715132879420164, + "grad_norm": 1.8755552768707275, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6997725963592529, + "num_tokens": 113466809.0, + "step": 4436 + }, + { + "epoch": 0.4872611464968153, + "grad_norm": 2.261885404586792, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.6929100751876831, + "num_tokens": 113490166.0, + "step": 4437 + }, + { + "epoch": 0.48737096419942894, + "grad_norm": 1.991119623184204, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7014158368110657, + "num_tokens": 113522452.0, + "step": 4438 + }, + { + "epoch": 0.48748078190204264, + "grad_norm": 2.156355381011963, + "learning_rate": 1e-06, + "loss": 1.0107, + "mean_token_accuracy": 0.6880694031715393, + "num_tokens": 113549217.0, + "step": 4439 + }, + { + "epoch": 0.4875905996046563, + "grad_norm": 2.213954210281372, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.700032114982605, + "num_tokens": 113574667.0, + "step": 4440 + }, + { + "epoch": 0.48770041730726993, + "grad_norm": 2.139834403991699, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7332383990287781, + "num_tokens": 113599652.0, + "step": 4441 + }, + { + "epoch": 0.4878102350098836, + "grad_norm": 2.2584376335144043, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.71427321434021, + "num_tokens": 113623442.0, + "step": 4442 + }, + { + "epoch": 0.4879200527124973, + "grad_norm": 2.1296701431274414, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6972954869270325, + "num_tokens": 113650763.0, + "step": 4443 + }, + { + "epoch": 0.4880298704151109, + "grad_norm": 2.4562435150146484, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7305064797401428, + "num_tokens": 113669455.0, + "step": 4444 + }, + { + "epoch": 0.48813968811772457, + "grad_norm": 2.265927314758301, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6897896528244019, + "num_tokens": 113693812.0, + "step": 4445 + }, + { + "epoch": 0.4882495058203382, + "grad_norm": 2.3697383403778076, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.7016426920890808, + "num_tokens": 113716094.0, + "step": 4446 + }, + { + "epoch": 0.4883593235229519, + "grad_norm": 2.373703956604004, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.707643985748291, + "num_tokens": 113739043.0, + "step": 4447 + }, + { + "epoch": 0.48846914122556556, + "grad_norm": 2.4415464401245117, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.721489429473877, + "num_tokens": 113758847.0, + "step": 4448 + }, + { + "epoch": 0.4885789589281792, + "grad_norm": 2.2388713359832764, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7074931859970093, + "num_tokens": 113782897.0, + "step": 4449 + }, + { + "epoch": 0.4886887766307929, + "grad_norm": 2.382746934890747, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7217088937759399, + "num_tokens": 113804531.0, + "step": 4450 + }, + { + "epoch": 0.48879859433340656, + "grad_norm": 2.11507511138916, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6954070329666138, + "num_tokens": 113833323.0, + "step": 4451 + }, + { + "epoch": 0.4889084120360202, + "grad_norm": 2.0734384059906006, + "learning_rate": 1e-06, + "loss": 1.036, + "mean_token_accuracy": 0.6808500289916992, + "num_tokens": 113861437.0, + "step": 4452 + }, + { + "epoch": 0.48901822973863385, + "grad_norm": 1.9860167503356934, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7125053405761719, + "num_tokens": 113891036.0, + "step": 4453 + }, + { + "epoch": 0.48912804744124755, + "grad_norm": 2.2024943828582764, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7222426533699036, + "num_tokens": 113914808.0, + "step": 4454 + }, + { + "epoch": 0.4892378651438612, + "grad_norm": 2.1648497581481934, + "learning_rate": 1e-06, + "loss": 1.0302, + "mean_token_accuracy": 0.6845816373825073, + "num_tokens": 113943142.0, + "step": 4455 + }, + { + "epoch": 0.48934768284647484, + "grad_norm": 2.4471378326416016, + "learning_rate": 1e-06, + "loss": 1.0616, + "mean_token_accuracy": 0.6843650937080383, + "num_tokens": 113966878.0, + "step": 4456 + }, + { + "epoch": 0.4894575005490885, + "grad_norm": 2.191295623779297, + "learning_rate": 1e-06, + "loss": 1.049, + "mean_token_accuracy": 0.6834858059883118, + "num_tokens": 113992592.0, + "step": 4457 + }, + { + "epoch": 0.4895673182517022, + "grad_norm": 2.173116445541382, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7139760255813599, + "num_tokens": 114020556.0, + "step": 4458 + }, + { + "epoch": 0.48967713595431583, + "grad_norm": 2.4253618717193604, + "learning_rate": 1e-06, + "loss": 1.0464, + "mean_token_accuracy": 0.6858173608779907, + "num_tokens": 114041860.0, + "step": 4459 + }, + { + "epoch": 0.4897869536569295, + "grad_norm": 2.4713315963745117, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7167871594429016, + "num_tokens": 114062212.0, + "step": 4460 + }, + { + "epoch": 0.4898967713595432, + "grad_norm": 2.3087637424468994, + "learning_rate": 1e-06, + "loss": 1.0595, + "mean_token_accuracy": 0.6865700483322144, + "num_tokens": 114087317.0, + "step": 4461 + }, + { + "epoch": 0.4900065890621568, + "grad_norm": 2.038203001022339, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7006111145019531, + "num_tokens": 114116597.0, + "step": 4462 + }, + { + "epoch": 0.4901164067647705, + "grad_norm": 2.066309690475464, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7091219425201416, + "num_tokens": 114144124.0, + "step": 4463 + }, + { + "epoch": 0.4902262244673841, + "grad_norm": 2.446927547454834, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7134883403778076, + "num_tokens": 114165680.0, + "step": 4464 + }, + { + "epoch": 0.4903360421699978, + "grad_norm": 2.894096612930298, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7208184599876404, + "num_tokens": 114181939.0, + "step": 4465 + }, + { + "epoch": 0.49044585987261147, + "grad_norm": 2.145183563232422, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7192484140396118, + "num_tokens": 114206419.0, + "step": 4466 + }, + { + "epoch": 0.4905556775752251, + "grad_norm": 2.0926191806793213, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.7015361785888672, + "num_tokens": 114234503.0, + "step": 4467 + }, + { + "epoch": 0.4906654952778388, + "grad_norm": 1.8681178092956543, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.6985771656036377, + "num_tokens": 114267365.0, + "step": 4468 + }, + { + "epoch": 0.49077531298045246, + "grad_norm": 2.116787910461426, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6978901028633118, + "num_tokens": 114294483.0, + "step": 4469 + }, + { + "epoch": 0.4908851306830661, + "grad_norm": 1.7786036729812622, + "learning_rate": 1e-06, + "loss": 1.0713, + "mean_token_accuracy": 0.6765453815460205, + "num_tokens": 114334404.0, + "step": 4470 + }, + { + "epoch": 0.49099494838567975, + "grad_norm": 2.3236217498779297, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6943657398223877, + "num_tokens": 114357891.0, + "step": 4471 + }, + { + "epoch": 0.49110476608829345, + "grad_norm": 2.036680221557617, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.6878491044044495, + "num_tokens": 114387245.0, + "step": 4472 + }, + { + "epoch": 0.4912145837909071, + "grad_norm": 2.004776954650879, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.6939691305160522, + "num_tokens": 114415710.0, + "step": 4473 + }, + { + "epoch": 0.49132440149352075, + "grad_norm": 2.7618653774261475, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7194262146949768, + "num_tokens": 114433107.0, + "step": 4474 + }, + { + "epoch": 0.4914342191961344, + "grad_norm": 1.976873755455017, + "learning_rate": 1e-06, + "loss": 1.0822, + "mean_token_accuracy": 0.6697815656661987, + "num_tokens": 114464409.0, + "step": 4475 + }, + { + "epoch": 0.4915440368987481, + "grad_norm": 2.5517942905426025, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7060509324073792, + "num_tokens": 114483439.0, + "step": 4476 + }, + { + "epoch": 0.49165385460136174, + "grad_norm": 2.1464195251464844, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7110222578048706, + "num_tokens": 114510274.0, + "step": 4477 + }, + { + "epoch": 0.4917636723039754, + "grad_norm": 2.040027618408203, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6904988288879395, + "num_tokens": 114542168.0, + "step": 4478 + }, + { + "epoch": 0.4918734900065891, + "grad_norm": 2.0026044845581055, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.6994349956512451, + "num_tokens": 114571879.0, + "step": 4479 + }, + { + "epoch": 0.49198330770920273, + "grad_norm": 2.300769567489624, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.710108757019043, + "num_tokens": 114596288.0, + "step": 4480 + }, + { + "epoch": 0.4920931254118164, + "grad_norm": 2.1129050254821777, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.706545352935791, + "num_tokens": 114622202.0, + "step": 4481 + }, + { + "epoch": 0.49220294311443, + "grad_norm": 2.1974141597747803, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7261455655097961, + "num_tokens": 114648850.0, + "step": 4482 + }, + { + "epoch": 0.4923127608170437, + "grad_norm": 2.0516395568847656, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.696807324886322, + "num_tokens": 114678609.0, + "step": 4483 + }, + { + "epoch": 0.4924225785196574, + "grad_norm": 2.209695339202881, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.6894470453262329, + "num_tokens": 114703422.0, + "step": 4484 + }, + { + "epoch": 0.492532396222271, + "grad_norm": 2.5857160091400146, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7134677767753601, + "num_tokens": 114723635.0, + "step": 4485 + }, + { + "epoch": 0.49264221392488466, + "grad_norm": 2.2691562175750732, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6961804628372192, + "num_tokens": 114750471.0, + "step": 4486 + }, + { + "epoch": 0.49275203162749837, + "grad_norm": 2.033740758895874, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.7067422866821289, + "num_tokens": 114779917.0, + "step": 4487 + }, + { + "epoch": 0.492861849330112, + "grad_norm": 2.3488399982452393, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7083226442337036, + "num_tokens": 114804333.0, + "step": 4488 + }, + { + "epoch": 0.49297166703272566, + "grad_norm": 2.222529888153076, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7115060091018677, + "num_tokens": 114828961.0, + "step": 4489 + }, + { + "epoch": 0.49308148473533936, + "grad_norm": 2.238750457763672, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7025724649429321, + "num_tokens": 114853454.0, + "step": 4490 + }, + { + "epoch": 0.493191302437953, + "grad_norm": 2.2014949321746826, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7002266645431519, + "num_tokens": 114878876.0, + "step": 4491 + }, + { + "epoch": 0.49330112014056665, + "grad_norm": 2.082808017730713, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7014114260673523, + "num_tokens": 114906792.0, + "step": 4492 + }, + { + "epoch": 0.4934109378431803, + "grad_norm": 2.2297861576080322, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.6970074772834778, + "num_tokens": 114932748.0, + "step": 4493 + }, + { + "epoch": 0.493520755545794, + "grad_norm": 2.330632209777832, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7092881202697754, + "num_tokens": 114956212.0, + "step": 4494 + }, + { + "epoch": 0.49363057324840764, + "grad_norm": 2.287292003631592, + "learning_rate": 1e-06, + "loss": 1.068, + "mean_token_accuracy": 0.6802611351013184, + "num_tokens": 114980627.0, + "step": 4495 + }, + { + "epoch": 0.4937403909510213, + "grad_norm": 1.936207890510559, + "learning_rate": 1e-06, + "loss": 1.0591, + "mean_token_accuracy": 0.6806435585021973, + "num_tokens": 115015789.0, + "step": 4496 + }, + { + "epoch": 0.493850208653635, + "grad_norm": 2.407057046890259, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.7015180587768555, + "num_tokens": 115038929.0, + "step": 4497 + }, + { + "epoch": 0.49396002635624864, + "grad_norm": 2.4181408882141113, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.723759651184082, + "num_tokens": 115060391.0, + "step": 4498 + }, + { + "epoch": 0.4940698440588623, + "grad_norm": 2.2775917053222656, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7132492065429688, + "num_tokens": 115083412.0, + "step": 4499 + }, + { + "epoch": 0.49417966176147593, + "grad_norm": 2.1968820095062256, + "learning_rate": 1e-06, + "loss": 1.0399, + "mean_token_accuracy": 0.6890383958816528, + "num_tokens": 115111273.0, + "step": 4500 + }, + { + "epoch": 0.49428947946408963, + "grad_norm": 2.2606914043426514, + "learning_rate": 1e-06, + "loss": 1.0668, + "mean_token_accuracy": 0.6985163688659668, + "num_tokens": 115134770.0, + "step": 4501 + }, + { + "epoch": 0.4943992971667033, + "grad_norm": 2.0680699348449707, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.6996935606002808, + "num_tokens": 115163975.0, + "step": 4502 + }, + { + "epoch": 0.4945091148693169, + "grad_norm": 2.196617841720581, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.70877605676651, + "num_tokens": 115189367.0, + "step": 4503 + }, + { + "epoch": 0.49461893257193057, + "grad_norm": 2.147233724594116, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7136384844779968, + "num_tokens": 115214316.0, + "step": 4504 + }, + { + "epoch": 0.49472875027454427, + "grad_norm": 2.288740634918213, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.734220027923584, + "num_tokens": 115236919.0, + "step": 4505 + }, + { + "epoch": 0.4948385679771579, + "grad_norm": 2.2219154834747314, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7114776968955994, + "num_tokens": 115259231.0, + "step": 4506 + }, + { + "epoch": 0.49494838567977156, + "grad_norm": 2.231417417526245, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.6905744671821594, + "num_tokens": 115284561.0, + "step": 4507 + }, + { + "epoch": 0.49505820338238526, + "grad_norm": 2.2343404293060303, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7086734771728516, + "num_tokens": 115308045.0, + "step": 4508 + }, + { + "epoch": 0.4951680210849989, + "grad_norm": 2.105973958969116, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7207872867584229, + "num_tokens": 115336182.0, + "step": 4509 + }, + { + "epoch": 0.49527783878761256, + "grad_norm": 2.321221113204956, + "learning_rate": 1e-06, + "loss": 1.0875, + "mean_token_accuracy": 0.6815820932388306, + "num_tokens": 115362638.0, + "step": 4510 + }, + { + "epoch": 0.4953876564902262, + "grad_norm": 2.100142002105713, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.7088437676429749, + "num_tokens": 115391323.0, + "step": 4511 + }, + { + "epoch": 0.4954974741928399, + "grad_norm": 2.4260072708129883, + "learning_rate": 1e-06, + "loss": 1.068, + "mean_token_accuracy": 0.6949338912963867, + "num_tokens": 115414870.0, + "step": 4512 + }, + { + "epoch": 0.49560729189545355, + "grad_norm": 2.1962594985961914, + "learning_rate": 1e-06, + "loss": 1.0475, + "mean_token_accuracy": 0.681415855884552, + "num_tokens": 115440616.0, + "step": 4513 + }, + { + "epoch": 0.4957171095980672, + "grad_norm": 2.2470240592956543, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7455195188522339, + "num_tokens": 115462342.0, + "step": 4514 + }, + { + "epoch": 0.4958269273006809, + "grad_norm": 2.429776668548584, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7127360105514526, + "num_tokens": 115482444.0, + "step": 4515 + }, + { + "epoch": 0.49593674500329454, + "grad_norm": 2.0697615146636963, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.7070249319076538, + "num_tokens": 115511288.0, + "step": 4516 + }, + { + "epoch": 0.4960465627059082, + "grad_norm": 2.3682644367218018, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.7035357356071472, + "num_tokens": 115532200.0, + "step": 4517 + }, + { + "epoch": 0.49615638040852184, + "grad_norm": 2.2639877796173096, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7057191133499146, + "num_tokens": 115556681.0, + "step": 4518 + }, + { + "epoch": 0.49626619811113554, + "grad_norm": 2.351720094680786, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7047882676124573, + "num_tokens": 115578582.0, + "step": 4519 + }, + { + "epoch": 0.4963760158137492, + "grad_norm": 2.4887654781341553, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7224934101104736, + "num_tokens": 115600319.0, + "step": 4520 + }, + { + "epoch": 0.49648583351636283, + "grad_norm": 2.068446397781372, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7049205303192139, + "num_tokens": 115629363.0, + "step": 4521 + }, + { + "epoch": 0.4965956512189765, + "grad_norm": 2.1440224647521973, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6923723816871643, + "num_tokens": 115659616.0, + "step": 4522 + }, + { + "epoch": 0.4967054689215902, + "grad_norm": 2.3477249145507812, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6989023685455322, + "num_tokens": 115683559.0, + "step": 4523 + }, + { + "epoch": 0.4968152866242038, + "grad_norm": 2.0527772903442383, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.71689373254776, + "num_tokens": 115712429.0, + "step": 4524 + }, + { + "epoch": 0.49692510432681747, + "grad_norm": 2.4718830585479736, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7166935801506042, + "num_tokens": 115732371.0, + "step": 4525 + }, + { + "epoch": 0.49703492202943117, + "grad_norm": 2.3909289836883545, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7167460918426514, + "num_tokens": 115754539.0, + "step": 4526 + }, + { + "epoch": 0.4971447397320448, + "grad_norm": 2.185235023498535, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.7027482986450195, + "num_tokens": 115779438.0, + "step": 4527 + }, + { + "epoch": 0.49725455743465846, + "grad_norm": 2.4715688228607178, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7014179229736328, + "num_tokens": 115799641.0, + "step": 4528 + }, + { + "epoch": 0.4973643751372721, + "grad_norm": 2.217848062515259, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7086759805679321, + "num_tokens": 115823242.0, + "step": 4529 + }, + { + "epoch": 0.4974741928398858, + "grad_norm": 2.3838555812835693, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.702820360660553, + "num_tokens": 115846791.0, + "step": 4530 + }, + { + "epoch": 0.49758401054249946, + "grad_norm": 2.2336175441741943, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7076553106307983, + "num_tokens": 115870798.0, + "step": 4531 + }, + { + "epoch": 0.4976938282451131, + "grad_norm": 2.198626756668091, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7089858651161194, + "num_tokens": 115896824.0, + "step": 4532 + }, + { + "epoch": 0.49780364594772675, + "grad_norm": 2.222581386566162, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7006280422210693, + "num_tokens": 115920549.0, + "step": 4533 + }, + { + "epoch": 0.49791346365034045, + "grad_norm": 2.5439000129699707, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7367224097251892, + "num_tokens": 115939964.0, + "step": 4534 + }, + { + "epoch": 0.4980232813529541, + "grad_norm": 2.297642469406128, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6962364912033081, + "num_tokens": 115962039.0, + "step": 4535 + }, + { + "epoch": 0.49813309905556774, + "grad_norm": 2.141146421432495, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7221837043762207, + "num_tokens": 115989182.0, + "step": 4536 + }, + { + "epoch": 0.49824291675818144, + "grad_norm": 2.3121960163116455, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7078573703765869, + "num_tokens": 116013178.0, + "step": 4537 + }, + { + "epoch": 0.4983527344607951, + "grad_norm": 2.1983234882354736, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.7012511491775513, + "num_tokens": 116038225.0, + "step": 4538 + }, + { + "epoch": 0.49846255216340873, + "grad_norm": 2.09281587600708, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7167204022407532, + "num_tokens": 116065983.0, + "step": 4539 + }, + { + "epoch": 0.4985723698660224, + "grad_norm": 2.2302823066711426, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.6882365345954895, + "num_tokens": 116091585.0, + "step": 4540 + }, + { + "epoch": 0.4986821875686361, + "grad_norm": 2.1558616161346436, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7111757397651672, + "num_tokens": 116117032.0, + "step": 4541 + }, + { + "epoch": 0.4987920052712497, + "grad_norm": 2.679110050201416, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6965014934539795, + "num_tokens": 116135610.0, + "step": 4542 + }, + { + "epoch": 0.4989018229738634, + "grad_norm": 2.3144919872283936, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.705439567565918, + "num_tokens": 116158326.0, + "step": 4543 + }, + { + "epoch": 0.4990116406764771, + "grad_norm": 2.029839515686035, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7131106853485107, + "num_tokens": 116184912.0, + "step": 4544 + }, + { + "epoch": 0.4991214583790907, + "grad_norm": 2.2198805809020996, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6867808699607849, + "num_tokens": 116211020.0, + "step": 4545 + }, + { + "epoch": 0.49923127608170437, + "grad_norm": 2.1629652976989746, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6973835825920105, + "num_tokens": 116238238.0, + "step": 4546 + }, + { + "epoch": 0.499341093784318, + "grad_norm": 2.250222682952881, + "learning_rate": 1e-06, + "loss": 1.0316, + "mean_token_accuracy": 0.687006413936615, + "num_tokens": 116265239.0, + "step": 4547 + }, + { + "epoch": 0.4994509114869317, + "grad_norm": 2.1373178958892822, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7037380933761597, + "num_tokens": 116292499.0, + "step": 4548 + }, + { + "epoch": 0.49956072918954536, + "grad_norm": 2.2032463550567627, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6944107413291931, + "num_tokens": 116317186.0, + "step": 4549 + }, + { + "epoch": 0.499670546892159, + "grad_norm": 1.9906941652297974, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7049401998519897, + "num_tokens": 116345169.0, + "step": 4550 + }, + { + "epoch": 0.49978036459477265, + "grad_norm": 2.048727035522461, + "learning_rate": 1e-06, + "loss": 1.0542, + "mean_token_accuracy": 0.6811161041259766, + "num_tokens": 116374407.0, + "step": 4551 + }, + { + "epoch": 0.49989018229738635, + "grad_norm": 2.299420118331909, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7062094211578369, + "num_tokens": 116397067.0, + "step": 4552 + }, + { + "epoch": 0.5, + "grad_norm": 2.0668673515319824, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7065768241882324, + "num_tokens": 116424208.0, + "step": 4553 + }, + { + "epoch": 0.5001098177026136, + "grad_norm": 2.1650781631469727, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7100070714950562, + "num_tokens": 116449437.0, + "step": 4554 + }, + { + "epoch": 0.5002196354052273, + "grad_norm": 2.142477512359619, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7117058038711548, + "num_tokens": 116475110.0, + "step": 4555 + }, + { + "epoch": 0.5003294531078409, + "grad_norm": 2.165140151977539, + "learning_rate": 1e-06, + "loss": 1.0735, + "mean_token_accuracy": 0.6753106117248535, + "num_tokens": 116501900.0, + "step": 4556 + }, + { + "epoch": 0.5004392708104547, + "grad_norm": 2.3199141025543213, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6895266771316528, + "num_tokens": 116526352.0, + "step": 4557 + }, + { + "epoch": 0.5005490885130683, + "grad_norm": 1.9604806900024414, + "learning_rate": 1e-06, + "loss": 1.1069, + "mean_token_accuracy": 0.6692266464233398, + "num_tokens": 116557588.0, + "step": 4558 + }, + { + "epoch": 0.500658906215682, + "grad_norm": 2.2008960247039795, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7041231393814087, + "num_tokens": 116583597.0, + "step": 4559 + }, + { + "epoch": 0.5007687239182956, + "grad_norm": 2.7039034366607666, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7325316667556763, + "num_tokens": 116599869.0, + "step": 4560 + }, + { + "epoch": 0.5008785416209093, + "grad_norm": 2.25305438041687, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.693420946598053, + "num_tokens": 116623110.0, + "step": 4561 + }, + { + "epoch": 0.5009883593235229, + "grad_norm": 2.486806869506836, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7011587619781494, + "num_tokens": 116643397.0, + "step": 4562 + }, + { + "epoch": 0.5010981770261366, + "grad_norm": 2.112537384033203, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7142648696899414, + "num_tokens": 116668231.0, + "step": 4563 + }, + { + "epoch": 0.5012079947287503, + "grad_norm": 2.4639971256256104, + "learning_rate": 1e-06, + "loss": 1.0117, + "mean_token_accuracy": 0.7117760181427002, + "num_tokens": 116688029.0, + "step": 4564 + }, + { + "epoch": 0.501317812431364, + "grad_norm": 2.2090351581573486, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.6974970102310181, + "num_tokens": 116713151.0, + "step": 4565 + }, + { + "epoch": 0.5014276301339776, + "grad_norm": 2.576144218444824, + "learning_rate": 1e-06, + "loss": 1.0275, + "mean_token_accuracy": 0.691735029220581, + "num_tokens": 116733202.0, + "step": 4566 + }, + { + "epoch": 0.5015374478365913, + "grad_norm": 2.4691240787506104, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6925997734069824, + "num_tokens": 116756133.0, + "step": 4567 + }, + { + "epoch": 0.5016472655392049, + "grad_norm": 2.286586046218872, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7063167095184326, + "num_tokens": 116780899.0, + "step": 4568 + }, + { + "epoch": 0.5017570832418186, + "grad_norm": 2.2962327003479004, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7185070514678955, + "num_tokens": 116804691.0, + "step": 4569 + }, + { + "epoch": 0.5018669009444322, + "grad_norm": 2.3665401935577393, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.6882458925247192, + "num_tokens": 116828122.0, + "step": 4570 + }, + { + "epoch": 0.5019767186470458, + "grad_norm": 2.236353874206543, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.720016598701477, + "num_tokens": 116853125.0, + "step": 4571 + }, + { + "epoch": 0.5020865363496596, + "grad_norm": 2.232738494873047, + "learning_rate": 1e-06, + "loss": 1.0442, + "mean_token_accuracy": 0.685020923614502, + "num_tokens": 116880004.0, + "step": 4572 + }, + { + "epoch": 0.5021963540522733, + "grad_norm": 2.1927571296691895, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6919281482696533, + "num_tokens": 116905412.0, + "step": 4573 + }, + { + "epoch": 0.5023061717548869, + "grad_norm": 2.4117534160614014, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6923250555992126, + "num_tokens": 116927969.0, + "step": 4574 + }, + { + "epoch": 0.5024159894575005, + "grad_norm": 2.170652389526367, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6979273557662964, + "num_tokens": 116955154.0, + "step": 4575 + }, + { + "epoch": 0.5025258071601142, + "grad_norm": 2.4483373165130615, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.7060239315032959, + "num_tokens": 116978210.0, + "step": 4576 + }, + { + "epoch": 0.5026356248627278, + "grad_norm": 2.2524149417877197, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7030820250511169, + "num_tokens": 117001470.0, + "step": 4577 + }, + { + "epoch": 0.5027454425653415, + "grad_norm": 2.557324171066284, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7290515899658203, + "num_tokens": 117021625.0, + "step": 4578 + }, + { + "epoch": 0.5028552602679552, + "grad_norm": 2.5101840496063232, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7121757864952087, + "num_tokens": 117043385.0, + "step": 4579 + }, + { + "epoch": 0.5029650779705689, + "grad_norm": 2.2877860069274902, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.698239803314209, + "num_tokens": 117066474.0, + "step": 4580 + }, + { + "epoch": 0.5030748956731825, + "grad_norm": 2.044447183609009, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.712388277053833, + "num_tokens": 117093207.0, + "step": 4581 + }, + { + "epoch": 0.5031847133757962, + "grad_norm": 2.285658359527588, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7104607820510864, + "num_tokens": 117116955.0, + "step": 4582 + }, + { + "epoch": 0.5032945310784098, + "grad_norm": 2.3955252170562744, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.7021009922027588, + "num_tokens": 117140644.0, + "step": 4583 + }, + { + "epoch": 0.5034043487810235, + "grad_norm": 2.3713364601135254, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7095479965209961, + "num_tokens": 117163111.0, + "step": 4584 + }, + { + "epoch": 0.5035141664836371, + "grad_norm": 1.885058045387268, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7137348651885986, + "num_tokens": 117194805.0, + "step": 4585 + }, + { + "epoch": 0.5036239841862509, + "grad_norm": 1.979578971862793, + "learning_rate": 1e-06, + "loss": 1.0803, + "mean_token_accuracy": 0.6794871091842651, + "num_tokens": 117226768.0, + "step": 4586 + }, + { + "epoch": 0.5037338018888645, + "grad_norm": 1.8827996253967285, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6957207918167114, + "num_tokens": 117257966.0, + "step": 4587 + }, + { + "epoch": 0.5038436195914782, + "grad_norm": 2.2020339965820312, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.7077810764312744, + "num_tokens": 117283101.0, + "step": 4588 + }, + { + "epoch": 0.5039534372940918, + "grad_norm": 2.186685562133789, + "learning_rate": 1e-06, + "loss": 1.0451, + "mean_token_accuracy": 0.6960458755493164, + "num_tokens": 117309478.0, + "step": 4589 + }, + { + "epoch": 0.5040632549967055, + "grad_norm": 1.837653636932373, + "learning_rate": 1e-06, + "loss": 1.1413, + "mean_token_accuracy": 0.672980546951294, + "num_tokens": 117347297.0, + "step": 4590 + }, + { + "epoch": 0.5041730726993191, + "grad_norm": 2.01595139503479, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7016266584396362, + "num_tokens": 117377837.0, + "step": 4591 + }, + { + "epoch": 0.5042828904019327, + "grad_norm": 2.207771062850952, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7141321897506714, + "num_tokens": 117403235.0, + "step": 4592 + }, + { + "epoch": 0.5043927081045465, + "grad_norm": 2.127023935317993, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.7016603946685791, + "num_tokens": 117431137.0, + "step": 4593 + }, + { + "epoch": 0.5045025258071602, + "grad_norm": 2.470526933670044, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7121386528015137, + "num_tokens": 117453842.0, + "step": 4594 + }, + { + "epoch": 0.5046123435097738, + "grad_norm": 2.2262022495269775, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6942785382270813, + "num_tokens": 117480841.0, + "step": 4595 + }, + { + "epoch": 0.5047221612123874, + "grad_norm": 2.137101888656616, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7282123565673828, + "num_tokens": 117507167.0, + "step": 4596 + }, + { + "epoch": 0.5048319789150011, + "grad_norm": 2.2026586532592773, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.7037820816040039, + "num_tokens": 117532766.0, + "step": 4597 + }, + { + "epoch": 0.5049417966176147, + "grad_norm": 2.3069777488708496, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7023528814315796, + "num_tokens": 117555616.0, + "step": 4598 + }, + { + "epoch": 0.5050516143202284, + "grad_norm": 2.2051477432250977, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.6981212496757507, + "num_tokens": 117581581.0, + "step": 4599 + }, + { + "epoch": 0.505161432022842, + "grad_norm": 2.301076889038086, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6901540160179138, + "num_tokens": 117606634.0, + "step": 4600 + }, + { + "epoch": 0.5052712497254558, + "grad_norm": 2.4273993968963623, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6959566473960876, + "num_tokens": 117629078.0, + "step": 4601 + }, + { + "epoch": 0.5053810674280694, + "grad_norm": 2.105496883392334, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.6922208070755005, + "num_tokens": 117655725.0, + "step": 4602 + }, + { + "epoch": 0.5054908851306831, + "grad_norm": 2.17132568359375, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6906110048294067, + "num_tokens": 117682131.0, + "step": 4603 + }, + { + "epoch": 0.5056007028332967, + "grad_norm": 2.296182870864868, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7154613137245178, + "num_tokens": 117705301.0, + "step": 4604 + }, + { + "epoch": 0.5057105205359104, + "grad_norm": 2.2428505420684814, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7186403274536133, + "num_tokens": 117730567.0, + "step": 4605 + }, + { + "epoch": 0.505820338238524, + "grad_norm": 2.210557699203491, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.6894976496696472, + "num_tokens": 117757868.0, + "step": 4606 + }, + { + "epoch": 0.5059301559411377, + "grad_norm": 2.111456871032715, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7094979286193848, + "num_tokens": 117784781.0, + "step": 4607 + }, + { + "epoch": 0.5060399736437514, + "grad_norm": 2.110738515853882, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.6946318745613098, + "num_tokens": 117811256.0, + "step": 4608 + }, + { + "epoch": 0.5061497913463651, + "grad_norm": 2.2857301235198975, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.698371171951294, + "num_tokens": 117836652.0, + "step": 4609 + }, + { + "epoch": 0.5062596090489787, + "grad_norm": 2.0087099075317383, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6987329721450806, + "num_tokens": 117868212.0, + "step": 4610 + }, + { + "epoch": 0.5063694267515924, + "grad_norm": 2.1930506229400635, + "learning_rate": 1e-06, + "loss": 1.1121, + "mean_token_accuracy": 0.6688512563705444, + "num_tokens": 117896713.0, + "step": 4611 + }, + { + "epoch": 0.506479244454206, + "grad_norm": 1.9364186525344849, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.6931247711181641, + "num_tokens": 117929225.0, + "step": 4612 + }, + { + "epoch": 0.5065890621568196, + "grad_norm": 2.0415306091308594, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7249228954315186, + "num_tokens": 117956791.0, + "step": 4613 + }, + { + "epoch": 0.5066988798594333, + "grad_norm": 2.200007915496826, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7048807144165039, + "num_tokens": 117982263.0, + "step": 4614 + }, + { + "epoch": 0.506808697562047, + "grad_norm": 2.1899335384368896, + "learning_rate": 1e-06, + "loss": 1.0518, + "mean_token_accuracy": 0.6795867681503296, + "num_tokens": 118009375.0, + "step": 4615 + }, + { + "epoch": 0.5069185152646607, + "grad_norm": 2.2406060695648193, + "learning_rate": 1e-06, + "loss": 1.0639, + "mean_token_accuracy": 0.6838354468345642, + "num_tokens": 118035589.0, + "step": 4616 + }, + { + "epoch": 0.5070283329672743, + "grad_norm": 2.048841714859009, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6939815282821655, + "num_tokens": 118064131.0, + "step": 4617 + }, + { + "epoch": 0.507138150669888, + "grad_norm": 2.117776393890381, + "learning_rate": 1e-06, + "loss": 1.028, + "mean_token_accuracy": 0.6877668499946594, + "num_tokens": 118091633.0, + "step": 4618 + }, + { + "epoch": 0.5072479683725016, + "grad_norm": 2.2414181232452393, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7131506204605103, + "num_tokens": 118117300.0, + "step": 4619 + }, + { + "epoch": 0.5073577860751153, + "grad_norm": 2.3228306770324707, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7261447906494141, + "num_tokens": 118138246.0, + "step": 4620 + }, + { + "epoch": 0.5074676037777289, + "grad_norm": 2.256399393081665, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7020631432533264, + "num_tokens": 118162276.0, + "step": 4621 + }, + { + "epoch": 0.5075774214803427, + "grad_norm": 2.3962438106536865, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6917843222618103, + "num_tokens": 118185802.0, + "step": 4622 + }, + { + "epoch": 0.5076872391829563, + "grad_norm": 2.074526071548462, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7141671180725098, + "num_tokens": 118213262.0, + "step": 4623 + }, + { + "epoch": 0.50779705688557, + "grad_norm": 2.1922190189361572, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7064074277877808, + "num_tokens": 118238159.0, + "step": 4624 + }, + { + "epoch": 0.5079068745881836, + "grad_norm": 2.051309108734131, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7039139270782471, + "num_tokens": 118267294.0, + "step": 4625 + }, + { + "epoch": 0.5080166922907973, + "grad_norm": 2.421950340270996, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6947081685066223, + "num_tokens": 118290171.0, + "step": 4626 + }, + { + "epoch": 0.5081265099934109, + "grad_norm": 2.153331756591797, + "learning_rate": 1e-06, + "loss": 1.0488, + "mean_token_accuracy": 0.6861811876296997, + "num_tokens": 118317546.0, + "step": 4627 + }, + { + "epoch": 0.5082363276960246, + "grad_norm": 2.0175118446350098, + "learning_rate": 1e-06, + "loss": 1.0677, + "mean_token_accuracy": 0.6791955232620239, + "num_tokens": 118346952.0, + "step": 4628 + }, + { + "epoch": 0.5083461453986382, + "grad_norm": 1.956490397453308, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7093234062194824, + "num_tokens": 118377254.0, + "step": 4629 + }, + { + "epoch": 0.508455963101252, + "grad_norm": 2.142007350921631, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7044166326522827, + "num_tokens": 118403932.0, + "step": 4630 + }, + { + "epoch": 0.5085657808038656, + "grad_norm": 2.151423931121826, + "learning_rate": 1e-06, + "loss": 1.048, + "mean_token_accuracy": 0.6879993081092834, + "num_tokens": 118432169.0, + "step": 4631 + }, + { + "epoch": 0.5086755985064793, + "grad_norm": 2.2006421089172363, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6941676735877991, + "num_tokens": 118459862.0, + "step": 4632 + }, + { + "epoch": 0.5087854162090929, + "grad_norm": 2.2689528465270996, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.6973772644996643, + "num_tokens": 118484576.0, + "step": 4633 + }, + { + "epoch": 0.5088952339117065, + "grad_norm": 2.379108190536499, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.7048157453536987, + "num_tokens": 118507716.0, + "step": 4634 + }, + { + "epoch": 0.5090050516143202, + "grad_norm": 2.2067055702209473, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7317637205123901, + "num_tokens": 118531126.0, + "step": 4635 + }, + { + "epoch": 0.5091148693169338, + "grad_norm": 2.3557772636413574, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7087136507034302, + "num_tokens": 118555264.0, + "step": 4636 + }, + { + "epoch": 0.5092246870195476, + "grad_norm": 2.2547504901885986, + "learning_rate": 1e-06, + "loss": 1.0416, + "mean_token_accuracy": 0.686096727848053, + "num_tokens": 118581391.0, + "step": 4637 + }, + { + "epoch": 0.5093345047221612, + "grad_norm": 2.0966506004333496, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7150710225105286, + "num_tokens": 118607392.0, + "step": 4638 + }, + { + "epoch": 0.5094443224247749, + "grad_norm": 1.923669695854187, + "learning_rate": 1e-06, + "loss": 1.0274, + "mean_token_accuracy": 0.6969996690750122, + "num_tokens": 118640382.0, + "step": 4639 + }, + { + "epoch": 0.5095541401273885, + "grad_norm": 2.2992961406707764, + "learning_rate": 1e-06, + "loss": 1.0392, + "mean_token_accuracy": 0.689691960811615, + "num_tokens": 118665480.0, + "step": 4640 + }, + { + "epoch": 0.5096639578300022, + "grad_norm": 2.236382484436035, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7077445983886719, + "num_tokens": 118690914.0, + "step": 4641 + }, + { + "epoch": 0.5097737755326158, + "grad_norm": 2.1489920616149902, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.7013243436813354, + "num_tokens": 118717659.0, + "step": 4642 + }, + { + "epoch": 0.5098835932352295, + "grad_norm": 1.917945384979248, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7165875434875488, + "num_tokens": 118748768.0, + "step": 4643 + }, + { + "epoch": 0.5099934109378432, + "grad_norm": 2.419290065765381, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7128299474716187, + "num_tokens": 118769864.0, + "step": 4644 + }, + { + "epoch": 0.5101032286404569, + "grad_norm": 1.964856505393982, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7038291692733765, + "num_tokens": 118801295.0, + "step": 4645 + }, + { + "epoch": 0.5102130463430705, + "grad_norm": 2.5370190143585205, + "learning_rate": 1e-06, + "loss": 1.0487, + "mean_token_accuracy": 0.6851611733436584, + "num_tokens": 118822925.0, + "step": 4646 + }, + { + "epoch": 0.5103228640456842, + "grad_norm": 2.5995073318481445, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6986985206604004, + "num_tokens": 118843670.0, + "step": 4647 + }, + { + "epoch": 0.5104326817482978, + "grad_norm": 1.993188500404358, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.6995232105255127, + "num_tokens": 118872178.0, + "step": 4648 + }, + { + "epoch": 0.5105424994509115, + "grad_norm": 2.2181897163391113, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6904029846191406, + "num_tokens": 118897724.0, + "step": 4649 + }, + { + "epoch": 0.5106523171535251, + "grad_norm": 2.0879976749420166, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.7030949592590332, + "num_tokens": 118924671.0, + "step": 4650 + }, + { + "epoch": 0.5107621348561389, + "grad_norm": 2.2914798259735107, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7311955094337463, + "num_tokens": 118946753.0, + "step": 4651 + }, + { + "epoch": 0.5108719525587525, + "grad_norm": 2.44793963432312, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.693397045135498, + "num_tokens": 118967808.0, + "step": 4652 + }, + { + "epoch": 0.5109817702613662, + "grad_norm": 2.2241978645324707, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7057597637176514, + "num_tokens": 118992841.0, + "step": 4653 + }, + { + "epoch": 0.5110915879639798, + "grad_norm": 2.606581211090088, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.715704083442688, + "num_tokens": 119018692.0, + "step": 4654 + }, + { + "epoch": 0.5112014056665934, + "grad_norm": 2.099790096282959, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7215815782546997, + "num_tokens": 119044173.0, + "step": 4655 + }, + { + "epoch": 0.5113112233692071, + "grad_norm": 1.8685952425003052, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7273852229118347, + "num_tokens": 119073834.0, + "step": 4656 + }, + { + "epoch": 0.5114210410718207, + "grad_norm": 1.8970226049423218, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6985887885093689, + "num_tokens": 119106436.0, + "step": 4657 + }, + { + "epoch": 0.5115308587744345, + "grad_norm": 2.190755844116211, + "learning_rate": 1e-06, + "loss": 1.0328, + "mean_token_accuracy": 0.684482216835022, + "num_tokens": 119133788.0, + "step": 4658 + }, + { + "epoch": 0.5116406764770481, + "grad_norm": 2.5946102142333984, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7063753008842468, + "num_tokens": 119153333.0, + "step": 4659 + }, + { + "epoch": 0.5117504941796618, + "grad_norm": 2.0869803428649902, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.6915675401687622, + "num_tokens": 119183772.0, + "step": 4660 + }, + { + "epoch": 0.5118603118822754, + "grad_norm": 1.976183295249939, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7103258371353149, + "num_tokens": 119212911.0, + "step": 4661 + }, + { + "epoch": 0.5119701295848891, + "grad_norm": 2.103367567062378, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.6928629279136658, + "num_tokens": 119241119.0, + "step": 4662 + }, + { + "epoch": 0.5120799472875027, + "grad_norm": 1.985398530960083, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6987587213516235, + "num_tokens": 119272025.0, + "step": 4663 + }, + { + "epoch": 0.5121897649901164, + "grad_norm": 2.2924070358276367, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7176826596260071, + "num_tokens": 119294658.0, + "step": 4664 + }, + { + "epoch": 0.51229958269273, + "grad_norm": 2.417858123779297, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7073831558227539, + "num_tokens": 119316786.0, + "step": 4665 + }, + { + "epoch": 0.5124094003953438, + "grad_norm": 2.044637680053711, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7012348175048828, + "num_tokens": 119347823.0, + "step": 4666 + }, + { + "epoch": 0.5125192180979574, + "grad_norm": 2.3850481510162354, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.7083413600921631, + "num_tokens": 119371831.0, + "step": 4667 + }, + { + "epoch": 0.5126290358005711, + "grad_norm": 2.081242561340332, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7169015407562256, + "num_tokens": 119398715.0, + "step": 4668 + }, + { + "epoch": 0.5127388535031847, + "grad_norm": 2.5807552337646484, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7352293133735657, + "num_tokens": 119416560.0, + "step": 4669 + }, + { + "epoch": 0.5128486712057984, + "grad_norm": 2.1995489597320557, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7104484438896179, + "num_tokens": 119443012.0, + "step": 4670 + }, + { + "epoch": 0.512958488908412, + "grad_norm": 2.329045295715332, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7025932669639587, + "num_tokens": 119466034.0, + "step": 4671 + }, + { + "epoch": 0.5130683066110256, + "grad_norm": 2.5326120853424072, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.6987029314041138, + "num_tokens": 119487626.0, + "step": 4672 + }, + { + "epoch": 0.5131781243136394, + "grad_norm": 2.335088014602661, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.707068681716919, + "num_tokens": 119510858.0, + "step": 4673 + }, + { + "epoch": 0.513287942016253, + "grad_norm": 2.1312549114227295, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6937036514282227, + "num_tokens": 119538523.0, + "step": 4674 + }, + { + "epoch": 0.5133977597188667, + "grad_norm": 2.4655518531799316, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7149572968482971, + "num_tokens": 119559215.0, + "step": 4675 + }, + { + "epoch": 0.5135075774214803, + "grad_norm": 2.4332191944122314, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7092538475990295, + "num_tokens": 119580507.0, + "step": 4676 + }, + { + "epoch": 0.513617395124094, + "grad_norm": 2.0268919467926025, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7160173654556274, + "num_tokens": 119609962.0, + "step": 4677 + }, + { + "epoch": 0.5137272128267076, + "grad_norm": 2.286510944366455, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7296977043151855, + "num_tokens": 119633394.0, + "step": 4678 + }, + { + "epoch": 0.5138370305293213, + "grad_norm": 2.814318895339966, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7151284217834473, + "num_tokens": 119650270.0, + "step": 4679 + }, + { + "epoch": 0.513946848231935, + "grad_norm": 2.6960999965667725, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7130872011184692, + "num_tokens": 119669593.0, + "step": 4680 + }, + { + "epoch": 0.5140566659345487, + "grad_norm": 2.488464832305908, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7222559452056885, + "num_tokens": 119688756.0, + "step": 4681 + }, + { + "epoch": 0.5141664836371623, + "grad_norm": 2.423807382583618, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7101413011550903, + "num_tokens": 119709721.0, + "step": 4682 + }, + { + "epoch": 0.514276301339776, + "grad_norm": 2.11643123626709, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.7079175710678101, + "num_tokens": 119737035.0, + "step": 4683 + }, + { + "epoch": 0.5143861190423896, + "grad_norm": 2.143616199493408, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7163314819335938, + "num_tokens": 119763204.0, + "step": 4684 + }, + { + "epoch": 0.5144959367450033, + "grad_norm": 2.14984130859375, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7287589907646179, + "num_tokens": 119789313.0, + "step": 4685 + }, + { + "epoch": 0.5146057544476169, + "grad_norm": 2.215402603149414, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7076330184936523, + "num_tokens": 119817937.0, + "step": 4686 + }, + { + "epoch": 0.5147155721502307, + "grad_norm": 2.356804847717285, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.694687008857727, + "num_tokens": 119840623.0, + "step": 4687 + }, + { + "epoch": 0.5148253898528443, + "grad_norm": 2.69331693649292, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.6919608116149902, + "num_tokens": 119860746.0, + "step": 4688 + }, + { + "epoch": 0.514935207555458, + "grad_norm": 2.1869547367095947, + "learning_rate": 1e-06, + "loss": 1.0252, + "mean_token_accuracy": 0.6920132637023926, + "num_tokens": 119886029.0, + "step": 4689 + }, + { + "epoch": 0.5150450252580716, + "grad_norm": 2.264404773712158, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7187885046005249, + "num_tokens": 119911706.0, + "step": 4690 + }, + { + "epoch": 0.5151548429606853, + "grad_norm": 2.1262223720550537, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.712378978729248, + "num_tokens": 119939086.0, + "step": 4691 + }, + { + "epoch": 0.5152646606632989, + "grad_norm": 1.9954396486282349, + "learning_rate": 1e-06, + "loss": 1.0619, + "mean_token_accuracy": 0.6749825477600098, + "num_tokens": 119969178.0, + "step": 4692 + }, + { + "epoch": 0.5153744783659125, + "grad_norm": 2.382197380065918, + "learning_rate": 1e-06, + "loss": 1.0726, + "mean_token_accuracy": 0.6808856725692749, + "num_tokens": 119994013.0, + "step": 4693 + }, + { + "epoch": 0.5154842960685262, + "grad_norm": 2.5289692878723145, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7213829159736633, + "num_tokens": 120013367.0, + "step": 4694 + }, + { + "epoch": 0.51559411377114, + "grad_norm": 2.393315553665161, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.6949498653411865, + "num_tokens": 120037080.0, + "step": 4695 + }, + { + "epoch": 0.5157039314737536, + "grad_norm": 2.3879411220550537, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7293887138366699, + "num_tokens": 120059184.0, + "step": 4696 + }, + { + "epoch": 0.5158137491763672, + "grad_norm": 2.211155414581299, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.720076858997345, + "num_tokens": 120084106.0, + "step": 4697 + }, + { + "epoch": 0.5159235668789809, + "grad_norm": 2.213409662246704, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7356447577476501, + "num_tokens": 120108098.0, + "step": 4698 + }, + { + "epoch": 0.5160333845815945, + "grad_norm": 1.9909158945083618, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6895878314971924, + "num_tokens": 120140839.0, + "step": 4699 + }, + { + "epoch": 0.5161432022842082, + "grad_norm": 2.1230099201202393, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7204508781433105, + "num_tokens": 120167151.0, + "step": 4700 + }, + { + "epoch": 0.5162530199868218, + "grad_norm": 2.185818910598755, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7020119428634644, + "num_tokens": 120192882.0, + "step": 4701 + }, + { + "epoch": 0.5163628376894356, + "grad_norm": 2.1190826892852783, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7043961882591248, + "num_tokens": 120220259.0, + "step": 4702 + }, + { + "epoch": 0.5164726553920492, + "grad_norm": 2.1660990715026855, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7191835641860962, + "num_tokens": 120246565.0, + "step": 4703 + }, + { + "epoch": 0.5165824730946629, + "grad_norm": 1.841709017753601, + "learning_rate": 1e-06, + "loss": 1.0264, + "mean_token_accuracy": 0.6962332129478455, + "num_tokens": 120280230.0, + "step": 4704 + }, + { + "epoch": 0.5166922907972765, + "grad_norm": 2.2102713584899902, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7169845104217529, + "num_tokens": 120303158.0, + "step": 4705 + }, + { + "epoch": 0.5168021084998902, + "grad_norm": 2.115145683288574, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7100112438201904, + "num_tokens": 120329331.0, + "step": 4706 + }, + { + "epoch": 0.5169119262025038, + "grad_norm": 2.334406614303589, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7118702530860901, + "num_tokens": 120351285.0, + "step": 4707 + }, + { + "epoch": 0.5170217439051175, + "grad_norm": 2.6628293991088867, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7160167098045349, + "num_tokens": 120372842.0, + "step": 4708 + }, + { + "epoch": 0.5171315616077312, + "grad_norm": 2.324932336807251, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7207352519035339, + "num_tokens": 120394533.0, + "step": 4709 + }, + { + "epoch": 0.5172413793103449, + "grad_norm": 2.463618040084839, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7272211313247681, + "num_tokens": 120414014.0, + "step": 4710 + }, + { + "epoch": 0.5173511970129585, + "grad_norm": 2.119192361831665, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7227339744567871, + "num_tokens": 120441047.0, + "step": 4711 + }, + { + "epoch": 0.5174610147155722, + "grad_norm": 2.521979331970215, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7017437815666199, + "num_tokens": 120462852.0, + "step": 4712 + }, + { + "epoch": 0.5175708324181858, + "grad_norm": 2.4321601390838623, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7188660502433777, + "num_tokens": 120484164.0, + "step": 4713 + }, + { + "epoch": 0.5176806501207994, + "grad_norm": 2.149899482727051, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.694148600101471, + "num_tokens": 120509272.0, + "step": 4714 + }, + { + "epoch": 0.5177904678234131, + "grad_norm": 2.2670977115631104, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.7017319202423096, + "num_tokens": 120533389.0, + "step": 4715 + }, + { + "epoch": 0.5179002855260268, + "grad_norm": 2.2158315181732178, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.6991561651229858, + "num_tokens": 120558553.0, + "step": 4716 + }, + { + "epoch": 0.5180101032286405, + "grad_norm": 2.1019351482391357, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6909358501434326, + "num_tokens": 120587170.0, + "step": 4717 + }, + { + "epoch": 0.5181199209312541, + "grad_norm": 2.5528507232666016, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.708070695400238, + "num_tokens": 120608475.0, + "step": 4718 + }, + { + "epoch": 0.5182297386338678, + "grad_norm": 2.0219826698303223, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.7032022476196289, + "num_tokens": 120639135.0, + "step": 4719 + }, + { + "epoch": 0.5183395563364814, + "grad_norm": 2.0773513317108154, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7051679491996765, + "num_tokens": 120665876.0, + "step": 4720 + }, + { + "epoch": 0.5184493740390951, + "grad_norm": 1.8229371309280396, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7051196694374084, + "num_tokens": 120697601.0, + "step": 4721 + }, + { + "epoch": 0.5185591917417087, + "grad_norm": 2.073761224746704, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7233152985572815, + "num_tokens": 120726362.0, + "step": 4722 + }, + { + "epoch": 0.5186690094443224, + "grad_norm": 2.116727352142334, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.7007954120635986, + "num_tokens": 120753101.0, + "step": 4723 + }, + { + "epoch": 0.5187788271469361, + "grad_norm": 2.13275146484375, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6964400410652161, + "num_tokens": 120779156.0, + "step": 4724 + }, + { + "epoch": 0.5188886448495498, + "grad_norm": 2.043170690536499, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.6908426880836487, + "num_tokens": 120808806.0, + "step": 4725 + }, + { + "epoch": 0.5189984625521634, + "grad_norm": 2.353731155395508, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7027701139450073, + "num_tokens": 120831408.0, + "step": 4726 + }, + { + "epoch": 0.5191082802547771, + "grad_norm": 2.450477361679077, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7196677923202515, + "num_tokens": 120852003.0, + "step": 4727 + }, + { + "epoch": 0.5192180979573907, + "grad_norm": 2.099148750305176, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6967267990112305, + "num_tokens": 120880888.0, + "step": 4728 + }, + { + "epoch": 0.5193279156600044, + "grad_norm": 2.2397851943969727, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6971668004989624, + "num_tokens": 120906622.0, + "step": 4729 + }, + { + "epoch": 0.519437733362618, + "grad_norm": 2.435561418533325, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7173010110855103, + "num_tokens": 120927688.0, + "step": 4730 + }, + { + "epoch": 0.5195475510652318, + "grad_norm": 2.1981616020202637, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7301620841026306, + "num_tokens": 120952091.0, + "step": 4731 + }, + { + "epoch": 0.5196573687678454, + "grad_norm": 2.1766462326049805, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7152879238128662, + "num_tokens": 120978970.0, + "step": 4732 + }, + { + "epoch": 0.519767186470459, + "grad_norm": 2.0677590370178223, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6943166255950928, + "num_tokens": 121005205.0, + "step": 4733 + }, + { + "epoch": 0.5198770041730727, + "grad_norm": 2.1874685287475586, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7267987132072449, + "num_tokens": 121031647.0, + "step": 4734 + }, + { + "epoch": 0.5199868218756863, + "grad_norm": 2.317232847213745, + "learning_rate": 1e-06, + "loss": 1.1047, + "mean_token_accuracy": 0.6691442131996155, + "num_tokens": 121054636.0, + "step": 4735 + }, + { + "epoch": 0.5200966395783, + "grad_norm": 2.11187481880188, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6886200904846191, + "num_tokens": 121083376.0, + "step": 4736 + }, + { + "epoch": 0.5202064572809136, + "grad_norm": 2.2631735801696777, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7162673473358154, + "num_tokens": 121107038.0, + "step": 4737 + }, + { + "epoch": 0.5203162749835274, + "grad_norm": 2.1142501831054688, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7145574688911438, + "num_tokens": 121133107.0, + "step": 4738 + }, + { + "epoch": 0.520426092686141, + "grad_norm": 2.157677173614502, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.7009205222129822, + "num_tokens": 121159893.0, + "step": 4739 + }, + { + "epoch": 0.5205359103887547, + "grad_norm": 2.3087003231048584, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7031028270721436, + "num_tokens": 121184540.0, + "step": 4740 + }, + { + "epoch": 0.5206457280913683, + "grad_norm": 2.1147077083587646, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7181714177131653, + "num_tokens": 121209978.0, + "step": 4741 + }, + { + "epoch": 0.520755545793982, + "grad_norm": 2.3093454837799072, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7124614715576172, + "num_tokens": 121232568.0, + "step": 4742 + }, + { + "epoch": 0.5208653634965956, + "grad_norm": 2.023228168487549, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7182730436325073, + "num_tokens": 121263122.0, + "step": 4743 + }, + { + "epoch": 0.5209751811992093, + "grad_norm": 2.328624963760376, + "learning_rate": 1e-06, + "loss": 1.0894, + "mean_token_accuracy": 0.675899863243103, + "num_tokens": 121287768.0, + "step": 4744 + }, + { + "epoch": 0.521084998901823, + "grad_norm": 2.169783592224121, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7124419212341309, + "num_tokens": 121313202.0, + "step": 4745 + }, + { + "epoch": 0.5211948166044367, + "grad_norm": 2.261834144592285, + "learning_rate": 1e-06, + "loss": 1.0694, + "mean_token_accuracy": 0.6872767210006714, + "num_tokens": 121337713.0, + "step": 4746 + }, + { + "epoch": 0.5213046343070503, + "grad_norm": 2.2731456756591797, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7165013551712036, + "num_tokens": 121360813.0, + "step": 4747 + }, + { + "epoch": 0.521414452009664, + "grad_norm": 2.340120792388916, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7007073163986206, + "num_tokens": 121383501.0, + "step": 4748 + }, + { + "epoch": 0.5215242697122776, + "grad_norm": 1.8732755184173584, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.6906670331954956, + "num_tokens": 121418194.0, + "step": 4749 + }, + { + "epoch": 0.5216340874148913, + "grad_norm": 2.486780881881714, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7060315012931824, + "num_tokens": 121438131.0, + "step": 4750 + }, + { + "epoch": 0.5217439051175049, + "grad_norm": 2.305166006088257, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.7104576826095581, + "num_tokens": 121461419.0, + "step": 4751 + }, + { + "epoch": 0.5218537228201185, + "grad_norm": 2.3150248527526855, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7197878360748291, + "num_tokens": 121485394.0, + "step": 4752 + }, + { + "epoch": 0.5219635405227323, + "grad_norm": 2.1275930404663086, + "learning_rate": 1e-06, + "loss": 1.0264, + "mean_token_accuracy": 0.6893297433853149, + "num_tokens": 121513614.0, + "step": 4753 + }, + { + "epoch": 0.522073358225346, + "grad_norm": 2.054476022720337, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7035127878189087, + "num_tokens": 121543832.0, + "step": 4754 + }, + { + "epoch": 0.5221831759279596, + "grad_norm": 1.9582045078277588, + "learning_rate": 1e-06, + "loss": 1.0342, + "mean_token_accuracy": 0.6884310245513916, + "num_tokens": 121576446.0, + "step": 4755 + }, + { + "epoch": 0.5222929936305732, + "grad_norm": 2.247243881225586, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7014665603637695, + "num_tokens": 121601237.0, + "step": 4756 + }, + { + "epoch": 0.5224028113331869, + "grad_norm": 2.0900213718414307, + "learning_rate": 1e-06, + "loss": 1.0834, + "mean_token_accuracy": 0.6737778186798096, + "num_tokens": 121631208.0, + "step": 4757 + }, + { + "epoch": 0.5225126290358005, + "grad_norm": 2.0598299503326416, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.6919263601303101, + "num_tokens": 121658226.0, + "step": 4758 + }, + { + "epoch": 0.5226224467384142, + "grad_norm": 1.9477753639221191, + "learning_rate": 1e-06, + "loss": 1.0723, + "mean_token_accuracy": 0.6813400983810425, + "num_tokens": 121692661.0, + "step": 4759 + }, + { + "epoch": 0.5227322644410279, + "grad_norm": 2.100090265274048, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7273744344711304, + "num_tokens": 121718464.0, + "step": 4760 + }, + { + "epoch": 0.5228420821436416, + "grad_norm": 2.006777763366699, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7120218276977539, + "num_tokens": 121747171.0, + "step": 4761 + }, + { + "epoch": 0.5229518998462552, + "grad_norm": 2.280532121658325, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7194654941558838, + "num_tokens": 121771495.0, + "step": 4762 + }, + { + "epoch": 0.5230617175488689, + "grad_norm": 2.1666040420532227, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7317554950714111, + "num_tokens": 121794591.0, + "step": 4763 + }, + { + "epoch": 0.5231715352514825, + "grad_norm": 2.3638806343078613, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7168385982513428, + "num_tokens": 121816675.0, + "step": 4764 + }, + { + "epoch": 0.5232813529540962, + "grad_norm": 2.375485897064209, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7105296850204468, + "num_tokens": 121837262.0, + "step": 4765 + }, + { + "epoch": 0.5233911706567098, + "grad_norm": 2.236574172973633, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6987377405166626, + "num_tokens": 121863444.0, + "step": 4766 + }, + { + "epoch": 0.5235009883593236, + "grad_norm": 2.062061309814453, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7014364004135132, + "num_tokens": 121891654.0, + "step": 4767 + }, + { + "epoch": 0.5236108060619372, + "grad_norm": 2.5495705604553223, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7025071382522583, + "num_tokens": 121911678.0, + "step": 4768 + }, + { + "epoch": 0.5237206237645509, + "grad_norm": 2.232846736907959, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7063319683074951, + "num_tokens": 121938591.0, + "step": 4769 + }, + { + "epoch": 0.5238304414671645, + "grad_norm": 1.971879243850708, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7111784219741821, + "num_tokens": 121968266.0, + "step": 4770 + }, + { + "epoch": 0.5239402591697782, + "grad_norm": 2.209442615509033, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7358956933021545, + "num_tokens": 121991725.0, + "step": 4771 + }, + { + "epoch": 0.5240500768723918, + "grad_norm": 1.8915303945541382, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7260931730270386, + "num_tokens": 122022607.0, + "step": 4772 + }, + { + "epoch": 0.5241598945750054, + "grad_norm": 2.2034096717834473, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7209949493408203, + "num_tokens": 122044601.0, + "step": 4773 + }, + { + "epoch": 0.5242697122776192, + "grad_norm": 2.1490423679351807, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7082395553588867, + "num_tokens": 122069744.0, + "step": 4774 + }, + { + "epoch": 0.5243795299802329, + "grad_norm": 2.1548712253570557, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.6965981721878052, + "num_tokens": 122096271.0, + "step": 4775 + }, + { + "epoch": 0.5244893476828465, + "grad_norm": 2.0584657192230225, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6946403384208679, + "num_tokens": 122123325.0, + "step": 4776 + }, + { + "epoch": 0.5245991653854601, + "grad_norm": 2.317131519317627, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7350025177001953, + "num_tokens": 122144850.0, + "step": 4777 + }, + { + "epoch": 0.5247089830880738, + "grad_norm": 2.235830783843994, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7404036521911621, + "num_tokens": 122168893.0, + "step": 4778 + }, + { + "epoch": 0.5248188007906874, + "grad_norm": 2.7076635360717773, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.6934939026832581, + "num_tokens": 122188216.0, + "step": 4779 + }, + { + "epoch": 0.5249286184933011, + "grad_norm": 2.1168062686920166, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.7074718475341797, + "num_tokens": 122215797.0, + "step": 4780 + }, + { + "epoch": 0.5250384361959147, + "grad_norm": 2.183147430419922, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6941561698913574, + "num_tokens": 122243610.0, + "step": 4781 + }, + { + "epoch": 0.5251482538985285, + "grad_norm": 2.2804360389709473, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.7016444206237793, + "num_tokens": 122267806.0, + "step": 4782 + }, + { + "epoch": 0.5252580716011421, + "grad_norm": 2.184884786605835, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6946526765823364, + "num_tokens": 122295529.0, + "step": 4783 + }, + { + "epoch": 0.5253678893037558, + "grad_norm": 2.0688107013702393, + "learning_rate": 1e-06, + "loss": 1.0687, + "mean_token_accuracy": 0.679989755153656, + "num_tokens": 122327571.0, + "step": 4784 + }, + { + "epoch": 0.5254777070063694, + "grad_norm": 2.4096665382385254, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7207897901535034, + "num_tokens": 122347298.0, + "step": 4785 + }, + { + "epoch": 0.5255875247089831, + "grad_norm": 2.300511598587036, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6936916708946228, + "num_tokens": 122370964.0, + "step": 4786 + }, + { + "epoch": 0.5256973424115967, + "grad_norm": 2.2803235054016113, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.6948539018630981, + "num_tokens": 122393459.0, + "step": 4787 + }, + { + "epoch": 0.5258071601142104, + "grad_norm": 2.3717591762542725, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7124252319335938, + "num_tokens": 122415380.0, + "step": 4788 + }, + { + "epoch": 0.5259169778168241, + "grad_norm": 2.2285218238830566, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7248390913009644, + "num_tokens": 122438090.0, + "step": 4789 + }, + { + "epoch": 0.5260267955194378, + "grad_norm": 2.374185562133789, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7032108902931213, + "num_tokens": 122459849.0, + "step": 4790 + }, + { + "epoch": 0.5261366132220514, + "grad_norm": 2.3239102363586426, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7057147026062012, + "num_tokens": 122485059.0, + "step": 4791 + }, + { + "epoch": 0.526246430924665, + "grad_norm": 2.3194830417633057, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.6971778273582458, + "num_tokens": 122510057.0, + "step": 4792 + }, + { + "epoch": 0.5263562486272787, + "grad_norm": 2.337334156036377, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7082421183586121, + "num_tokens": 122532450.0, + "step": 4793 + }, + { + "epoch": 0.5264660663298923, + "grad_norm": 2.160604953765869, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7003490328788757, + "num_tokens": 122558104.0, + "step": 4794 + }, + { + "epoch": 0.526575884032506, + "grad_norm": 2.0094211101531982, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.699836015701294, + "num_tokens": 122585769.0, + "step": 4795 + }, + { + "epoch": 0.5266857017351197, + "grad_norm": 2.178534984588623, + "learning_rate": 1e-06, + "loss": 1.0494, + "mean_token_accuracy": 0.6799812912940979, + "num_tokens": 122611595.0, + "step": 4796 + }, + { + "epoch": 0.5267955194377334, + "grad_norm": 2.181809663772583, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7160430550575256, + "num_tokens": 122635205.0, + "step": 4797 + }, + { + "epoch": 0.526905337140347, + "grad_norm": 2.266587972640991, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7194700837135315, + "num_tokens": 122657731.0, + "step": 4798 + }, + { + "epoch": 0.5270151548429607, + "grad_norm": 2.222749710083008, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6991128325462341, + "num_tokens": 122684736.0, + "step": 4799 + }, + { + "epoch": 0.5271249725455743, + "grad_norm": 2.422792434692383, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7001457214355469, + "num_tokens": 122707186.0, + "step": 4800 + }, + { + "epoch": 0.527234790248188, + "grad_norm": 2.1115775108337402, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6942151188850403, + "num_tokens": 122732299.0, + "step": 4801 + }, + { + "epoch": 0.5273446079508016, + "grad_norm": 2.3991827964782715, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.7046456336975098, + "num_tokens": 122755775.0, + "step": 4802 + }, + { + "epoch": 0.5274544256534154, + "grad_norm": 2.3043150901794434, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6953428983688354, + "num_tokens": 122779417.0, + "step": 4803 + }, + { + "epoch": 0.527564243356029, + "grad_norm": 2.117471218109131, + "learning_rate": 1e-06, + "loss": 1.0557, + "mean_token_accuracy": 0.6775684356689453, + "num_tokens": 122806394.0, + "step": 4804 + }, + { + "epoch": 0.5276740610586427, + "grad_norm": 2.157705545425415, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7167856097221375, + "num_tokens": 122832201.0, + "step": 4805 + }, + { + "epoch": 0.5277838787612563, + "grad_norm": 2.2547292709350586, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7184419631958008, + "num_tokens": 122856277.0, + "step": 4806 + }, + { + "epoch": 0.52789369646387, + "grad_norm": 2.1649134159088135, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7105457782745361, + "num_tokens": 122886387.0, + "step": 4807 + }, + { + "epoch": 0.5280035141664836, + "grad_norm": 2.378671407699585, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6995164155960083, + "num_tokens": 122908355.0, + "step": 4808 + }, + { + "epoch": 0.5281133318690973, + "grad_norm": 2.046009063720703, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6994907855987549, + "num_tokens": 122937529.0, + "step": 4809 + }, + { + "epoch": 0.528223149571711, + "grad_norm": 2.0948092937469482, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7082659006118774, + "num_tokens": 122965890.0, + "step": 4810 + }, + { + "epoch": 0.5283329672743247, + "grad_norm": 2.1568963527679443, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.6936834454536438, + "num_tokens": 122992997.0, + "step": 4811 + }, + { + "epoch": 0.5284427849769383, + "grad_norm": 2.271864891052246, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7157820463180542, + "num_tokens": 123018615.0, + "step": 4812 + }, + { + "epoch": 0.528552602679552, + "grad_norm": 2.1800081729888916, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7022608518600464, + "num_tokens": 123045809.0, + "step": 4813 + }, + { + "epoch": 0.5286624203821656, + "grad_norm": 2.4559431076049805, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.7077634334564209, + "num_tokens": 123066918.0, + "step": 4814 + }, + { + "epoch": 0.5287722380847792, + "grad_norm": 1.9758321046829224, + "learning_rate": 1e-06, + "loss": 1.0714, + "mean_token_accuracy": 0.6856175661087036, + "num_tokens": 123096480.0, + "step": 4815 + }, + { + "epoch": 0.5288820557873929, + "grad_norm": 2.1855967044830322, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6932423114776611, + "num_tokens": 123122940.0, + "step": 4816 + }, + { + "epoch": 0.5289918734900065, + "grad_norm": 2.2767114639282227, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7073870897293091, + "num_tokens": 123145184.0, + "step": 4817 + }, + { + "epoch": 0.5291016911926203, + "grad_norm": 1.8980870246887207, + "learning_rate": 1e-06, + "loss": 1.0719, + "mean_token_accuracy": 0.6767159700393677, + "num_tokens": 123177677.0, + "step": 4818 + }, + { + "epoch": 0.5292115088952339, + "grad_norm": 2.233658790588379, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6969042420387268, + "num_tokens": 123205231.0, + "step": 4819 + }, + { + "epoch": 0.5293213265978476, + "grad_norm": 2.2792820930480957, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7027915716171265, + "num_tokens": 123230923.0, + "step": 4820 + }, + { + "epoch": 0.5294311443004612, + "grad_norm": 2.16351056098938, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6933892965316772, + "num_tokens": 123258512.0, + "step": 4821 + }, + { + "epoch": 0.5295409620030749, + "grad_norm": 1.965206503868103, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6969385147094727, + "num_tokens": 123288630.0, + "step": 4822 + }, + { + "epoch": 0.5296507797056885, + "grad_norm": 2.1032214164733887, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.7017823457717896, + "num_tokens": 123317084.0, + "step": 4823 + }, + { + "epoch": 0.5297605974083022, + "grad_norm": 2.145024061203003, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7140772342681885, + "num_tokens": 123343572.0, + "step": 4824 + }, + { + "epoch": 0.5298704151109159, + "grad_norm": 2.2341766357421875, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7050759792327881, + "num_tokens": 123369963.0, + "step": 4825 + }, + { + "epoch": 0.5299802328135296, + "grad_norm": 2.246811628341675, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7127566337585449, + "num_tokens": 123393934.0, + "step": 4826 + }, + { + "epoch": 0.5300900505161432, + "grad_norm": 2.38212513923645, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7165277004241943, + "num_tokens": 123417409.0, + "step": 4827 + }, + { + "epoch": 0.5301998682187569, + "grad_norm": 2.1571848392486572, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.7026364803314209, + "num_tokens": 123442446.0, + "step": 4828 + }, + { + "epoch": 0.5303096859213705, + "grad_norm": 2.2804811000823975, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7186384201049805, + "num_tokens": 123464266.0, + "step": 4829 + }, + { + "epoch": 0.5304195036239842, + "grad_norm": 2.081954002380371, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.7007966637611389, + "num_tokens": 123494886.0, + "step": 4830 + }, + { + "epoch": 0.5305293213265978, + "grad_norm": 2.083552837371826, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7251969575881958, + "num_tokens": 123521837.0, + "step": 4831 + }, + { + "epoch": 0.5306391390292116, + "grad_norm": 2.2025325298309326, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6964056491851807, + "num_tokens": 123546720.0, + "step": 4832 + }, + { + "epoch": 0.5307489567318252, + "grad_norm": 2.600566864013672, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.6968353986740112, + "num_tokens": 123567344.0, + "step": 4833 + }, + { + "epoch": 0.5308587744344389, + "grad_norm": 2.3005900382995605, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7377450466156006, + "num_tokens": 123590065.0, + "step": 4834 + }, + { + "epoch": 0.5309685921370525, + "grad_norm": 2.367389678955078, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.6983557343482971, + "num_tokens": 123615828.0, + "step": 4835 + }, + { + "epoch": 0.5310784098396661, + "grad_norm": 2.3293144702911377, + "learning_rate": 1e-06, + "loss": 1.057, + "mean_token_accuracy": 0.693774938583374, + "num_tokens": 123639381.0, + "step": 4836 + }, + { + "epoch": 0.5311882275422798, + "grad_norm": 2.220400810241699, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7199063301086426, + "num_tokens": 123664242.0, + "step": 4837 + }, + { + "epoch": 0.5312980452448934, + "grad_norm": 2.168668031692505, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7135464549064636, + "num_tokens": 123689868.0, + "step": 4838 + }, + { + "epoch": 0.5314078629475072, + "grad_norm": 2.2165467739105225, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6933979988098145, + "num_tokens": 123716015.0, + "step": 4839 + }, + { + "epoch": 0.5315176806501208, + "grad_norm": 2.385568618774414, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7141456604003906, + "num_tokens": 123736081.0, + "step": 4840 + }, + { + "epoch": 0.5316274983527345, + "grad_norm": 1.9928014278411865, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6824055910110474, + "num_tokens": 123766371.0, + "step": 4841 + }, + { + "epoch": 0.5317373160553481, + "grad_norm": 2.3101963996887207, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7018625736236572, + "num_tokens": 123789379.0, + "step": 4842 + }, + { + "epoch": 0.5318471337579618, + "grad_norm": 2.1287171840667725, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.72245854139328, + "num_tokens": 123813304.0, + "step": 4843 + }, + { + "epoch": 0.5319569514605754, + "grad_norm": 2.4477903842926025, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.7085827589035034, + "num_tokens": 123833392.0, + "step": 4844 + }, + { + "epoch": 0.5320667691631891, + "grad_norm": 2.255563735961914, + "learning_rate": 1e-06, + "loss": 1.0519, + "mean_token_accuracy": 0.6787552833557129, + "num_tokens": 123859031.0, + "step": 4845 + }, + { + "epoch": 0.5321765868658027, + "grad_norm": 2.2905945777893066, + "learning_rate": 1e-06, + "loss": 1.0274, + "mean_token_accuracy": 0.6939120292663574, + "num_tokens": 123882552.0, + "step": 4846 + }, + { + "epoch": 0.5322864045684165, + "grad_norm": 2.1854147911071777, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.7066181302070618, + "num_tokens": 123905690.0, + "step": 4847 + }, + { + "epoch": 0.5323962222710301, + "grad_norm": 1.9259510040283203, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.6993256211280823, + "num_tokens": 123935721.0, + "step": 4848 + }, + { + "epoch": 0.5325060399736438, + "grad_norm": 2.272749185562134, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7280349731445312, + "num_tokens": 123959674.0, + "step": 4849 + }, + { + "epoch": 0.5326158576762574, + "grad_norm": 2.209850311279297, + "learning_rate": 1e-06, + "loss": 1.0428, + "mean_token_accuracy": 0.6876154541969299, + "num_tokens": 123986043.0, + "step": 4850 + }, + { + "epoch": 0.532725675378871, + "grad_norm": 2.2241010665893555, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.7041618824005127, + "num_tokens": 124012205.0, + "step": 4851 + }, + { + "epoch": 0.5328354930814847, + "grad_norm": 2.300410032272339, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.7030772566795349, + "num_tokens": 124034046.0, + "step": 4852 + }, + { + "epoch": 0.5329453107840983, + "grad_norm": 2.153398275375366, + "learning_rate": 1e-06, + "loss": 1.036, + "mean_token_accuracy": 0.6817431449890137, + "num_tokens": 124063024.0, + "step": 4853 + }, + { + "epoch": 0.5330551284867121, + "grad_norm": 2.7011828422546387, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7029362916946411, + "num_tokens": 124082332.0, + "step": 4854 + }, + { + "epoch": 0.5331649461893258, + "grad_norm": 2.13541316986084, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6998415589332581, + "num_tokens": 124110987.0, + "step": 4855 + }, + { + "epoch": 0.5332747638919394, + "grad_norm": 2.051900625228882, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.705342710018158, + "num_tokens": 124137958.0, + "step": 4856 + }, + { + "epoch": 0.533384581594553, + "grad_norm": 2.097318649291992, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7084391117095947, + "num_tokens": 124164204.0, + "step": 4857 + }, + { + "epoch": 0.5334943992971667, + "grad_norm": 2.2514593601226807, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7205171585083008, + "num_tokens": 124189022.0, + "step": 4858 + }, + { + "epoch": 0.5336042169997803, + "grad_norm": 2.0000553131103516, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7030617594718933, + "num_tokens": 124216775.0, + "step": 4859 + }, + { + "epoch": 0.533714034702394, + "grad_norm": 2.044804811477661, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.6989279389381409, + "num_tokens": 124247504.0, + "step": 4860 + }, + { + "epoch": 0.5338238524050077, + "grad_norm": 2.042227029800415, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6971178650856018, + "num_tokens": 124276516.0, + "step": 4861 + }, + { + "epoch": 0.5339336701076214, + "grad_norm": 2.200413942337036, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7047666311264038, + "num_tokens": 124302095.0, + "step": 4862 + }, + { + "epoch": 0.534043487810235, + "grad_norm": 1.8656868934631348, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6964863538742065, + "num_tokens": 124335755.0, + "step": 4863 + }, + { + "epoch": 0.5341533055128487, + "grad_norm": 2.0395612716674805, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.7017542123794556, + "num_tokens": 124362763.0, + "step": 4864 + }, + { + "epoch": 0.5342631232154623, + "grad_norm": 2.1202757358551025, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.6969501972198486, + "num_tokens": 124390226.0, + "step": 4865 + }, + { + "epoch": 0.534372940918076, + "grad_norm": 2.5567994117736816, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7069768309593201, + "num_tokens": 124411119.0, + "step": 4866 + }, + { + "epoch": 0.5344827586206896, + "grad_norm": 2.474330186843872, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7055860757827759, + "num_tokens": 124434710.0, + "step": 4867 + }, + { + "epoch": 0.5345925763233034, + "grad_norm": 2.520834445953369, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.70844566822052, + "num_tokens": 124456073.0, + "step": 4868 + }, + { + "epoch": 0.534702394025917, + "grad_norm": 2.089240789413452, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7217286229133606, + "num_tokens": 124481372.0, + "step": 4869 + }, + { + "epoch": 0.5348122117285307, + "grad_norm": 2.314095973968506, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7001356482505798, + "num_tokens": 124505071.0, + "step": 4870 + }, + { + "epoch": 0.5349220294311443, + "grad_norm": 2.243215799331665, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6924477815628052, + "num_tokens": 124531739.0, + "step": 4871 + }, + { + "epoch": 0.535031847133758, + "grad_norm": 2.1999495029449463, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7209564447402954, + "num_tokens": 124553972.0, + "step": 4872 + }, + { + "epoch": 0.5351416648363716, + "grad_norm": 2.193706512451172, + "learning_rate": 1e-06, + "loss": 1.1052, + "mean_token_accuracy": 0.6750420331954956, + "num_tokens": 124584746.0, + "step": 4873 + }, + { + "epoch": 0.5352514825389852, + "grad_norm": 2.141707420349121, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7065019607543945, + "num_tokens": 124610006.0, + "step": 4874 + }, + { + "epoch": 0.5353613002415989, + "grad_norm": 2.733607530593872, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.7042729258537292, + "num_tokens": 124629449.0, + "step": 4875 + }, + { + "epoch": 0.5354711179442126, + "grad_norm": 2.7419371604919434, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7153407335281372, + "num_tokens": 124648230.0, + "step": 4876 + }, + { + "epoch": 0.5355809356468263, + "grad_norm": 2.330531120300293, + "learning_rate": 1e-06, + "loss": 1.043, + "mean_token_accuracy": 0.6839492321014404, + "num_tokens": 124671846.0, + "step": 4877 + }, + { + "epoch": 0.5356907533494399, + "grad_norm": 2.1389708518981934, + "learning_rate": 1e-06, + "loss": 1.0898, + "mean_token_accuracy": 0.674129843711853, + "num_tokens": 124704019.0, + "step": 4878 + }, + { + "epoch": 0.5358005710520536, + "grad_norm": 2.0471272468566895, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6959229707717896, + "num_tokens": 124731867.0, + "step": 4879 + }, + { + "epoch": 0.5359103887546672, + "grad_norm": 2.226834774017334, + "learning_rate": 1e-06, + "loss": 1.079, + "mean_token_accuracy": 0.6822097301483154, + "num_tokens": 124757203.0, + "step": 4880 + }, + { + "epoch": 0.5360202064572809, + "grad_norm": 2.061704397201538, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7085539698600769, + "num_tokens": 124783386.0, + "step": 4881 + }, + { + "epoch": 0.5361300241598945, + "grad_norm": 2.394960403442383, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7046625018119812, + "num_tokens": 124804471.0, + "step": 4882 + }, + { + "epoch": 0.5362398418625083, + "grad_norm": 2.3311493396759033, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7273901700973511, + "num_tokens": 124827154.0, + "step": 4883 + }, + { + "epoch": 0.5363496595651219, + "grad_norm": 2.408267021179199, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7224084138870239, + "num_tokens": 124847875.0, + "step": 4884 + }, + { + "epoch": 0.5364594772677356, + "grad_norm": 2.389982223510742, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7063241004943848, + "num_tokens": 124870793.0, + "step": 4885 + }, + { + "epoch": 0.5365692949703492, + "grad_norm": 2.134075164794922, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7083788514137268, + "num_tokens": 124896541.0, + "step": 4886 + }, + { + "epoch": 0.5366791126729629, + "grad_norm": 2.329516649246216, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.7041539549827576, + "num_tokens": 124920341.0, + "step": 4887 + }, + { + "epoch": 0.5367889303755765, + "grad_norm": 2.0356674194335938, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7096638083457947, + "num_tokens": 124951340.0, + "step": 4888 + }, + { + "epoch": 0.5368987480781902, + "grad_norm": 2.374521017074585, + "learning_rate": 1e-06, + "loss": 1.0586, + "mean_token_accuracy": 0.6822842359542847, + "num_tokens": 124976815.0, + "step": 4889 + }, + { + "epoch": 0.5370085657808039, + "grad_norm": 1.9404447078704834, + "learning_rate": 1e-06, + "loss": 1.0246, + "mean_token_accuracy": 0.6891400218009949, + "num_tokens": 125009262.0, + "step": 4890 + }, + { + "epoch": 0.5371183834834176, + "grad_norm": 2.4240689277648926, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7020151615142822, + "num_tokens": 125032511.0, + "step": 4891 + }, + { + "epoch": 0.5372282011860312, + "grad_norm": 2.394501209259033, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7168835401535034, + "num_tokens": 125055990.0, + "step": 4892 + }, + { + "epoch": 0.5373380188886449, + "grad_norm": 1.9075512886047363, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.6859368085861206, + "num_tokens": 125089102.0, + "step": 4893 + }, + { + "epoch": 0.5374478365912585, + "grad_norm": 2.258430004119873, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7130646705627441, + "num_tokens": 125112462.0, + "step": 4894 + }, + { + "epoch": 0.5375576542938721, + "grad_norm": 2.1391210556030273, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6909558773040771, + "num_tokens": 125139465.0, + "step": 4895 + }, + { + "epoch": 0.5376674719964858, + "grad_norm": 2.322753429412842, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.714011549949646, + "num_tokens": 125160968.0, + "step": 4896 + }, + { + "epoch": 0.5377772896990995, + "grad_norm": 2.203061580657959, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.6984522342681885, + "num_tokens": 125190124.0, + "step": 4897 + }, + { + "epoch": 0.5378871074017132, + "grad_norm": 2.295290946960449, + "learning_rate": 1e-06, + "loss": 1.0216, + "mean_token_accuracy": 0.6910808086395264, + "num_tokens": 125215130.0, + "step": 4898 + }, + { + "epoch": 0.5379969251043268, + "grad_norm": 2.4800174236297607, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.7033231258392334, + "num_tokens": 125236835.0, + "step": 4899 + }, + { + "epoch": 0.5381067428069405, + "grad_norm": 2.271918773651123, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.707573413848877, + "num_tokens": 125261657.0, + "step": 4900 + }, + { + "epoch": 0.5382165605095541, + "grad_norm": 2.272127866744995, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.6924920082092285, + "num_tokens": 125288211.0, + "step": 4901 + }, + { + "epoch": 0.5383263782121678, + "grad_norm": 2.1596336364746094, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.6907957196235657, + "num_tokens": 125313489.0, + "step": 4902 + }, + { + "epoch": 0.5384361959147814, + "grad_norm": 2.0306575298309326, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.6932132840156555, + "num_tokens": 125344306.0, + "step": 4903 + }, + { + "epoch": 0.5385460136173951, + "grad_norm": 2.132660388946533, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7025933861732483, + "num_tokens": 125373411.0, + "step": 4904 + }, + { + "epoch": 0.5386558313200088, + "grad_norm": 2.3042304515838623, + "learning_rate": 1e-06, + "loss": 1.0818, + "mean_token_accuracy": 0.6790891885757446, + "num_tokens": 125398340.0, + "step": 4905 + }, + { + "epoch": 0.5387656490226225, + "grad_norm": 2.437382698059082, + "learning_rate": 1e-06, + "loss": 1.0918, + "mean_token_accuracy": 0.6785780191421509, + "num_tokens": 125421099.0, + "step": 4906 + }, + { + "epoch": 0.5388754667252361, + "grad_norm": 2.3758533000946045, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7227756977081299, + "num_tokens": 125442555.0, + "step": 4907 + }, + { + "epoch": 0.5389852844278498, + "grad_norm": 2.1767046451568604, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7250546216964722, + "num_tokens": 125466853.0, + "step": 4908 + }, + { + "epoch": 0.5390951021304634, + "grad_norm": 1.9610942602157593, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7119970321655273, + "num_tokens": 125496612.0, + "step": 4909 + }, + { + "epoch": 0.5392049198330771, + "grad_norm": 2.224769115447998, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7196975946426392, + "num_tokens": 125523594.0, + "step": 4910 + }, + { + "epoch": 0.5393147375356907, + "grad_norm": 2.2446095943450928, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.7065041065216064, + "num_tokens": 125546703.0, + "step": 4911 + }, + { + "epoch": 0.5394245552383045, + "grad_norm": 2.4670214653015137, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6929704546928406, + "num_tokens": 125568286.0, + "step": 4912 + }, + { + "epoch": 0.5395343729409181, + "grad_norm": 2.2090747356414795, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.6977588534355164, + "num_tokens": 125593991.0, + "step": 4913 + }, + { + "epoch": 0.5396441906435318, + "grad_norm": 2.173434019088745, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.700381875038147, + "num_tokens": 125620538.0, + "step": 4914 + }, + { + "epoch": 0.5397540083461454, + "grad_norm": 2.0815529823303223, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7054632902145386, + "num_tokens": 125646598.0, + "step": 4915 + }, + { + "epoch": 0.539863826048759, + "grad_norm": 2.0972847938537598, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7138149738311768, + "num_tokens": 125674204.0, + "step": 4916 + }, + { + "epoch": 0.5399736437513727, + "grad_norm": 2.143768072128296, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6911468505859375, + "num_tokens": 125700336.0, + "step": 4917 + }, + { + "epoch": 0.5400834614539863, + "grad_norm": 2.2991943359375, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7117174863815308, + "num_tokens": 125722887.0, + "step": 4918 + }, + { + "epoch": 0.5401932791566001, + "grad_norm": 2.4063682556152344, + "learning_rate": 1e-06, + "loss": 1.0367, + "mean_token_accuracy": 0.6778695583343506, + "num_tokens": 125745722.0, + "step": 4919 + }, + { + "epoch": 0.5403030968592137, + "grad_norm": 2.626767635345459, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7036802768707275, + "num_tokens": 125765011.0, + "step": 4920 + }, + { + "epoch": 0.5404129145618274, + "grad_norm": 2.0005340576171875, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7035508751869202, + "num_tokens": 125795344.0, + "step": 4921 + }, + { + "epoch": 0.540522732264441, + "grad_norm": 1.972278118133545, + "learning_rate": 1e-06, + "loss": 1.0306, + "mean_token_accuracy": 0.6873660087585449, + "num_tokens": 125825144.0, + "step": 4922 + }, + { + "epoch": 0.5406325499670547, + "grad_norm": 2.307785749435425, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.6941269040107727, + "num_tokens": 125850159.0, + "step": 4923 + }, + { + "epoch": 0.5407423676696683, + "grad_norm": 2.5717954635620117, + "learning_rate": 1e-06, + "loss": 1.0114, + "mean_token_accuracy": 0.6976640820503235, + "num_tokens": 125870039.0, + "step": 4924 + }, + { + "epoch": 0.540852185372282, + "grad_norm": 2.0869431495666504, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.6853218078613281, + "num_tokens": 125900239.0, + "step": 4925 + }, + { + "epoch": 0.5409620030748957, + "grad_norm": 2.2788355350494385, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7228065729141235, + "num_tokens": 125923141.0, + "step": 4926 + }, + { + "epoch": 0.5410718207775094, + "grad_norm": 2.1409568786621094, + "learning_rate": 1e-06, + "loss": 1.0771, + "mean_token_accuracy": 0.6813281774520874, + "num_tokens": 125949284.0, + "step": 4927 + }, + { + "epoch": 0.541181638480123, + "grad_norm": 2.144275188446045, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7229015827178955, + "num_tokens": 125974994.0, + "step": 4928 + }, + { + "epoch": 0.5412914561827367, + "grad_norm": 2.2798590660095215, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.6955941319465637, + "num_tokens": 125999394.0, + "step": 4929 + }, + { + "epoch": 0.5414012738853503, + "grad_norm": 2.596773147583008, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7107041478157043, + "num_tokens": 126018623.0, + "step": 4930 + }, + { + "epoch": 0.541511091587964, + "grad_norm": 2.311450242996216, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7179348468780518, + "num_tokens": 126040369.0, + "step": 4931 + }, + { + "epoch": 0.5416209092905776, + "grad_norm": 2.2678778171539307, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6983484029769897, + "num_tokens": 126064294.0, + "step": 4932 + }, + { + "epoch": 0.5417307269931912, + "grad_norm": 2.075213670730591, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7055460810661316, + "num_tokens": 126091099.0, + "step": 4933 + }, + { + "epoch": 0.541840544695805, + "grad_norm": 2.14882230758667, + "learning_rate": 1e-06, + "loss": 1.042, + "mean_token_accuracy": 0.6892826557159424, + "num_tokens": 126119275.0, + "step": 4934 + }, + { + "epoch": 0.5419503623984187, + "grad_norm": 2.505627393722534, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7131417393684387, + "num_tokens": 126139101.0, + "step": 4935 + }, + { + "epoch": 0.5420601801010323, + "grad_norm": 2.06954288482666, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7279622554779053, + "num_tokens": 126166893.0, + "step": 4936 + }, + { + "epoch": 0.5421699978036459, + "grad_norm": 2.250967025756836, + "learning_rate": 1e-06, + "loss": 1.0945, + "mean_token_accuracy": 0.675764262676239, + "num_tokens": 126191009.0, + "step": 4937 + }, + { + "epoch": 0.5422798155062596, + "grad_norm": 2.108222484588623, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6977635025978088, + "num_tokens": 126220129.0, + "step": 4938 + }, + { + "epoch": 0.5423896332088732, + "grad_norm": 2.1806070804595947, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.7008967399597168, + "num_tokens": 126246945.0, + "step": 4939 + }, + { + "epoch": 0.5424994509114869, + "grad_norm": 2.21008038520813, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.704007625579834, + "num_tokens": 126272295.0, + "step": 4940 + }, + { + "epoch": 0.5426092686141006, + "grad_norm": 2.3895061016082764, + "learning_rate": 1e-06, + "loss": 1.0546, + "mean_token_accuracy": 0.6830296516418457, + "num_tokens": 126296773.0, + "step": 4941 + }, + { + "epoch": 0.5427190863167143, + "grad_norm": 2.2433393001556396, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7260779142379761, + "num_tokens": 126320260.0, + "step": 4942 + }, + { + "epoch": 0.5428289040193279, + "grad_norm": 2.072019338607788, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7094855308532715, + "num_tokens": 126347046.0, + "step": 4943 + }, + { + "epoch": 0.5429387217219416, + "grad_norm": 2.2562098503112793, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7194509506225586, + "num_tokens": 126370730.0, + "step": 4944 + }, + { + "epoch": 0.5430485394245552, + "grad_norm": 2.2327158451080322, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7163723111152649, + "num_tokens": 126395207.0, + "step": 4945 + }, + { + "epoch": 0.5431583571271689, + "grad_norm": 1.9704883098602295, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.6958290338516235, + "num_tokens": 126426664.0, + "step": 4946 + }, + { + "epoch": 0.5432681748297825, + "grad_norm": 2.3429818153381348, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7043951749801636, + "num_tokens": 126449306.0, + "step": 4947 + }, + { + "epoch": 0.5433779925323963, + "grad_norm": 2.2852272987365723, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7080062031745911, + "num_tokens": 126472294.0, + "step": 4948 + }, + { + "epoch": 0.5434878102350099, + "grad_norm": 2.2056398391723633, + "learning_rate": 1e-06, + "loss": 1.0342, + "mean_token_accuracy": 0.6900293231010437, + "num_tokens": 126497716.0, + "step": 4949 + }, + { + "epoch": 0.5435976279376236, + "grad_norm": 2.05617094039917, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.702602744102478, + "num_tokens": 126526325.0, + "step": 4950 + }, + { + "epoch": 0.5437074456402372, + "grad_norm": 2.28135347366333, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7005981206893921, + "num_tokens": 126550192.0, + "step": 4951 + }, + { + "epoch": 0.5438172633428509, + "grad_norm": 2.540968418121338, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7033286094665527, + "num_tokens": 126570541.0, + "step": 4952 + }, + { + "epoch": 0.5439270810454645, + "grad_norm": 2.115469455718994, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7111804485321045, + "num_tokens": 126598008.0, + "step": 4953 + }, + { + "epoch": 0.5440368987480781, + "grad_norm": 2.0003883838653564, + "learning_rate": 1e-06, + "loss": 1.068, + "mean_token_accuracy": 0.6758391857147217, + "num_tokens": 126629533.0, + "step": 4954 + }, + { + "epoch": 0.5441467164506919, + "grad_norm": 1.9445722103118896, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.7011063098907471, + "num_tokens": 126659673.0, + "step": 4955 + }, + { + "epoch": 0.5442565341533055, + "grad_norm": 2.2067904472351074, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7073913812637329, + "num_tokens": 126686075.0, + "step": 4956 + }, + { + "epoch": 0.5443663518559192, + "grad_norm": 2.379695177078247, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7170138955116272, + "num_tokens": 126707903.0, + "step": 4957 + }, + { + "epoch": 0.5444761695585328, + "grad_norm": 2.321446418762207, + "learning_rate": 1e-06, + "loss": 1.0417, + "mean_token_accuracy": 0.6939336061477661, + "num_tokens": 126731234.0, + "step": 4958 + }, + { + "epoch": 0.5445859872611465, + "grad_norm": 2.4819743633270264, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7233191728591919, + "num_tokens": 126751244.0, + "step": 4959 + }, + { + "epoch": 0.5446958049637601, + "grad_norm": 2.235586166381836, + "learning_rate": 1e-06, + "loss": 1.0959, + "mean_token_accuracy": 0.6726274490356445, + "num_tokens": 126777993.0, + "step": 4960 + }, + { + "epoch": 0.5448056226663738, + "grad_norm": 2.217158794403076, + "learning_rate": 1e-06, + "loss": 1.06, + "mean_token_accuracy": 0.6817270517349243, + "num_tokens": 126803177.0, + "step": 4961 + }, + { + "epoch": 0.5449154403689875, + "grad_norm": 2.543915271759033, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7156170606613159, + "num_tokens": 126823654.0, + "step": 4962 + }, + { + "epoch": 0.5450252580716012, + "grad_norm": 2.3427414894104004, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7119646072387695, + "num_tokens": 126846619.0, + "step": 4963 + }, + { + "epoch": 0.5451350757742148, + "grad_norm": 2.2779979705810547, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7003765106201172, + "num_tokens": 126871001.0, + "step": 4964 + }, + { + "epoch": 0.5452448934768285, + "grad_norm": 2.5637197494506836, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7234165668487549, + "num_tokens": 126892042.0, + "step": 4965 + }, + { + "epoch": 0.5453547111794421, + "grad_norm": 2.2388198375701904, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.702617883682251, + "num_tokens": 126916766.0, + "step": 4966 + }, + { + "epoch": 0.5454645288820558, + "grad_norm": 2.039740800857544, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.7055133581161499, + "num_tokens": 126947866.0, + "step": 4967 + }, + { + "epoch": 0.5455743465846694, + "grad_norm": 2.45759916305542, + "learning_rate": 1e-06, + "loss": 1.0683, + "mean_token_accuracy": 0.6783312559127808, + "num_tokens": 126969749.0, + "step": 4968 + }, + { + "epoch": 0.5456841642872831, + "grad_norm": 2.086665153503418, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.709577739238739, + "num_tokens": 126998643.0, + "step": 4969 + }, + { + "epoch": 0.5457939819898968, + "grad_norm": 2.1892707347869873, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7131708860397339, + "num_tokens": 127024138.0, + "step": 4970 + }, + { + "epoch": 0.5459037996925105, + "grad_norm": 2.616119384765625, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7220802307128906, + "num_tokens": 127043889.0, + "step": 4971 + }, + { + "epoch": 0.5460136173951241, + "grad_norm": 2.047884702682495, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7046473026275635, + "num_tokens": 127075036.0, + "step": 4972 + }, + { + "epoch": 0.5461234350977378, + "grad_norm": 2.4180636405944824, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7014014720916748, + "num_tokens": 127097421.0, + "step": 4973 + }, + { + "epoch": 0.5462332528003514, + "grad_norm": 2.1417484283447266, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7019299268722534, + "num_tokens": 127124085.0, + "step": 4974 + }, + { + "epoch": 0.546343070502965, + "grad_norm": 2.1212170124053955, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7070519924163818, + "num_tokens": 127149534.0, + "step": 4975 + }, + { + "epoch": 0.5464528882055787, + "grad_norm": 2.113809823989868, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.701148271560669, + "num_tokens": 127177192.0, + "step": 4976 + }, + { + "epoch": 0.5465627059081924, + "grad_norm": 2.212568521499634, + "learning_rate": 1e-06, + "loss": 1.0256, + "mean_token_accuracy": 0.6882979869842529, + "num_tokens": 127203172.0, + "step": 4977 + }, + { + "epoch": 0.5466725236108061, + "grad_norm": 2.0576767921447754, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6909072995185852, + "num_tokens": 127232240.0, + "step": 4978 + }, + { + "epoch": 0.5467823413134197, + "grad_norm": 2.1064140796661377, + "learning_rate": 1e-06, + "loss": 1.0334, + "mean_token_accuracy": 0.6870710849761963, + "num_tokens": 127259417.0, + "step": 4979 + }, + { + "epoch": 0.5468921590160334, + "grad_norm": 2.0763044357299805, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7171306610107422, + "num_tokens": 127283842.0, + "step": 4980 + }, + { + "epoch": 0.547001976718647, + "grad_norm": 2.4751479625701904, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6942786574363708, + "num_tokens": 127305139.0, + "step": 4981 + }, + { + "epoch": 0.5471117944212607, + "grad_norm": 2.383143424987793, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7187002897262573, + "num_tokens": 127326541.0, + "step": 4982 + }, + { + "epoch": 0.5472216121238743, + "grad_norm": 2.600264310836792, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.696457028388977, + "num_tokens": 127344611.0, + "step": 4983 + }, + { + "epoch": 0.5473314298264881, + "grad_norm": 2.077094316482544, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7195895910263062, + "num_tokens": 127372399.0, + "step": 4984 + }, + { + "epoch": 0.5474412475291017, + "grad_norm": 2.5745913982391357, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7104830145835876, + "num_tokens": 127392950.0, + "step": 4985 + }, + { + "epoch": 0.5475510652317154, + "grad_norm": 2.126347303390503, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7251067161560059, + "num_tokens": 127419198.0, + "step": 4986 + }, + { + "epoch": 0.547660882934329, + "grad_norm": 2.4108293056488037, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7109529376029968, + "num_tokens": 127442227.0, + "step": 4987 + }, + { + "epoch": 0.5477707006369427, + "grad_norm": 2.1271378993988037, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.6925954818725586, + "num_tokens": 127469634.0, + "step": 4988 + }, + { + "epoch": 0.5478805183395563, + "grad_norm": 2.1177868843078613, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7139958143234253, + "num_tokens": 127496580.0, + "step": 4989 + }, + { + "epoch": 0.54799033604217, + "grad_norm": 2.3584794998168945, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7185846567153931, + "num_tokens": 127522478.0, + "step": 4990 + }, + { + "epoch": 0.5481001537447837, + "grad_norm": 2.52536940574646, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.7007033228874207, + "num_tokens": 127542754.0, + "step": 4991 + }, + { + "epoch": 0.5482099714473974, + "grad_norm": 2.313016176223755, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7120730876922607, + "num_tokens": 127566654.0, + "step": 4992 + }, + { + "epoch": 0.548319789150011, + "grad_norm": 2.090106725692749, + "learning_rate": 1e-06, + "loss": 1.1005, + "mean_token_accuracy": 0.6721562743186951, + "num_tokens": 127595688.0, + "step": 4993 + }, + { + "epoch": 0.5484296068526247, + "grad_norm": 2.223167896270752, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7229461073875427, + "num_tokens": 127619445.0, + "step": 4994 + }, + { + "epoch": 0.5485394245552383, + "grad_norm": 1.8583042621612549, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7065572142601013, + "num_tokens": 127651153.0, + "step": 4995 + }, + { + "epoch": 0.5486492422578519, + "grad_norm": 2.098271608352661, + "learning_rate": 1e-06, + "loss": 1.1006, + "mean_token_accuracy": 0.6684770584106445, + "num_tokens": 127680810.0, + "step": 4996 + }, + { + "epoch": 0.5487590599604656, + "grad_norm": 2.230088949203491, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7113150358200073, + "num_tokens": 127705391.0, + "step": 4997 + }, + { + "epoch": 0.5488688776630792, + "grad_norm": 2.366262912750244, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7124192714691162, + "num_tokens": 127729524.0, + "step": 4998 + }, + { + "epoch": 0.548978695365693, + "grad_norm": 2.065133571624756, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6924227476119995, + "num_tokens": 127756776.0, + "step": 4999 + }, + { + "epoch": 0.5490885130683066, + "grad_norm": 2.6135966777801514, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6937659382820129, + "num_tokens": 127775733.0, + "step": 5000 + }, + { + "epoch": 0.5491983307709203, + "grad_norm": 2.0701346397399902, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7137686610221863, + "num_tokens": 127803903.0, + "step": 5001 + }, + { + "epoch": 0.5493081484735339, + "grad_norm": 2.1410553455352783, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7059627771377563, + "num_tokens": 127830601.0, + "step": 5002 + }, + { + "epoch": 0.5494179661761476, + "grad_norm": 2.291360855102539, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7257617712020874, + "num_tokens": 127852221.0, + "step": 5003 + }, + { + "epoch": 0.5495277838787612, + "grad_norm": 2.051663875579834, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7247503399848938, + "num_tokens": 127881392.0, + "step": 5004 + }, + { + "epoch": 0.5496376015813749, + "grad_norm": 2.2076122760772705, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7298213839530945, + "num_tokens": 127905730.0, + "step": 5005 + }, + { + "epoch": 0.5497474192839886, + "grad_norm": 2.134838342666626, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7193167209625244, + "num_tokens": 127930169.0, + "step": 5006 + }, + { + "epoch": 0.5498572369866023, + "grad_norm": 2.225151300430298, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7132701277732849, + "num_tokens": 127954728.0, + "step": 5007 + }, + { + "epoch": 0.5499670546892159, + "grad_norm": 2.378338575363159, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6998708248138428, + "num_tokens": 127978320.0, + "step": 5008 + }, + { + "epoch": 0.5500768723918296, + "grad_norm": 2.2789947986602783, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7085413932800293, + "num_tokens": 128001136.0, + "step": 5009 + }, + { + "epoch": 0.5501866900944432, + "grad_norm": 2.466393232345581, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.708283007144928, + "num_tokens": 128021236.0, + "step": 5010 + }, + { + "epoch": 0.5502965077970569, + "grad_norm": 2.324777126312256, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7246807217597961, + "num_tokens": 128041790.0, + "step": 5011 + }, + { + "epoch": 0.5504063254996705, + "grad_norm": 2.1339476108551025, + "learning_rate": 1e-06, + "loss": 1.0629, + "mean_token_accuracy": 0.690254271030426, + "num_tokens": 128070541.0, + "step": 5012 + }, + { + "epoch": 0.5505161432022843, + "grad_norm": 2.1875991821289062, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.6897799968719482, + "num_tokens": 128095181.0, + "step": 5013 + }, + { + "epoch": 0.5506259609048979, + "grad_norm": 2.1247212886810303, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7339621782302856, + "num_tokens": 128120225.0, + "step": 5014 + }, + { + "epoch": 0.5507357786075116, + "grad_norm": 2.4193150997161865, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7193289995193481, + "num_tokens": 128145416.0, + "step": 5015 + }, + { + "epoch": 0.5508455963101252, + "grad_norm": 2.1359355449676514, + "learning_rate": 1e-06, + "loss": 1.0568, + "mean_token_accuracy": 0.6839460730552673, + "num_tokens": 128171898.0, + "step": 5016 + }, + { + "epoch": 0.5509554140127388, + "grad_norm": 2.3928568363189697, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.714542806148529, + "num_tokens": 128193674.0, + "step": 5017 + }, + { + "epoch": 0.5510652317153525, + "grad_norm": 2.383450984954834, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7124435901641846, + "num_tokens": 128216681.0, + "step": 5018 + }, + { + "epoch": 0.5511750494179661, + "grad_norm": 2.0813252925872803, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7042949199676514, + "num_tokens": 128243685.0, + "step": 5019 + }, + { + "epoch": 0.5512848671205799, + "grad_norm": 2.288857936859131, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7226181626319885, + "num_tokens": 128269284.0, + "step": 5020 + }, + { + "epoch": 0.5513946848231935, + "grad_norm": 2.131502628326416, + "learning_rate": 1e-06, + "loss": 1.1035, + "mean_token_accuracy": 0.6785382032394409, + "num_tokens": 128298649.0, + "step": 5021 + }, + { + "epoch": 0.5515045025258072, + "grad_norm": 2.3139750957489014, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7239640951156616, + "num_tokens": 128320306.0, + "step": 5022 + }, + { + "epoch": 0.5516143202284208, + "grad_norm": 2.525084972381592, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.744102954864502, + "num_tokens": 128339402.0, + "step": 5023 + }, + { + "epoch": 0.5517241379310345, + "grad_norm": 2.0133779048919678, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6928141713142395, + "num_tokens": 128369268.0, + "step": 5024 + }, + { + "epoch": 0.5518339556336481, + "grad_norm": 2.1066622734069824, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.6969890594482422, + "num_tokens": 128396845.0, + "step": 5025 + }, + { + "epoch": 0.5519437733362618, + "grad_norm": 2.397265672683716, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7125158309936523, + "num_tokens": 128419178.0, + "step": 5026 + }, + { + "epoch": 0.5520535910388754, + "grad_norm": 2.1297054290771484, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7141693830490112, + "num_tokens": 128446287.0, + "step": 5027 + }, + { + "epoch": 0.5521634087414892, + "grad_norm": 2.3746438026428223, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.711887001991272, + "num_tokens": 128470998.0, + "step": 5028 + }, + { + "epoch": 0.5522732264441028, + "grad_norm": 2.6397552490234375, + "learning_rate": 1e-06, + "loss": 1.0518, + "mean_token_accuracy": 0.6771963834762573, + "num_tokens": 128491180.0, + "step": 5029 + }, + { + "epoch": 0.5523830441467165, + "grad_norm": 2.398550271987915, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7005220651626587, + "num_tokens": 128512845.0, + "step": 5030 + }, + { + "epoch": 0.5524928618493301, + "grad_norm": 2.4128530025482178, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6947791576385498, + "num_tokens": 128536178.0, + "step": 5031 + }, + { + "epoch": 0.5526026795519438, + "grad_norm": 1.983520746231079, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7108430862426758, + "num_tokens": 128565394.0, + "step": 5032 + }, + { + "epoch": 0.5527124972545574, + "grad_norm": 2.145681381225586, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.6968322992324829, + "num_tokens": 128593522.0, + "step": 5033 + }, + { + "epoch": 0.552822314957171, + "grad_norm": 2.043837308883667, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7060278058052063, + "num_tokens": 128620307.0, + "step": 5034 + }, + { + "epoch": 0.5529321326597848, + "grad_norm": 2.364976644515991, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7197315692901611, + "num_tokens": 128641134.0, + "step": 5035 + }, + { + "epoch": 0.5530419503623984, + "grad_norm": 2.1090850830078125, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7091637849807739, + "num_tokens": 128669753.0, + "step": 5036 + }, + { + "epoch": 0.5531517680650121, + "grad_norm": 2.5553743839263916, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7294429540634155, + "num_tokens": 128688005.0, + "step": 5037 + }, + { + "epoch": 0.5532615857676257, + "grad_norm": 2.0753977298736572, + "learning_rate": 1e-06, + "loss": 1.0465, + "mean_token_accuracy": 0.6888086795806885, + "num_tokens": 128717178.0, + "step": 5038 + }, + { + "epoch": 0.5533714034702394, + "grad_norm": 2.079237699508667, + "learning_rate": 1e-06, + "loss": 1.023, + "mean_token_accuracy": 0.6985759139060974, + "num_tokens": 128746657.0, + "step": 5039 + }, + { + "epoch": 0.553481221172853, + "grad_norm": 2.105654239654541, + "learning_rate": 1e-06, + "loss": 1.1221, + "mean_token_accuracy": 0.6674110889434814, + "num_tokens": 128775720.0, + "step": 5040 + }, + { + "epoch": 0.5535910388754667, + "grad_norm": 2.058506965637207, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.6923636794090271, + "num_tokens": 128806040.0, + "step": 5041 + }, + { + "epoch": 0.5537008565780804, + "grad_norm": 2.4375882148742676, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7165404558181763, + "num_tokens": 128827821.0, + "step": 5042 + }, + { + "epoch": 0.5538106742806941, + "grad_norm": 2.1838157176971436, + "learning_rate": 1e-06, + "loss": 1.0532, + "mean_token_accuracy": 0.6786552667617798, + "num_tokens": 128853437.0, + "step": 5043 + }, + { + "epoch": 0.5539204919833077, + "grad_norm": 2.176583766937256, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7057909369468689, + "num_tokens": 128879252.0, + "step": 5044 + }, + { + "epoch": 0.5540303096859214, + "grad_norm": 1.9673748016357422, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7036792039871216, + "num_tokens": 128909475.0, + "step": 5045 + }, + { + "epoch": 0.554140127388535, + "grad_norm": 2.328765630722046, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7099699378013611, + "num_tokens": 128933012.0, + "step": 5046 + }, + { + "epoch": 0.5542499450911487, + "grad_norm": 2.212756872177124, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7094264030456543, + "num_tokens": 128960771.0, + "step": 5047 + }, + { + "epoch": 0.5543597627937623, + "grad_norm": 2.1442179679870605, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7148783206939697, + "num_tokens": 128987100.0, + "step": 5048 + }, + { + "epoch": 0.5544695804963761, + "grad_norm": 2.20528507232666, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7205865979194641, + "num_tokens": 129011051.0, + "step": 5049 + }, + { + "epoch": 0.5545793981989897, + "grad_norm": 2.4469151496887207, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.697687029838562, + "num_tokens": 129033403.0, + "step": 5050 + }, + { + "epoch": 0.5546892159016034, + "grad_norm": 2.6684000492095947, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7085754871368408, + "num_tokens": 129051037.0, + "step": 5051 + }, + { + "epoch": 0.554799033604217, + "grad_norm": 2.026993989944458, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7190945148468018, + "num_tokens": 129080855.0, + "step": 5052 + }, + { + "epoch": 0.5549088513068307, + "grad_norm": 2.1508290767669678, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6988476514816284, + "num_tokens": 129107544.0, + "step": 5053 + }, + { + "epoch": 0.5550186690094443, + "grad_norm": 2.155378818511963, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7097728252410889, + "num_tokens": 129133551.0, + "step": 5054 + }, + { + "epoch": 0.555128486712058, + "grad_norm": 2.346705436706543, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.6952642202377319, + "num_tokens": 129158661.0, + "step": 5055 + }, + { + "epoch": 0.5552383044146716, + "grad_norm": 2.1525495052337646, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.6974916458129883, + "num_tokens": 129185180.0, + "step": 5056 + }, + { + "epoch": 0.5553481221172853, + "grad_norm": 1.9889934062957764, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.6970330476760864, + "num_tokens": 129216908.0, + "step": 5057 + }, + { + "epoch": 0.555457939819899, + "grad_norm": 2.266425609588623, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6997499465942383, + "num_tokens": 129240645.0, + "step": 5058 + }, + { + "epoch": 0.5555677575225126, + "grad_norm": 2.5283515453338623, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6896694898605347, + "num_tokens": 129263269.0, + "step": 5059 + }, + { + "epoch": 0.5556775752251263, + "grad_norm": 2.1089870929718018, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6875525116920471, + "num_tokens": 129291202.0, + "step": 5060 + }, + { + "epoch": 0.5557873929277399, + "grad_norm": 2.3976969718933105, + "learning_rate": 1e-06, + "loss": 1.0521, + "mean_token_accuracy": 0.6923681497573853, + "num_tokens": 129313788.0, + "step": 5061 + }, + { + "epoch": 0.5558972106303536, + "grad_norm": 2.2221155166625977, + "learning_rate": 1e-06, + "loss": 1.0407, + "mean_token_accuracy": 0.6910138130187988, + "num_tokens": 129341482.0, + "step": 5062 + }, + { + "epoch": 0.5560070283329672, + "grad_norm": 2.2449207305908203, + "learning_rate": 1e-06, + "loss": 1.0355, + "mean_token_accuracy": 0.6832020878791809, + "num_tokens": 129368740.0, + "step": 5063 + }, + { + "epoch": 0.556116846035581, + "grad_norm": 2.304928779602051, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7039645910263062, + "num_tokens": 129393181.0, + "step": 5064 + }, + { + "epoch": 0.5562266637381946, + "grad_norm": 1.9899821281433105, + "learning_rate": 1e-06, + "loss": 1.0557, + "mean_token_accuracy": 0.6879861950874329, + "num_tokens": 129425575.0, + "step": 5065 + }, + { + "epoch": 0.5563364814408083, + "grad_norm": 2.38919997215271, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6958308815956116, + "num_tokens": 129448928.0, + "step": 5066 + }, + { + "epoch": 0.5564462991434219, + "grad_norm": 2.3103322982788086, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7184603214263916, + "num_tokens": 129473237.0, + "step": 5067 + }, + { + "epoch": 0.5565561168460356, + "grad_norm": 2.1315996646881104, + "learning_rate": 1e-06, + "loss": 1.0872, + "mean_token_accuracy": 0.680729866027832, + "num_tokens": 129504499.0, + "step": 5068 + }, + { + "epoch": 0.5566659345486492, + "grad_norm": 2.238438606262207, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.6936643123626709, + "num_tokens": 129527623.0, + "step": 5069 + }, + { + "epoch": 0.5567757522512629, + "grad_norm": 2.162125587463379, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.697528064250946, + "num_tokens": 129553694.0, + "step": 5070 + }, + { + "epoch": 0.5568855699538766, + "grad_norm": 2.502309560775757, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.734368085861206, + "num_tokens": 129572232.0, + "step": 5071 + }, + { + "epoch": 0.5569953876564903, + "grad_norm": 2.173901081085205, + "learning_rate": 1e-06, + "loss": 1.0459, + "mean_token_accuracy": 0.6847389340400696, + "num_tokens": 129596755.0, + "step": 5072 + }, + { + "epoch": 0.5571052053591039, + "grad_norm": 2.0959584712982178, + "learning_rate": 1e-06, + "loss": 1.0844, + "mean_token_accuracy": 0.6761282682418823, + "num_tokens": 129623534.0, + "step": 5073 + }, + { + "epoch": 0.5572150230617176, + "grad_norm": 2.051041841506958, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.6957440972328186, + "num_tokens": 129651797.0, + "step": 5074 + }, + { + "epoch": 0.5573248407643312, + "grad_norm": 2.26629900932312, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6982277631759644, + "num_tokens": 129675569.0, + "step": 5075 + }, + { + "epoch": 0.5574346584669448, + "grad_norm": 2.4155256748199463, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7153082489967346, + "num_tokens": 129697291.0, + "step": 5076 + }, + { + "epoch": 0.5575444761695585, + "grad_norm": 2.2103271484375, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6851522922515869, + "num_tokens": 129722207.0, + "step": 5077 + }, + { + "epoch": 0.5576542938721722, + "grad_norm": 2.1656301021575928, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.718275249004364, + "num_tokens": 129748643.0, + "step": 5078 + }, + { + "epoch": 0.5577641115747859, + "grad_norm": 2.196539878845215, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.7003163695335388, + "num_tokens": 129773588.0, + "step": 5079 + }, + { + "epoch": 0.5578739292773995, + "grad_norm": 2.225179433822632, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.7032533288002014, + "num_tokens": 129798359.0, + "step": 5080 + }, + { + "epoch": 0.5579837469800132, + "grad_norm": 2.205226421356201, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7138279676437378, + "num_tokens": 129823472.0, + "step": 5081 + }, + { + "epoch": 0.5580935646826268, + "grad_norm": 2.1723086833953857, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7199996709823608, + "num_tokens": 129847121.0, + "step": 5082 + }, + { + "epoch": 0.5582033823852405, + "grad_norm": 2.112834930419922, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7016241550445557, + "num_tokens": 129873027.0, + "step": 5083 + }, + { + "epoch": 0.5583132000878541, + "grad_norm": 2.1616880893707275, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6989632248878479, + "num_tokens": 129898511.0, + "step": 5084 + }, + { + "epoch": 0.5584230177904678, + "grad_norm": 2.1299240589141846, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7041513919830322, + "num_tokens": 129925022.0, + "step": 5085 + }, + { + "epoch": 0.5585328354930815, + "grad_norm": 2.404477834701538, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6915920972824097, + "num_tokens": 129947349.0, + "step": 5086 + }, + { + "epoch": 0.5586426531956952, + "grad_norm": 2.0313141345977783, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7058807015419006, + "num_tokens": 129976134.0, + "step": 5087 + }, + { + "epoch": 0.5587524708983088, + "grad_norm": 2.4025001525878906, + "learning_rate": 1e-06, + "loss": 1.0427, + "mean_token_accuracy": 0.6889500617980957, + "num_tokens": 129998906.0, + "step": 5088 + }, + { + "epoch": 0.5588622886009225, + "grad_norm": 2.4756200313568115, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6937487125396729, + "num_tokens": 130019820.0, + "step": 5089 + }, + { + "epoch": 0.5589721063035361, + "grad_norm": 2.1601638793945312, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.689195454120636, + "num_tokens": 130045203.0, + "step": 5090 + }, + { + "epoch": 0.5590819240061498, + "grad_norm": 2.207146406173706, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7127647995948792, + "num_tokens": 130069001.0, + "step": 5091 + }, + { + "epoch": 0.5591917417087634, + "grad_norm": 2.297982931137085, + "learning_rate": 1e-06, + "loss": 1.0554, + "mean_token_accuracy": 0.6913641691207886, + "num_tokens": 130097128.0, + "step": 5092 + }, + { + "epoch": 0.5593015594113772, + "grad_norm": 2.0059564113616943, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.709915280342102, + "num_tokens": 130124830.0, + "step": 5093 + }, + { + "epoch": 0.5594113771139908, + "grad_norm": 2.075662136077881, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.6936473250389099, + "num_tokens": 130154558.0, + "step": 5094 + }, + { + "epoch": 0.5595211948166045, + "grad_norm": 2.0955400466918945, + "learning_rate": 1e-06, + "loss": 1.0366, + "mean_token_accuracy": 0.6918289065361023, + "num_tokens": 130182991.0, + "step": 5095 + }, + { + "epoch": 0.5596310125192181, + "grad_norm": 2.068308115005493, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7011652588844299, + "num_tokens": 130212755.0, + "step": 5096 + }, + { + "epoch": 0.5597408302218317, + "grad_norm": 1.974555492401123, + "learning_rate": 1e-06, + "loss": 1.039, + "mean_token_accuracy": 0.6836885809898376, + "num_tokens": 130243454.0, + "step": 5097 + }, + { + "epoch": 0.5598506479244454, + "grad_norm": 2.2974367141723633, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.691723108291626, + "num_tokens": 130268166.0, + "step": 5098 + }, + { + "epoch": 0.559960465627059, + "grad_norm": 2.196768283843994, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7038006782531738, + "num_tokens": 130298217.0, + "step": 5099 + }, + { + "epoch": 0.5600702833296728, + "grad_norm": 2.2009215354919434, + "learning_rate": 1e-06, + "loss": 1.0411, + "mean_token_accuracy": 0.6988941431045532, + "num_tokens": 130325713.0, + "step": 5100 + }, + { + "epoch": 0.5601801010322864, + "grad_norm": 2.410098075866699, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7037432193756104, + "num_tokens": 130347506.0, + "step": 5101 + }, + { + "epoch": 0.5602899187349001, + "grad_norm": 2.261455535888672, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.6924908757209778, + "num_tokens": 130371255.0, + "step": 5102 + }, + { + "epoch": 0.5603997364375137, + "grad_norm": 2.1740338802337646, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.6973206400871277, + "num_tokens": 130398489.0, + "step": 5103 + }, + { + "epoch": 0.5605095541401274, + "grad_norm": 2.317408561706543, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6954141855239868, + "num_tokens": 130421423.0, + "step": 5104 + }, + { + "epoch": 0.560619371842741, + "grad_norm": 2.0422844886779785, + "learning_rate": 1e-06, + "loss": 1.0546, + "mean_token_accuracy": 0.6858506798744202, + "num_tokens": 130452419.0, + "step": 5105 + }, + { + "epoch": 0.5607291895453547, + "grad_norm": 2.16965389251709, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6962348222732544, + "num_tokens": 130477551.0, + "step": 5106 + }, + { + "epoch": 0.5608390072479684, + "grad_norm": 2.2207887172698975, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7329311370849609, + "num_tokens": 130504436.0, + "step": 5107 + }, + { + "epoch": 0.5609488249505821, + "grad_norm": 2.4011199474334717, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7045818567276001, + "num_tokens": 130526454.0, + "step": 5108 + }, + { + "epoch": 0.5610586426531957, + "grad_norm": 2.2201473712921143, + "learning_rate": 1e-06, + "loss": 1.039, + "mean_token_accuracy": 0.6823638081550598, + "num_tokens": 130552812.0, + "step": 5109 + }, + { + "epoch": 0.5611684603558094, + "grad_norm": 2.4743072986602783, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7051807641983032, + "num_tokens": 130573620.0, + "step": 5110 + }, + { + "epoch": 0.561278278058423, + "grad_norm": 1.9923433065414429, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.6948366761207581, + "num_tokens": 130602186.0, + "step": 5111 + }, + { + "epoch": 0.5613880957610367, + "grad_norm": 2.1424896717071533, + "learning_rate": 1e-06, + "loss": 1.0731, + "mean_token_accuracy": 0.6946439743041992, + "num_tokens": 130628140.0, + "step": 5112 + }, + { + "epoch": 0.5614979134636503, + "grad_norm": 2.3743770122528076, + "learning_rate": 1e-06, + "loss": 1.0837, + "mean_token_accuracy": 0.6803247928619385, + "num_tokens": 130651849.0, + "step": 5113 + }, + { + "epoch": 0.5616077311662641, + "grad_norm": 2.338592290878296, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7056053280830383, + "num_tokens": 130673447.0, + "step": 5114 + }, + { + "epoch": 0.5617175488688777, + "grad_norm": 2.1639294624328613, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.6959699392318726, + "num_tokens": 130700542.0, + "step": 5115 + }, + { + "epoch": 0.5618273665714913, + "grad_norm": 2.0790741443634033, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.73012775182724, + "num_tokens": 130727175.0, + "step": 5116 + }, + { + "epoch": 0.561937184274105, + "grad_norm": 2.2571187019348145, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7263250350952148, + "num_tokens": 130751100.0, + "step": 5117 + }, + { + "epoch": 0.5620470019767186, + "grad_norm": 1.9245409965515137, + "learning_rate": 1e-06, + "loss": 1.0888, + "mean_token_accuracy": 0.6737624406814575, + "num_tokens": 130785380.0, + "step": 5118 + }, + { + "epoch": 0.5621568196793323, + "grad_norm": 2.5137789249420166, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7162479162216187, + "num_tokens": 130806666.0, + "step": 5119 + }, + { + "epoch": 0.5622666373819459, + "grad_norm": 2.1045989990234375, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.7038072347640991, + "num_tokens": 130836003.0, + "step": 5120 + }, + { + "epoch": 0.5623764550845596, + "grad_norm": 2.009446620941162, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.6914982795715332, + "num_tokens": 130866172.0, + "step": 5121 + }, + { + "epoch": 0.5624862727871733, + "grad_norm": 2.1265780925750732, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.719960629940033, + "num_tokens": 130890879.0, + "step": 5122 + }, + { + "epoch": 0.562596090489787, + "grad_norm": 1.9774398803710938, + "learning_rate": 1e-06, + "loss": 1.0853, + "mean_token_accuracy": 0.6742534637451172, + "num_tokens": 130924470.0, + "step": 5123 + }, + { + "epoch": 0.5627059081924006, + "grad_norm": 2.207777261734009, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7198062539100647, + "num_tokens": 130949564.0, + "step": 5124 + }, + { + "epoch": 0.5628157258950143, + "grad_norm": 2.403461456298828, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.6936172246932983, + "num_tokens": 130974249.0, + "step": 5125 + }, + { + "epoch": 0.5629255435976279, + "grad_norm": 2.0938806533813477, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7016534805297852, + "num_tokens": 131002027.0, + "step": 5126 + }, + { + "epoch": 0.5630353613002416, + "grad_norm": 2.3049981594085693, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.716293454170227, + "num_tokens": 131024751.0, + "step": 5127 + }, + { + "epoch": 0.5631451790028552, + "grad_norm": 2.498826742172241, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.716262936592102, + "num_tokens": 131045039.0, + "step": 5128 + }, + { + "epoch": 0.563254996705469, + "grad_norm": 2.0683321952819824, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7096918225288391, + "num_tokens": 131073829.0, + "step": 5129 + }, + { + "epoch": 0.5633648144080826, + "grad_norm": 2.086602210998535, + "learning_rate": 1e-06, + "loss": 1.1122, + "mean_token_accuracy": 0.6735206842422485, + "num_tokens": 131105075.0, + "step": 5130 + }, + { + "epoch": 0.5634746321106963, + "grad_norm": 1.875575304031372, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.722611665725708, + "num_tokens": 131138379.0, + "step": 5131 + }, + { + "epoch": 0.5635844498133099, + "grad_norm": 2.2213504314422607, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.691340446472168, + "num_tokens": 131165504.0, + "step": 5132 + }, + { + "epoch": 0.5636942675159236, + "grad_norm": 2.303725481033325, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.706447958946228, + "num_tokens": 131188079.0, + "step": 5133 + }, + { + "epoch": 0.5638040852185372, + "grad_norm": 2.136275053024292, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7351043820381165, + "num_tokens": 131215267.0, + "step": 5134 + }, + { + "epoch": 0.5639139029211508, + "grad_norm": 2.056201219558716, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.7059656977653503, + "num_tokens": 131243507.0, + "step": 5135 + }, + { + "epoch": 0.5640237206237646, + "grad_norm": 2.241579055786133, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7019014358520508, + "num_tokens": 131267215.0, + "step": 5136 + }, + { + "epoch": 0.5641335383263782, + "grad_norm": 2.0836663246154785, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7074544429779053, + "num_tokens": 131294906.0, + "step": 5137 + }, + { + "epoch": 0.5642433560289919, + "grad_norm": 2.413874864578247, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.7033654451370239, + "num_tokens": 131318768.0, + "step": 5138 + }, + { + "epoch": 0.5643531737316055, + "grad_norm": 2.076267957687378, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7153878211975098, + "num_tokens": 131345066.0, + "step": 5139 + }, + { + "epoch": 0.5644629914342192, + "grad_norm": 2.15423583984375, + "learning_rate": 1e-06, + "loss": 1.0228, + "mean_token_accuracy": 0.691852331161499, + "num_tokens": 131373445.0, + "step": 5140 + }, + { + "epoch": 0.5645728091368328, + "grad_norm": 2.1780481338500977, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7262872457504272, + "num_tokens": 131400408.0, + "step": 5141 + }, + { + "epoch": 0.5646826268394465, + "grad_norm": 2.285525321960449, + "learning_rate": 1e-06, + "loss": 1.0436, + "mean_token_accuracy": 0.6856168508529663, + "num_tokens": 131425704.0, + "step": 5142 + }, + { + "epoch": 0.5647924445420602, + "grad_norm": 2.0796122550964355, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.7070777416229248, + "num_tokens": 131455919.0, + "step": 5143 + }, + { + "epoch": 0.5649022622446739, + "grad_norm": 2.3604729175567627, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7205420732498169, + "num_tokens": 131476750.0, + "step": 5144 + }, + { + "epoch": 0.5650120799472875, + "grad_norm": 2.258829116821289, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.706416130065918, + "num_tokens": 131500668.0, + "step": 5145 + }, + { + "epoch": 0.5651218976499012, + "grad_norm": 2.139446496963501, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7036283612251282, + "num_tokens": 131528435.0, + "step": 5146 + }, + { + "epoch": 0.5652317153525148, + "grad_norm": 2.230865955352783, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.683158278465271, + "num_tokens": 131553281.0, + "step": 5147 + }, + { + "epoch": 0.5653415330551285, + "grad_norm": 2.1167330741882324, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7061115503311157, + "num_tokens": 131579630.0, + "step": 5148 + }, + { + "epoch": 0.5654513507577421, + "grad_norm": 2.0566999912261963, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7205460667610168, + "num_tokens": 131607997.0, + "step": 5149 + }, + { + "epoch": 0.5655611684603558, + "grad_norm": 2.26053524017334, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.708196759223938, + "num_tokens": 131631264.0, + "step": 5150 + }, + { + "epoch": 0.5656709861629695, + "grad_norm": 2.32737135887146, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7072873711585999, + "num_tokens": 131655112.0, + "step": 5151 + }, + { + "epoch": 0.5657808038655832, + "grad_norm": 2.193033218383789, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7155853509902954, + "num_tokens": 131679783.0, + "step": 5152 + }, + { + "epoch": 0.5658906215681968, + "grad_norm": 2.171645402908325, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.6977136135101318, + "num_tokens": 131706914.0, + "step": 5153 + }, + { + "epoch": 0.5660004392708105, + "grad_norm": 2.0240607261657715, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7135035991668701, + "num_tokens": 131735920.0, + "step": 5154 + }, + { + "epoch": 0.5661102569734241, + "grad_norm": 2.130007266998291, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7120506167411804, + "num_tokens": 131764580.0, + "step": 5155 + }, + { + "epoch": 0.5662200746760377, + "grad_norm": 2.2134037017822266, + "learning_rate": 1e-06, + "loss": 1.034, + "mean_token_accuracy": 0.6868361234664917, + "num_tokens": 131791460.0, + "step": 5156 + }, + { + "epoch": 0.5663298923786514, + "grad_norm": 2.295806407928467, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7163032293319702, + "num_tokens": 131816981.0, + "step": 5157 + }, + { + "epoch": 0.5664397100812651, + "grad_norm": 2.1272737979888916, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7099831104278564, + "num_tokens": 131843831.0, + "step": 5158 + }, + { + "epoch": 0.5665495277838788, + "grad_norm": 2.667454481124878, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.699321985244751, + "num_tokens": 131861824.0, + "step": 5159 + }, + { + "epoch": 0.5666593454864924, + "grad_norm": 2.2544362545013428, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7295523881912231, + "num_tokens": 131886105.0, + "step": 5160 + }, + { + "epoch": 0.5667691631891061, + "grad_norm": 2.7114369869232178, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7387001514434814, + "num_tokens": 131902081.0, + "step": 5161 + }, + { + "epoch": 0.5668789808917197, + "grad_norm": 2.2707459926605225, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7194013595581055, + "num_tokens": 131924721.0, + "step": 5162 + }, + { + "epoch": 0.5669887985943334, + "grad_norm": 1.9646692276000977, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.7000858783721924, + "num_tokens": 131955805.0, + "step": 5163 + }, + { + "epoch": 0.567098616296947, + "grad_norm": 1.9920923709869385, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7144814729690552, + "num_tokens": 131986149.0, + "step": 5164 + }, + { + "epoch": 0.5672084339995608, + "grad_norm": 2.354201316833496, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.7009085416793823, + "num_tokens": 132008036.0, + "step": 5165 + }, + { + "epoch": 0.5673182517021744, + "grad_norm": 2.0753183364868164, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6991628408432007, + "num_tokens": 132039105.0, + "step": 5166 + }, + { + "epoch": 0.5674280694047881, + "grad_norm": 2.1122853755950928, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6982345581054688, + "num_tokens": 132068057.0, + "step": 5167 + }, + { + "epoch": 0.5675378871074017, + "grad_norm": 2.52435040473938, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7313644289970398, + "num_tokens": 132086962.0, + "step": 5168 + }, + { + "epoch": 0.5676477048100154, + "grad_norm": 1.975690484046936, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.688011109828949, + "num_tokens": 132117621.0, + "step": 5169 + }, + { + "epoch": 0.567757522512629, + "grad_norm": 2.520113945007324, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7103031873703003, + "num_tokens": 132137970.0, + "step": 5170 + }, + { + "epoch": 0.5678673402152427, + "grad_norm": 2.180495500564575, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7154670357704163, + "num_tokens": 132163907.0, + "step": 5171 + }, + { + "epoch": 0.5679771579178564, + "grad_norm": 2.48449444770813, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.6971420645713806, + "num_tokens": 132185708.0, + "step": 5172 + }, + { + "epoch": 0.5680869756204701, + "grad_norm": 2.1011219024658203, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7030408382415771, + "num_tokens": 132213477.0, + "step": 5173 + }, + { + "epoch": 0.5681967933230837, + "grad_norm": 2.5734658241271973, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7002685070037842, + "num_tokens": 132242847.0, + "step": 5174 + }, + { + "epoch": 0.5683066110256974, + "grad_norm": 2.455339193344116, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7229149341583252, + "num_tokens": 132263850.0, + "step": 5175 + }, + { + "epoch": 0.568416428728311, + "grad_norm": 2.1150853633880615, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.6968591213226318, + "num_tokens": 132292812.0, + "step": 5176 + }, + { + "epoch": 0.5685262464309246, + "grad_norm": 2.613741397857666, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7085167765617371, + "num_tokens": 132314497.0, + "step": 5177 + }, + { + "epoch": 0.5686360641335383, + "grad_norm": 1.8717641830444336, + "learning_rate": 1e-06, + "loss": 1.0381, + "mean_token_accuracy": 0.6872171759605408, + "num_tokens": 132349225.0, + "step": 5178 + }, + { + "epoch": 0.5687458818361519, + "grad_norm": 2.171372652053833, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.7012116312980652, + "num_tokens": 132376102.0, + "step": 5179 + }, + { + "epoch": 0.5688556995387657, + "grad_norm": 2.014157772064209, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6941030025482178, + "num_tokens": 132405206.0, + "step": 5180 + }, + { + "epoch": 0.5689655172413793, + "grad_norm": 2.148186683654785, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.691834568977356, + "num_tokens": 132432749.0, + "step": 5181 + }, + { + "epoch": 0.569075334943993, + "grad_norm": 2.133798599243164, + "learning_rate": 1e-06, + "loss": 1.0481, + "mean_token_accuracy": 0.6858314275741577, + "num_tokens": 132461406.0, + "step": 5182 + }, + { + "epoch": 0.5691851526466066, + "grad_norm": 1.8902556896209717, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.702989935874939, + "num_tokens": 132494024.0, + "step": 5183 + }, + { + "epoch": 0.5692949703492203, + "grad_norm": 2.3585824966430664, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6993553042411804, + "num_tokens": 132518596.0, + "step": 5184 + }, + { + "epoch": 0.5694047880518339, + "grad_norm": 2.001164197921753, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.6999626755714417, + "num_tokens": 132548695.0, + "step": 5185 + }, + { + "epoch": 0.5695146057544476, + "grad_norm": 2.2039332389831543, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7315380573272705, + "num_tokens": 132572081.0, + "step": 5186 + }, + { + "epoch": 0.5696244234570613, + "grad_norm": 2.3059349060058594, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6948357820510864, + "num_tokens": 132596491.0, + "step": 5187 + }, + { + "epoch": 0.569734241159675, + "grad_norm": 2.1926629543304443, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7294172048568726, + "num_tokens": 132622632.0, + "step": 5188 + }, + { + "epoch": 0.5698440588622886, + "grad_norm": 2.5061216354370117, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7223296761512756, + "num_tokens": 132642872.0, + "step": 5189 + }, + { + "epoch": 0.5699538765649023, + "grad_norm": 2.2852139472961426, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7092533707618713, + "num_tokens": 132666134.0, + "step": 5190 + }, + { + "epoch": 0.5700636942675159, + "grad_norm": 2.0722525119781494, + "learning_rate": 1e-06, + "loss": 1.0285, + "mean_token_accuracy": 0.6891306638717651, + "num_tokens": 132694521.0, + "step": 5191 + }, + { + "epoch": 0.5701735119701296, + "grad_norm": 2.011131525039673, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6960245966911316, + "num_tokens": 132722488.0, + "step": 5192 + }, + { + "epoch": 0.5702833296727432, + "grad_norm": 2.2863526344299316, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6994380354881287, + "num_tokens": 132745990.0, + "step": 5193 + }, + { + "epoch": 0.570393147375357, + "grad_norm": 2.46114444732666, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7267479300498962, + "num_tokens": 132765407.0, + "step": 5194 + }, + { + "epoch": 0.5705029650779706, + "grad_norm": 1.8740427494049072, + "learning_rate": 1e-06, + "loss": 1.0412, + "mean_token_accuracy": 0.6827425360679626, + "num_tokens": 132801604.0, + "step": 5195 + }, + { + "epoch": 0.5706127827805842, + "grad_norm": 2.0660176277160645, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7072588801383972, + "num_tokens": 132828279.0, + "step": 5196 + }, + { + "epoch": 0.5707226004831979, + "grad_norm": 2.1482038497924805, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.695851743221283, + "num_tokens": 132853731.0, + "step": 5197 + }, + { + "epoch": 0.5708324181858115, + "grad_norm": 2.3349368572235107, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7008805871009827, + "num_tokens": 132875473.0, + "step": 5198 + }, + { + "epoch": 0.5709422358884252, + "grad_norm": 2.2414262294769287, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6985551118850708, + "num_tokens": 132900491.0, + "step": 5199 + }, + { + "epoch": 0.5710520535910388, + "grad_norm": 2.204103469848633, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7191270589828491, + "num_tokens": 132925188.0, + "step": 5200 + }, + { + "epoch": 0.5711618712936526, + "grad_norm": 2.178283929824829, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6922187209129333, + "num_tokens": 132953279.0, + "step": 5201 + }, + { + "epoch": 0.5712716889962662, + "grad_norm": 1.983218789100647, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.6988095045089722, + "num_tokens": 132985638.0, + "step": 5202 + }, + { + "epoch": 0.5713815066988799, + "grad_norm": 2.117560386657715, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7096092700958252, + "num_tokens": 133010688.0, + "step": 5203 + }, + { + "epoch": 0.5714913244014935, + "grad_norm": 2.141160011291504, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7018729448318481, + "num_tokens": 133037216.0, + "step": 5204 + }, + { + "epoch": 0.5716011421041072, + "grad_norm": 2.5855398178100586, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7126110792160034, + "num_tokens": 133056750.0, + "step": 5205 + }, + { + "epoch": 0.5717109598067208, + "grad_norm": 2.2184226512908936, + "learning_rate": 1e-06, + "loss": 1.0433, + "mean_token_accuracy": 0.6915255188941956, + "num_tokens": 133083278.0, + "step": 5206 + }, + { + "epoch": 0.5718207775093345, + "grad_norm": 2.0706353187561035, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7308179140090942, + "num_tokens": 133110014.0, + "step": 5207 + }, + { + "epoch": 0.5719305952119481, + "grad_norm": 2.339143991470337, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.700867772102356, + "num_tokens": 133133227.0, + "step": 5208 + }, + { + "epoch": 0.5720404129145619, + "grad_norm": 2.210047721862793, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.6855577230453491, + "num_tokens": 133159442.0, + "step": 5209 + }, + { + "epoch": 0.5721502306171755, + "grad_norm": 2.151888608932495, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.6974372863769531, + "num_tokens": 133185713.0, + "step": 5210 + }, + { + "epoch": 0.5722600483197892, + "grad_norm": 2.4886977672576904, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.6992747783660889, + "num_tokens": 133207123.0, + "step": 5211 + }, + { + "epoch": 0.5723698660224028, + "grad_norm": 2.389796733856201, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7101632356643677, + "num_tokens": 133228422.0, + "step": 5212 + }, + { + "epoch": 0.5724796837250165, + "grad_norm": 2.1343600749969482, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.7013906836509705, + "num_tokens": 133257505.0, + "step": 5213 + }, + { + "epoch": 0.5725895014276301, + "grad_norm": 2.079450845718384, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.714850664138794, + "num_tokens": 133284121.0, + "step": 5214 + }, + { + "epoch": 0.5726993191302437, + "grad_norm": 2.295158863067627, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.6922764778137207, + "num_tokens": 133308728.0, + "step": 5215 + }, + { + "epoch": 0.5728091368328575, + "grad_norm": 2.0969457626342773, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.6976003050804138, + "num_tokens": 133336024.0, + "step": 5216 + }, + { + "epoch": 0.5729189545354711, + "grad_norm": 2.226456880569458, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.6858619451522827, + "num_tokens": 133362621.0, + "step": 5217 + }, + { + "epoch": 0.5730287722380848, + "grad_norm": 2.12762451171875, + "learning_rate": 1e-06, + "loss": 1.0505, + "mean_token_accuracy": 0.6901290416717529, + "num_tokens": 133393100.0, + "step": 5218 + }, + { + "epoch": 0.5731385899406984, + "grad_norm": 2.3217480182647705, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.7023499011993408, + "num_tokens": 133416697.0, + "step": 5219 + }, + { + "epoch": 0.5732484076433121, + "grad_norm": 2.456611156463623, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7223052382469177, + "num_tokens": 133437166.0, + "step": 5220 + }, + { + "epoch": 0.5733582253459257, + "grad_norm": 2.2031919956207275, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.6995227336883545, + "num_tokens": 133462846.0, + "step": 5221 + }, + { + "epoch": 0.5734680430485394, + "grad_norm": 2.2209274768829346, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.697110652923584, + "num_tokens": 133488753.0, + "step": 5222 + }, + { + "epoch": 0.5735778607511531, + "grad_norm": 2.672912836074829, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7091571092605591, + "num_tokens": 133507499.0, + "step": 5223 + }, + { + "epoch": 0.5736876784537668, + "grad_norm": 2.231971263885498, + "learning_rate": 1e-06, + "loss": 1.0407, + "mean_token_accuracy": 0.6909307241439819, + "num_tokens": 133534886.0, + "step": 5224 + }, + { + "epoch": 0.5737974961563804, + "grad_norm": 2.0116348266601562, + "learning_rate": 1e-06, + "loss": 1.0793, + "mean_token_accuracy": 0.6752097606658936, + "num_tokens": 133565521.0, + "step": 5225 + }, + { + "epoch": 0.5739073138589941, + "grad_norm": 2.159179449081421, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7057539224624634, + "num_tokens": 133593685.0, + "step": 5226 + }, + { + "epoch": 0.5740171315616077, + "grad_norm": 2.3818514347076416, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.715941309928894, + "num_tokens": 133615704.0, + "step": 5227 + }, + { + "epoch": 0.5741269492642214, + "grad_norm": 2.10730242729187, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.6913232803344727, + "num_tokens": 133644573.0, + "step": 5228 + }, + { + "epoch": 0.574236766966835, + "grad_norm": 2.4767048358917236, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7052225470542908, + "num_tokens": 133665230.0, + "step": 5229 + }, + { + "epoch": 0.5743465846694488, + "grad_norm": 2.010559320449829, + "learning_rate": 1e-06, + "loss": 1.0499, + "mean_token_accuracy": 0.6874926090240479, + "num_tokens": 133697766.0, + "step": 5230 + }, + { + "epoch": 0.5744564023720624, + "grad_norm": 2.2679786682128906, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.7068604230880737, + "num_tokens": 133723102.0, + "step": 5231 + }, + { + "epoch": 0.5745662200746761, + "grad_norm": 2.4066245555877686, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.71731036901474, + "num_tokens": 133744395.0, + "step": 5232 + }, + { + "epoch": 0.5746760377772897, + "grad_norm": 2.3116726875305176, + "learning_rate": 1e-06, + "loss": 1.0586, + "mean_token_accuracy": 0.6828966736793518, + "num_tokens": 133768410.0, + "step": 5233 + }, + { + "epoch": 0.5747858554799034, + "grad_norm": 2.0108566284179688, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7066856622695923, + "num_tokens": 133799111.0, + "step": 5234 + }, + { + "epoch": 0.574895673182517, + "grad_norm": 2.2645676136016846, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7115681171417236, + "num_tokens": 133823076.0, + "step": 5235 + }, + { + "epoch": 0.5750054908851306, + "grad_norm": 2.046687602996826, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7146927714347839, + "num_tokens": 133851203.0, + "step": 5236 + }, + { + "epoch": 0.5751153085877443, + "grad_norm": 2.4981017112731934, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7137110829353333, + "num_tokens": 133870637.0, + "step": 5237 + }, + { + "epoch": 0.575225126290358, + "grad_norm": 2.011878252029419, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7073957920074463, + "num_tokens": 133900478.0, + "step": 5238 + }, + { + "epoch": 0.5753349439929717, + "grad_norm": 2.297799587249756, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6975668668746948, + "num_tokens": 133924906.0, + "step": 5239 + }, + { + "epoch": 0.5754447616955853, + "grad_norm": 2.0349321365356445, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.697813868522644, + "num_tokens": 133956232.0, + "step": 5240 + }, + { + "epoch": 0.575554579398199, + "grad_norm": 2.4285261631011963, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7071639895439148, + "num_tokens": 133977653.0, + "step": 5241 + }, + { + "epoch": 0.5756643971008126, + "grad_norm": 1.867732048034668, + "learning_rate": 1e-06, + "loss": 1.0442, + "mean_token_accuracy": 0.6864049434661865, + "num_tokens": 134011831.0, + "step": 5242 + }, + { + "epoch": 0.5757742148034263, + "grad_norm": 2.379023790359497, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.700381338596344, + "num_tokens": 134036522.0, + "step": 5243 + }, + { + "epoch": 0.5758840325060399, + "grad_norm": 2.4852688312530518, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7239013910293579, + "num_tokens": 134056111.0, + "step": 5244 + }, + { + "epoch": 0.5759938502086537, + "grad_norm": 2.323112726211548, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7145013213157654, + "num_tokens": 134078893.0, + "step": 5245 + }, + { + "epoch": 0.5761036679112673, + "grad_norm": 2.2319445610046387, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.725853681564331, + "num_tokens": 134102337.0, + "step": 5246 + }, + { + "epoch": 0.576213485613881, + "grad_norm": 2.250452756881714, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7041491270065308, + "num_tokens": 134126539.0, + "step": 5247 + }, + { + "epoch": 0.5763233033164946, + "grad_norm": 1.9754196405410767, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.7000181078910828, + "num_tokens": 134155913.0, + "step": 5248 + }, + { + "epoch": 0.5764331210191083, + "grad_norm": 1.9975942373275757, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6842961311340332, + "num_tokens": 134189663.0, + "step": 5249 + }, + { + "epoch": 0.5765429387217219, + "grad_norm": 2.388068675994873, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.7032905220985413, + "num_tokens": 134213535.0, + "step": 5250 + }, + { + "epoch": 0.5766527564243356, + "grad_norm": 2.271507978439331, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7055532336235046, + "num_tokens": 134237142.0, + "step": 5251 + }, + { + "epoch": 0.5767625741269493, + "grad_norm": 2.2329764366149902, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.6990963220596313, + "num_tokens": 134260706.0, + "step": 5252 + }, + { + "epoch": 0.576872391829563, + "grad_norm": 2.1064529418945312, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6962153911590576, + "num_tokens": 134288554.0, + "step": 5253 + }, + { + "epoch": 0.5769822095321766, + "grad_norm": 2.4105021953582764, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.7043722867965698, + "num_tokens": 134312906.0, + "step": 5254 + }, + { + "epoch": 0.5770920272347903, + "grad_norm": 2.3380286693573, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.689510703086853, + "num_tokens": 134336373.0, + "step": 5255 + }, + { + "epoch": 0.5772018449374039, + "grad_norm": 2.206165313720703, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7030678987503052, + "num_tokens": 134361612.0, + "step": 5256 + }, + { + "epoch": 0.5773116626400175, + "grad_norm": 2.5263853073120117, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7197569608688354, + "num_tokens": 134383044.0, + "step": 5257 + }, + { + "epoch": 0.5774214803426312, + "grad_norm": 2.2857918739318848, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.6982623338699341, + "num_tokens": 134408907.0, + "step": 5258 + }, + { + "epoch": 0.577531298045245, + "grad_norm": 2.164426565170288, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7216979265213013, + "num_tokens": 134434983.0, + "step": 5259 + }, + { + "epoch": 0.5776411157478586, + "grad_norm": 1.9600861072540283, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7177273035049438, + "num_tokens": 134464952.0, + "step": 5260 + }, + { + "epoch": 0.5777509334504722, + "grad_norm": 2.043395519256592, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6946761608123779, + "num_tokens": 134494377.0, + "step": 5261 + }, + { + "epoch": 0.5778607511530859, + "grad_norm": 2.503385305404663, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7169157266616821, + "num_tokens": 134513519.0, + "step": 5262 + }, + { + "epoch": 0.5779705688556995, + "grad_norm": 2.160963296890259, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7184327840805054, + "num_tokens": 134539485.0, + "step": 5263 + }, + { + "epoch": 0.5780803865583132, + "grad_norm": 2.3676071166992188, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7046006917953491, + "num_tokens": 134562064.0, + "step": 5264 + }, + { + "epoch": 0.5781902042609268, + "grad_norm": 2.588430643081665, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.708868145942688, + "num_tokens": 134581780.0, + "step": 5265 + }, + { + "epoch": 0.5783000219635406, + "grad_norm": 2.1529293060302734, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7035061120986938, + "num_tokens": 134610479.0, + "step": 5266 + }, + { + "epoch": 0.5784098396661542, + "grad_norm": 2.3510210514068604, + "learning_rate": 1e-06, + "loss": 1.1146, + "mean_token_accuracy": 0.6768959164619446, + "num_tokens": 134636261.0, + "step": 5267 + }, + { + "epoch": 0.5785196573687679, + "grad_norm": 1.965215802192688, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7042492628097534, + "num_tokens": 134668337.0, + "step": 5268 + }, + { + "epoch": 0.5786294750713815, + "grad_norm": 2.6045262813568115, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.6867469549179077, + "num_tokens": 134689375.0, + "step": 5269 + }, + { + "epoch": 0.5787392927739952, + "grad_norm": 2.305982828140259, + "learning_rate": 1e-06, + "loss": 1.0698, + "mean_token_accuracy": 0.6831462979316711, + "num_tokens": 134713077.0, + "step": 5270 + }, + { + "epoch": 0.5788491104766088, + "grad_norm": 2.14768385887146, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7002355456352234, + "num_tokens": 134739992.0, + "step": 5271 + }, + { + "epoch": 0.5789589281792225, + "grad_norm": 2.244431734085083, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7086526155471802, + "num_tokens": 134764983.0, + "step": 5272 + }, + { + "epoch": 0.5790687458818361, + "grad_norm": 2.008049726486206, + "learning_rate": 1e-06, + "loss": 1.0622, + "mean_token_accuracy": 0.6838811039924622, + "num_tokens": 134795016.0, + "step": 5273 + }, + { + "epoch": 0.5791785635844499, + "grad_norm": 2.376450538635254, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7115902304649353, + "num_tokens": 134817406.0, + "step": 5274 + }, + { + "epoch": 0.5792883812870635, + "grad_norm": 2.377598524093628, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7313549518585205, + "num_tokens": 134838708.0, + "step": 5275 + }, + { + "epoch": 0.5793981989896771, + "grad_norm": 2.3747684955596924, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7158827781677246, + "num_tokens": 134859449.0, + "step": 5276 + }, + { + "epoch": 0.5795080166922908, + "grad_norm": 2.3294596672058105, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6939287185668945, + "num_tokens": 134882680.0, + "step": 5277 + }, + { + "epoch": 0.5796178343949044, + "grad_norm": 2.370047092437744, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7177176475524902, + "num_tokens": 134904622.0, + "step": 5278 + }, + { + "epoch": 0.5797276520975181, + "grad_norm": 2.0610570907592773, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6919966340065002, + "num_tokens": 134932908.0, + "step": 5279 + }, + { + "epoch": 0.5798374698001317, + "grad_norm": 2.3895063400268555, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.6998162865638733, + "num_tokens": 134956074.0, + "step": 5280 + }, + { + "epoch": 0.5799472875027455, + "grad_norm": 2.156578302383423, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7077143788337708, + "num_tokens": 134982141.0, + "step": 5281 + }, + { + "epoch": 0.5800571052053591, + "grad_norm": 2.262556552886963, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7279261946678162, + "num_tokens": 135004471.0, + "step": 5282 + }, + { + "epoch": 0.5801669229079728, + "grad_norm": 2.5727756023406982, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7096685767173767, + "num_tokens": 135024864.0, + "step": 5283 + }, + { + "epoch": 0.5802767406105864, + "grad_norm": 2.272317409515381, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7069960832595825, + "num_tokens": 135049103.0, + "step": 5284 + }, + { + "epoch": 0.5803865583132001, + "grad_norm": 2.164241313934326, + "learning_rate": 1e-06, + "loss": 1.0336, + "mean_token_accuracy": 0.692474365234375, + "num_tokens": 135076312.0, + "step": 5285 + }, + { + "epoch": 0.5804963760158137, + "grad_norm": 2.1518056392669678, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7114217877388, + "num_tokens": 135101948.0, + "step": 5286 + }, + { + "epoch": 0.5806061937184274, + "grad_norm": 2.2823657989501953, + "learning_rate": 1e-06, + "loss": 1.0222, + "mean_token_accuracy": 0.69402015209198, + "num_tokens": 135127338.0, + "step": 5287 + }, + { + "epoch": 0.5807160114210411, + "grad_norm": 2.1262638568878174, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7015013098716736, + "num_tokens": 135154438.0, + "step": 5288 + }, + { + "epoch": 0.5808258291236548, + "grad_norm": 2.274423122406006, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7282164692878723, + "num_tokens": 135178745.0, + "step": 5289 + }, + { + "epoch": 0.5809356468262684, + "grad_norm": 2.041285514831543, + "learning_rate": 1e-06, + "loss": 1.0683, + "mean_token_accuracy": 0.6729447841644287, + "num_tokens": 135207371.0, + "step": 5290 + }, + { + "epoch": 0.5810454645288821, + "grad_norm": 2.102325677871704, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.6958357095718384, + "num_tokens": 135234547.0, + "step": 5291 + }, + { + "epoch": 0.5811552822314957, + "grad_norm": 2.3036012649536133, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7320735454559326, + "num_tokens": 135263427.0, + "step": 5292 + }, + { + "epoch": 0.5812650999341094, + "grad_norm": 2.376906633377075, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7095665335655212, + "num_tokens": 135285978.0, + "step": 5293 + }, + { + "epoch": 0.581374917636723, + "grad_norm": 2.0865085124969482, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7225471138954163, + "num_tokens": 135312661.0, + "step": 5294 + }, + { + "epoch": 0.5814847353393368, + "grad_norm": 2.129005193710327, + "learning_rate": 1e-06, + "loss": 1.0417, + "mean_token_accuracy": 0.698786199092865, + "num_tokens": 135341264.0, + "step": 5295 + }, + { + "epoch": 0.5815945530419504, + "grad_norm": 2.0789682865142822, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7043500542640686, + "num_tokens": 135367495.0, + "step": 5296 + }, + { + "epoch": 0.581704370744564, + "grad_norm": 2.074681520462036, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7113746404647827, + "num_tokens": 135395662.0, + "step": 5297 + }, + { + "epoch": 0.5818141884471777, + "grad_norm": 2.3837361335754395, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7153599262237549, + "num_tokens": 135417551.0, + "step": 5298 + }, + { + "epoch": 0.5819240061497913, + "grad_norm": 2.2760415077209473, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.694574236869812, + "num_tokens": 135443168.0, + "step": 5299 + }, + { + "epoch": 0.582033823852405, + "grad_norm": 2.253973960876465, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6966695785522461, + "num_tokens": 135468159.0, + "step": 5300 + }, + { + "epoch": 0.5821436415550186, + "grad_norm": 1.930856466293335, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7305599451065063, + "num_tokens": 135497264.0, + "step": 5301 + }, + { + "epoch": 0.5822534592576323, + "grad_norm": 1.9967776536941528, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.6967792510986328, + "num_tokens": 135526328.0, + "step": 5302 + }, + { + "epoch": 0.582363276960246, + "grad_norm": 2.157819986343384, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.6966089010238647, + "num_tokens": 135552169.0, + "step": 5303 + }, + { + "epoch": 0.5824730946628597, + "grad_norm": 2.1124160289764404, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7289667129516602, + "num_tokens": 135577879.0, + "step": 5304 + }, + { + "epoch": 0.5825829123654733, + "grad_norm": 2.1372947692871094, + "learning_rate": 1e-06, + "loss": 1.0685, + "mean_token_accuracy": 0.6846901774406433, + "num_tokens": 135606215.0, + "step": 5305 + }, + { + "epoch": 0.582692730068087, + "grad_norm": 2.1692488193511963, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.6986602544784546, + "num_tokens": 135630936.0, + "step": 5306 + }, + { + "epoch": 0.5828025477707006, + "grad_norm": 2.124831438064575, + "learning_rate": 1e-06, + "loss": 1.0553, + "mean_token_accuracy": 0.678467869758606, + "num_tokens": 135657404.0, + "step": 5307 + }, + { + "epoch": 0.5829123654733143, + "grad_norm": 2.0433545112609863, + "learning_rate": 1e-06, + "loss": 1.0848, + "mean_token_accuracy": 0.6768622398376465, + "num_tokens": 135687776.0, + "step": 5308 + }, + { + "epoch": 0.5830221831759279, + "grad_norm": 2.1240854263305664, + "learning_rate": 1e-06, + "loss": 1.0835, + "mean_token_accuracy": 0.679086446762085, + "num_tokens": 135716266.0, + "step": 5309 + }, + { + "epoch": 0.5831320008785417, + "grad_norm": 2.4426982402801514, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7165090441703796, + "num_tokens": 135736616.0, + "step": 5310 + }, + { + "epoch": 0.5832418185811553, + "grad_norm": 2.125251531600952, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.7008824944496155, + "num_tokens": 135762965.0, + "step": 5311 + }, + { + "epoch": 0.583351636283769, + "grad_norm": 2.022073984146118, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7048896551132202, + "num_tokens": 135791974.0, + "step": 5312 + }, + { + "epoch": 0.5834614539863826, + "grad_norm": 2.3723700046539307, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7077864408493042, + "num_tokens": 135814712.0, + "step": 5313 + }, + { + "epoch": 0.5835712716889963, + "grad_norm": 2.1856625080108643, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.699596643447876, + "num_tokens": 135840413.0, + "step": 5314 + }, + { + "epoch": 0.5836810893916099, + "grad_norm": 2.0286941528320312, + "learning_rate": 1e-06, + "loss": 1.0465, + "mean_token_accuracy": 0.687652587890625, + "num_tokens": 135868362.0, + "step": 5315 + }, + { + "epoch": 0.5837909070942235, + "grad_norm": 2.4506590366363525, + "learning_rate": 1e-06, + "loss": 1.0381, + "mean_token_accuracy": 0.6907340288162231, + "num_tokens": 135889768.0, + "step": 5316 + }, + { + "epoch": 0.5839007247968373, + "grad_norm": 2.364684820175171, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7045695781707764, + "num_tokens": 135910913.0, + "step": 5317 + }, + { + "epoch": 0.584010542499451, + "grad_norm": 2.2083871364593506, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7052146196365356, + "num_tokens": 135935974.0, + "step": 5318 + }, + { + "epoch": 0.5841203602020646, + "grad_norm": 2.1646921634674072, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.722533643245697, + "num_tokens": 135961908.0, + "step": 5319 + }, + { + "epoch": 0.5842301779046782, + "grad_norm": 2.229759931564331, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.7018722295761108, + "num_tokens": 135985404.0, + "step": 5320 + }, + { + "epoch": 0.5843399956072919, + "grad_norm": 2.0649678707122803, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.6982519030570984, + "num_tokens": 136013544.0, + "step": 5321 + }, + { + "epoch": 0.5844498133099055, + "grad_norm": 2.315434455871582, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.6928778290748596, + "num_tokens": 136039388.0, + "step": 5322 + }, + { + "epoch": 0.5845596310125192, + "grad_norm": 1.8201138973236084, + "learning_rate": 1e-06, + "loss": 1.0745, + "mean_token_accuracy": 0.6800323724746704, + "num_tokens": 136074822.0, + "step": 5323 + }, + { + "epoch": 0.5846694487151329, + "grad_norm": 1.935565710067749, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6869075298309326, + "num_tokens": 136106995.0, + "step": 5324 + }, + { + "epoch": 0.5847792664177466, + "grad_norm": 1.963432788848877, + "learning_rate": 1e-06, + "loss": 1.0398, + "mean_token_accuracy": 0.6928228139877319, + "num_tokens": 136138878.0, + "step": 5325 + }, + { + "epoch": 0.5848890841203602, + "grad_norm": 2.3269758224487305, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7156783938407898, + "num_tokens": 136161237.0, + "step": 5326 + }, + { + "epoch": 0.5849989018229739, + "grad_norm": 2.3249213695526123, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7089858055114746, + "num_tokens": 136184706.0, + "step": 5327 + }, + { + "epoch": 0.5851087195255875, + "grad_norm": 2.260946035385132, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7099902033805847, + "num_tokens": 136208436.0, + "step": 5328 + }, + { + "epoch": 0.5852185372282012, + "grad_norm": 2.405003309249878, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7253791093826294, + "num_tokens": 136228979.0, + "step": 5329 + }, + { + "epoch": 0.5853283549308148, + "grad_norm": 2.4071524143218994, + "learning_rate": 1e-06, + "loss": 1.0296, + "mean_token_accuracy": 0.6867721080780029, + "num_tokens": 136251832.0, + "step": 5330 + }, + { + "epoch": 0.5854381726334285, + "grad_norm": 2.2424240112304688, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7266184687614441, + "num_tokens": 136274713.0, + "step": 5331 + }, + { + "epoch": 0.5855479903360422, + "grad_norm": 2.28714919090271, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.736009418964386, + "num_tokens": 136298179.0, + "step": 5332 + }, + { + "epoch": 0.5856578080386559, + "grad_norm": 2.032038688659668, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.7020332217216492, + "num_tokens": 136326522.0, + "step": 5333 + }, + { + "epoch": 0.5857676257412695, + "grad_norm": 2.4378914833068848, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7292335033416748, + "num_tokens": 136346109.0, + "step": 5334 + }, + { + "epoch": 0.5858774434438832, + "grad_norm": 2.215634346008301, + "learning_rate": 1e-06, + "loss": 1.0272, + "mean_token_accuracy": 0.6946288347244263, + "num_tokens": 136373614.0, + "step": 5335 + }, + { + "epoch": 0.5859872611464968, + "grad_norm": 2.2514665126800537, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.6849210262298584, + "num_tokens": 136400148.0, + "step": 5336 + }, + { + "epoch": 0.5860970788491104, + "grad_norm": 2.0200657844543457, + "learning_rate": 1e-06, + "loss": 1.0475, + "mean_token_accuracy": 0.6794916391372681, + "num_tokens": 136430638.0, + "step": 5337 + }, + { + "epoch": 0.5862068965517241, + "grad_norm": 2.284346342086792, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7353113889694214, + "num_tokens": 136453457.0, + "step": 5338 + }, + { + "epoch": 0.5863167142543378, + "grad_norm": 2.185500144958496, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.6967089176177979, + "num_tokens": 136480526.0, + "step": 5339 + }, + { + "epoch": 0.5864265319569515, + "grad_norm": 1.9509538412094116, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7063806056976318, + "num_tokens": 136509759.0, + "step": 5340 + }, + { + "epoch": 0.5865363496595651, + "grad_norm": 2.216169834136963, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7033293843269348, + "num_tokens": 136535521.0, + "step": 5341 + }, + { + "epoch": 0.5866461673621788, + "grad_norm": 2.3127665519714355, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6901450157165527, + "num_tokens": 136558044.0, + "step": 5342 + }, + { + "epoch": 0.5867559850647924, + "grad_norm": 2.1460344791412354, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7001857757568359, + "num_tokens": 136586912.0, + "step": 5343 + }, + { + "epoch": 0.5868658027674061, + "grad_norm": 2.0045268535614014, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6926512718200684, + "num_tokens": 136618169.0, + "step": 5344 + }, + { + "epoch": 0.5869756204700197, + "grad_norm": 2.19408917427063, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.731020450592041, + "num_tokens": 136642626.0, + "step": 5345 + }, + { + "epoch": 0.5870854381726335, + "grad_norm": 2.388819694519043, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7255960702896118, + "num_tokens": 136663204.0, + "step": 5346 + }, + { + "epoch": 0.5871952558752471, + "grad_norm": 2.4902827739715576, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7062215805053711, + "num_tokens": 136684590.0, + "step": 5347 + }, + { + "epoch": 0.5873050735778608, + "grad_norm": 2.0773074626922607, + "learning_rate": 1e-06, + "loss": 1.0766, + "mean_token_accuracy": 0.6779825687408447, + "num_tokens": 136715297.0, + "step": 5348 + }, + { + "epoch": 0.5874148912804744, + "grad_norm": 2.035658121109009, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6967229247093201, + "num_tokens": 136744394.0, + "step": 5349 + }, + { + "epoch": 0.5875247089830881, + "grad_norm": 2.04549241065979, + "learning_rate": 1e-06, + "loss": 1.1098, + "mean_token_accuracy": 0.6665891408920288, + "num_tokens": 136774668.0, + "step": 5350 + }, + { + "epoch": 0.5876345266857017, + "grad_norm": 2.0430872440338135, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7129660248756409, + "num_tokens": 136802990.0, + "step": 5351 + }, + { + "epoch": 0.5877443443883154, + "grad_norm": 2.3252577781677246, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.720517635345459, + "num_tokens": 136824366.0, + "step": 5352 + }, + { + "epoch": 0.5878541620909291, + "grad_norm": 2.1417832374572754, + "learning_rate": 1e-06, + "loss": 1.0895, + "mean_token_accuracy": 0.6928514242172241, + "num_tokens": 136850258.0, + "step": 5353 + }, + { + "epoch": 0.5879639797935428, + "grad_norm": 2.0790340900421143, + "learning_rate": 1e-06, + "loss": 1.0705, + "mean_token_accuracy": 0.6748162508010864, + "num_tokens": 136878326.0, + "step": 5354 + }, + { + "epoch": 0.5880737974961564, + "grad_norm": 2.155050277709961, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7054411172866821, + "num_tokens": 136903619.0, + "step": 5355 + }, + { + "epoch": 0.58818361519877, + "grad_norm": 2.1938223838806152, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6976556181907654, + "num_tokens": 136928272.0, + "step": 5356 + }, + { + "epoch": 0.5882934329013837, + "grad_norm": 2.6144602298736572, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7195217609405518, + "num_tokens": 136947548.0, + "step": 5357 + }, + { + "epoch": 0.5884032506039973, + "grad_norm": 2.3347890377044678, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.7039546966552734, + "num_tokens": 136971744.0, + "step": 5358 + }, + { + "epoch": 0.588513068306611, + "grad_norm": 2.5987486839294434, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7038650512695312, + "num_tokens": 136992093.0, + "step": 5359 + }, + { + "epoch": 0.5886228860092246, + "grad_norm": 2.0842151641845703, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6960094571113586, + "num_tokens": 137020926.0, + "step": 5360 + }, + { + "epoch": 0.5887327037118384, + "grad_norm": 2.0548293590545654, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.745278537273407, + "num_tokens": 137047386.0, + "step": 5361 + }, + { + "epoch": 0.588842521414452, + "grad_norm": 2.3642923831939697, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7039272785186768, + "num_tokens": 137069670.0, + "step": 5362 + }, + { + "epoch": 0.5889523391170657, + "grad_norm": 2.080733299255371, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7154250741004944, + "num_tokens": 137096273.0, + "step": 5363 + }, + { + "epoch": 0.5890621568196793, + "grad_norm": 2.257136583328247, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7233264446258545, + "num_tokens": 137117926.0, + "step": 5364 + }, + { + "epoch": 0.589171974522293, + "grad_norm": 2.4054176807403564, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7272894382476807, + "num_tokens": 137139104.0, + "step": 5365 + }, + { + "epoch": 0.5892817922249066, + "grad_norm": 2.4113364219665527, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.738505482673645, + "num_tokens": 137159837.0, + "step": 5366 + }, + { + "epoch": 0.5893916099275203, + "grad_norm": 2.492621660232544, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7149978876113892, + "num_tokens": 137179919.0, + "step": 5367 + }, + { + "epoch": 0.589501427630134, + "grad_norm": 2.1851093769073486, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7123774290084839, + "num_tokens": 137205175.0, + "step": 5368 + }, + { + "epoch": 0.5896112453327477, + "grad_norm": 2.199327230453491, + "learning_rate": 1e-06, + "loss": 1.0195, + "mean_token_accuracy": 0.6913931369781494, + "num_tokens": 137230232.0, + "step": 5369 + }, + { + "epoch": 0.5897210630353613, + "grad_norm": 2.565932035446167, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.713040828704834, + "num_tokens": 137250248.0, + "step": 5370 + }, + { + "epoch": 0.589830880737975, + "grad_norm": 2.021702527999878, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.7066731452941895, + "num_tokens": 137279607.0, + "step": 5371 + }, + { + "epoch": 0.5899406984405886, + "grad_norm": 2.914135217666626, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7256278991699219, + "num_tokens": 137294900.0, + "step": 5372 + }, + { + "epoch": 0.5900505161432023, + "grad_norm": 2.0817315578460693, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6965664029121399, + "num_tokens": 137324756.0, + "step": 5373 + }, + { + "epoch": 0.5901603338458159, + "grad_norm": 2.075937032699585, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7242046594619751, + "num_tokens": 137350849.0, + "step": 5374 + }, + { + "epoch": 0.5902701515484297, + "grad_norm": 2.1639163494110107, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.6966185569763184, + "num_tokens": 137378892.0, + "step": 5375 + }, + { + "epoch": 0.5903799692510433, + "grad_norm": 2.309326410293579, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6970471143722534, + "num_tokens": 137404685.0, + "step": 5376 + }, + { + "epoch": 0.590489786953657, + "grad_norm": 2.0321402549743652, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7082158327102661, + "num_tokens": 137432087.0, + "step": 5377 + }, + { + "epoch": 0.5905996046562706, + "grad_norm": 2.0937771797180176, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.704016923904419, + "num_tokens": 137459407.0, + "step": 5378 + }, + { + "epoch": 0.5907094223588842, + "grad_norm": 2.1169419288635254, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.7011134624481201, + "num_tokens": 137487143.0, + "step": 5379 + }, + { + "epoch": 0.5908192400614979, + "grad_norm": 2.042790651321411, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7147091627120972, + "num_tokens": 137516244.0, + "step": 5380 + }, + { + "epoch": 0.5909290577641115, + "grad_norm": 1.871619462966919, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.7022653818130493, + "num_tokens": 137548083.0, + "step": 5381 + }, + { + "epoch": 0.5910388754667253, + "grad_norm": 2.119415044784546, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.7006174325942993, + "num_tokens": 137575604.0, + "step": 5382 + }, + { + "epoch": 0.5911486931693389, + "grad_norm": 2.1641600131988525, + "learning_rate": 1e-06, + "loss": 1.0107, + "mean_token_accuracy": 0.6954028010368347, + "num_tokens": 137602927.0, + "step": 5383 + }, + { + "epoch": 0.5912585108719526, + "grad_norm": 2.4160499572753906, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7111133337020874, + "num_tokens": 137623974.0, + "step": 5384 + }, + { + "epoch": 0.5913683285745662, + "grad_norm": 2.133829355239868, + "learning_rate": 1e-06, + "loss": 1.0432, + "mean_token_accuracy": 0.6849881410598755, + "num_tokens": 137652946.0, + "step": 5385 + }, + { + "epoch": 0.5914781462771799, + "grad_norm": 2.169700860977173, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.6954527497291565, + "num_tokens": 137679740.0, + "step": 5386 + }, + { + "epoch": 0.5915879639797935, + "grad_norm": 2.2143261432647705, + "learning_rate": 1e-06, + "loss": 1.061, + "mean_token_accuracy": 0.6793835163116455, + "num_tokens": 137703595.0, + "step": 5387 + }, + { + "epoch": 0.5916977816824072, + "grad_norm": 2.019361972808838, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6941823959350586, + "num_tokens": 137733492.0, + "step": 5388 + }, + { + "epoch": 0.5918075993850208, + "grad_norm": 2.318848133087158, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.6975747346878052, + "num_tokens": 137757313.0, + "step": 5389 + }, + { + "epoch": 0.5919174170876346, + "grad_norm": 2.3386363983154297, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.706194281578064, + "num_tokens": 137781084.0, + "step": 5390 + }, + { + "epoch": 0.5920272347902482, + "grad_norm": 2.191598653793335, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7149968147277832, + "num_tokens": 137806038.0, + "step": 5391 + }, + { + "epoch": 0.5921370524928619, + "grad_norm": 2.1245124340057373, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7108228802680969, + "num_tokens": 137832852.0, + "step": 5392 + }, + { + "epoch": 0.5922468701954755, + "grad_norm": 1.9229226112365723, + "learning_rate": 1e-06, + "loss": 1.0457, + "mean_token_accuracy": 0.686616837978363, + "num_tokens": 137865293.0, + "step": 5393 + }, + { + "epoch": 0.5923566878980892, + "grad_norm": 2.114345073699951, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7024505734443665, + "num_tokens": 137891231.0, + "step": 5394 + }, + { + "epoch": 0.5924665056007028, + "grad_norm": 2.091761827468872, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7085679769515991, + "num_tokens": 137919454.0, + "step": 5395 + }, + { + "epoch": 0.5925763233033164, + "grad_norm": 2.4190919399261475, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.6956168413162231, + "num_tokens": 137941619.0, + "step": 5396 + }, + { + "epoch": 0.5926861410059302, + "grad_norm": 2.33449649810791, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7157808542251587, + "num_tokens": 137964841.0, + "step": 5397 + }, + { + "epoch": 0.5927959587085438, + "grad_norm": 2.1701602935791016, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6876338124275208, + "num_tokens": 137990311.0, + "step": 5398 + }, + { + "epoch": 0.5929057764111575, + "grad_norm": 2.1215908527374268, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7121777534484863, + "num_tokens": 138017260.0, + "step": 5399 + }, + { + "epoch": 0.5930155941137711, + "grad_norm": 2.424516439437866, + "learning_rate": 1e-06, + "loss": 1.0434, + "mean_token_accuracy": 0.6924723386764526, + "num_tokens": 138038310.0, + "step": 5400 + }, + { + "epoch": 0.5931254118163848, + "grad_norm": 2.375490427017212, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7185412049293518, + "num_tokens": 138061244.0, + "step": 5401 + }, + { + "epoch": 0.5932352295189984, + "grad_norm": 2.0867202281951904, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7064347267150879, + "num_tokens": 138088224.0, + "step": 5402 + }, + { + "epoch": 0.5933450472216121, + "grad_norm": 2.4022409915924072, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7124600410461426, + "num_tokens": 138109829.0, + "step": 5403 + }, + { + "epoch": 0.5934548649242258, + "grad_norm": 2.020941972732544, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7210696339607239, + "num_tokens": 138139543.0, + "step": 5404 + }, + { + "epoch": 0.5935646826268395, + "grad_norm": 2.1806650161743164, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6982930898666382, + "num_tokens": 138166481.0, + "step": 5405 + }, + { + "epoch": 0.5936745003294531, + "grad_norm": 2.3836510181427, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.7032301425933838, + "num_tokens": 138189247.0, + "step": 5406 + }, + { + "epoch": 0.5937843180320668, + "grad_norm": 2.3079464435577393, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7195724248886108, + "num_tokens": 138211598.0, + "step": 5407 + }, + { + "epoch": 0.5938941357346804, + "grad_norm": 2.2259576320648193, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7085574269294739, + "num_tokens": 138235480.0, + "step": 5408 + }, + { + "epoch": 0.5940039534372941, + "grad_norm": 2.0691869258880615, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7014778852462769, + "num_tokens": 138262608.0, + "step": 5409 + }, + { + "epoch": 0.5941137711399077, + "grad_norm": 2.243184804916382, + "learning_rate": 1e-06, + "loss": 1.0445, + "mean_token_accuracy": 0.6811659932136536, + "num_tokens": 138286746.0, + "step": 5410 + }, + { + "epoch": 0.5942235888425215, + "grad_norm": 2.3537232875823975, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7230703830718994, + "num_tokens": 138305856.0, + "step": 5411 + }, + { + "epoch": 0.5943334065451351, + "grad_norm": 1.9981813430786133, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7079118490219116, + "num_tokens": 138333008.0, + "step": 5412 + }, + { + "epoch": 0.5944432242477488, + "grad_norm": 2.0857861042022705, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.6862979531288147, + "num_tokens": 138360528.0, + "step": 5413 + }, + { + "epoch": 0.5945530419503624, + "grad_norm": 1.97495698928833, + "learning_rate": 1e-06, + "loss": 1.0373, + "mean_token_accuracy": 0.6899513006210327, + "num_tokens": 138391043.0, + "step": 5414 + }, + { + "epoch": 0.594662859652976, + "grad_norm": 2.3295648097991943, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.713892936706543, + "num_tokens": 138414157.0, + "step": 5415 + }, + { + "epoch": 0.5947726773555897, + "grad_norm": 2.422833204269409, + "learning_rate": 1e-06, + "loss": 1.0552, + "mean_token_accuracy": 0.6925188302993774, + "num_tokens": 138438607.0, + "step": 5416 + }, + { + "epoch": 0.5948824950582033, + "grad_norm": 2.108868360519409, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7184101939201355, + "num_tokens": 138466781.0, + "step": 5417 + }, + { + "epoch": 0.5949923127608171, + "grad_norm": 2.233900785446167, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7083629369735718, + "num_tokens": 138490074.0, + "step": 5418 + }, + { + "epoch": 0.5951021304634307, + "grad_norm": 1.9759331941604614, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7399367690086365, + "num_tokens": 138519313.0, + "step": 5419 + }, + { + "epoch": 0.5952119481660444, + "grad_norm": 2.1203970909118652, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.709947943687439, + "num_tokens": 138545232.0, + "step": 5420 + }, + { + "epoch": 0.595321765868658, + "grad_norm": 1.990991234779358, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.7008717060089111, + "num_tokens": 138574784.0, + "step": 5421 + }, + { + "epoch": 0.5954315835712717, + "grad_norm": 2.2105233669281006, + "learning_rate": 1e-06, + "loss": 1.0734, + "mean_token_accuracy": 0.6832815408706665, + "num_tokens": 138600896.0, + "step": 5422 + }, + { + "epoch": 0.5955414012738853, + "grad_norm": 2.5325706005096436, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6994076371192932, + "num_tokens": 138620515.0, + "step": 5423 + }, + { + "epoch": 0.595651218976499, + "grad_norm": 2.168677568435669, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.716768205165863, + "num_tokens": 138645664.0, + "step": 5424 + }, + { + "epoch": 0.5957610366791126, + "grad_norm": 2.4417951107025146, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.7009086608886719, + "num_tokens": 138667520.0, + "step": 5425 + }, + { + "epoch": 0.5958708543817264, + "grad_norm": 2.0520553588867188, + "learning_rate": 1e-06, + "loss": 1.0195, + "mean_token_accuracy": 0.6937882900238037, + "num_tokens": 138695914.0, + "step": 5426 + }, + { + "epoch": 0.59598067208434, + "grad_norm": 2.19313907623291, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7147538661956787, + "num_tokens": 138720699.0, + "step": 5427 + }, + { + "epoch": 0.5960904897869537, + "grad_norm": 2.1149091720581055, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7044700384140015, + "num_tokens": 138749805.0, + "step": 5428 + }, + { + "epoch": 0.5962003074895673, + "grad_norm": 2.1223971843719482, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.6901217699050903, + "num_tokens": 138777463.0, + "step": 5429 + }, + { + "epoch": 0.596310125192181, + "grad_norm": 2.443835973739624, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7123374938964844, + "num_tokens": 138800211.0, + "step": 5430 + }, + { + "epoch": 0.5964199428947946, + "grad_norm": 2.3497536182403564, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7047187685966492, + "num_tokens": 138822308.0, + "step": 5431 + }, + { + "epoch": 0.5965297605974083, + "grad_norm": 2.133669137954712, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7205413579940796, + "num_tokens": 138850484.0, + "step": 5432 + }, + { + "epoch": 0.596639578300022, + "grad_norm": 2.2421422004699707, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.7008097767829895, + "num_tokens": 138875397.0, + "step": 5433 + }, + { + "epoch": 0.5967493960026357, + "grad_norm": 2.1430938243865967, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.715735673904419, + "num_tokens": 138901879.0, + "step": 5434 + }, + { + "epoch": 0.5968592137052493, + "grad_norm": 2.0300865173339844, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7227108478546143, + "num_tokens": 138928396.0, + "step": 5435 + }, + { + "epoch": 0.596969031407863, + "grad_norm": 2.2968149185180664, + "learning_rate": 1e-06, + "loss": 1.0239, + "mean_token_accuracy": 0.6857172846794128, + "num_tokens": 138953818.0, + "step": 5436 + }, + { + "epoch": 0.5970788491104766, + "grad_norm": 2.294616222381592, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7103809714317322, + "num_tokens": 138977738.0, + "step": 5437 + }, + { + "epoch": 0.5971886668130902, + "grad_norm": 2.195504903793335, + "learning_rate": 1e-06, + "loss": 1.037, + "mean_token_accuracy": 0.6919108629226685, + "num_tokens": 139002758.0, + "step": 5438 + }, + { + "epoch": 0.5972984845157039, + "grad_norm": 2.195185661315918, + "learning_rate": 1e-06, + "loss": 1.1148, + "mean_token_accuracy": 0.6657540798187256, + "num_tokens": 139030325.0, + "step": 5439 + }, + { + "epoch": 0.5974083022183176, + "grad_norm": 2.244047164916992, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.697629451751709, + "num_tokens": 139055735.0, + "step": 5440 + }, + { + "epoch": 0.5975181199209313, + "grad_norm": 2.1475067138671875, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7077404260635376, + "num_tokens": 139081513.0, + "step": 5441 + }, + { + "epoch": 0.5976279376235449, + "grad_norm": 2.2790026664733887, + "learning_rate": 1e-06, + "loss": 1.0425, + "mean_token_accuracy": 0.6883483529090881, + "num_tokens": 139105458.0, + "step": 5442 + }, + { + "epoch": 0.5977377553261586, + "grad_norm": 2.072763442993164, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6958900690078735, + "num_tokens": 139131793.0, + "step": 5443 + }, + { + "epoch": 0.5978475730287722, + "grad_norm": 2.360793352127075, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7023825645446777, + "num_tokens": 139154007.0, + "step": 5444 + }, + { + "epoch": 0.5979573907313859, + "grad_norm": 2.2047502994537354, + "learning_rate": 1e-06, + "loss": 1.0368, + "mean_token_accuracy": 0.6846777200698853, + "num_tokens": 139179615.0, + "step": 5445 + }, + { + "epoch": 0.5980672084339995, + "grad_norm": 1.9076684713363647, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.6895367503166199, + "num_tokens": 139213492.0, + "step": 5446 + }, + { + "epoch": 0.5981770261366133, + "grad_norm": 2.220804452896118, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6777309775352478, + "num_tokens": 139238766.0, + "step": 5447 + }, + { + "epoch": 0.5982868438392269, + "grad_norm": 2.185032367706299, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7014877200126648, + "num_tokens": 139263260.0, + "step": 5448 + }, + { + "epoch": 0.5983966615418406, + "grad_norm": 2.1036243438720703, + "learning_rate": 1e-06, + "loss": 1.023, + "mean_token_accuracy": 0.6914762854576111, + "num_tokens": 139291891.0, + "step": 5449 + }, + { + "epoch": 0.5985064792444542, + "grad_norm": 2.4592342376708984, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7159557938575745, + "num_tokens": 139312408.0, + "step": 5450 + }, + { + "epoch": 0.5986162969470679, + "grad_norm": 2.3810031414031982, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.695040225982666, + "num_tokens": 139334400.0, + "step": 5451 + }, + { + "epoch": 0.5987261146496815, + "grad_norm": 2.3869199752807617, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7204530239105225, + "num_tokens": 139354883.0, + "step": 5452 + }, + { + "epoch": 0.5988359323522952, + "grad_norm": 2.023581027984619, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6994174718856812, + "num_tokens": 139386448.0, + "step": 5453 + }, + { + "epoch": 0.5989457500549088, + "grad_norm": 1.9177680015563965, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.6978455185890198, + "num_tokens": 139416985.0, + "step": 5454 + }, + { + "epoch": 0.5990555677575226, + "grad_norm": 2.1373941898345947, + "learning_rate": 1e-06, + "loss": 1.0392, + "mean_token_accuracy": 0.6889805197715759, + "num_tokens": 139444265.0, + "step": 5455 + }, + { + "epoch": 0.5991653854601362, + "grad_norm": 2.0801665782928467, + "learning_rate": 1e-06, + "loss": 1.0737, + "mean_token_accuracy": 0.6805123686790466, + "num_tokens": 139471572.0, + "step": 5456 + }, + { + "epoch": 0.5992752031627498, + "grad_norm": 2.0222036838531494, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.712124764919281, + "num_tokens": 139499810.0, + "step": 5457 + }, + { + "epoch": 0.5993850208653635, + "grad_norm": 2.358072519302368, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7078285813331604, + "num_tokens": 139523178.0, + "step": 5458 + }, + { + "epoch": 0.5994948385679771, + "grad_norm": 1.9711129665374756, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.7014869451522827, + "num_tokens": 139552522.0, + "step": 5459 + }, + { + "epoch": 0.5996046562705908, + "grad_norm": 2.145177125930786, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7260847091674805, + "num_tokens": 139576774.0, + "step": 5460 + }, + { + "epoch": 0.5997144739732044, + "grad_norm": 1.8915650844573975, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.6978452801704407, + "num_tokens": 139607657.0, + "step": 5461 + }, + { + "epoch": 0.5998242916758182, + "grad_norm": 2.177297353744507, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.6921370625495911, + "num_tokens": 139635936.0, + "step": 5462 + }, + { + "epoch": 0.5999341093784318, + "grad_norm": 2.4001736640930176, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7103123664855957, + "num_tokens": 139658750.0, + "step": 5463 + }, + { + "epoch": 0.6000439270810455, + "grad_norm": 2.3743550777435303, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.7036010026931763, + "num_tokens": 139680163.0, + "step": 5464 + }, + { + "epoch": 0.6001537447836591, + "grad_norm": 2.2922418117523193, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6940711140632629, + "num_tokens": 139706166.0, + "step": 5465 + }, + { + "epoch": 0.6002635624862728, + "grad_norm": 2.1303861141204834, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7239646911621094, + "num_tokens": 139732942.0, + "step": 5466 + }, + { + "epoch": 0.6003733801888864, + "grad_norm": 1.9978018999099731, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7238330841064453, + "num_tokens": 139763938.0, + "step": 5467 + }, + { + "epoch": 0.6004831978915001, + "grad_norm": 2.546170949935913, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7236598134040833, + "num_tokens": 139783383.0, + "step": 5468 + }, + { + "epoch": 0.6005930155941138, + "grad_norm": 2.3421225547790527, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6876004338264465, + "num_tokens": 139808474.0, + "step": 5469 + }, + { + "epoch": 0.6007028332967275, + "grad_norm": 2.1042590141296387, + "learning_rate": 1e-06, + "loss": 1.093, + "mean_token_accuracy": 0.6728971600532532, + "num_tokens": 139835648.0, + "step": 5470 + }, + { + "epoch": 0.6008126509993411, + "grad_norm": 2.3728411197662354, + "learning_rate": 1e-06, + "loss": 1.0415, + "mean_token_accuracy": 0.6978983879089355, + "num_tokens": 139859515.0, + "step": 5471 + }, + { + "epoch": 0.6009224687019548, + "grad_norm": 2.1159634590148926, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7119944095611572, + "num_tokens": 139885438.0, + "step": 5472 + }, + { + "epoch": 0.6010322864045684, + "grad_norm": 2.3178253173828125, + "learning_rate": 1e-06, + "loss": 1.06, + "mean_token_accuracy": 0.6813931465148926, + "num_tokens": 139912304.0, + "step": 5473 + }, + { + "epoch": 0.601142104107182, + "grad_norm": 2.109975576400757, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.709976315498352, + "num_tokens": 139937479.0, + "step": 5474 + }, + { + "epoch": 0.6012519218097957, + "grad_norm": 2.428147077560425, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6955729722976685, + "num_tokens": 139959450.0, + "step": 5475 + }, + { + "epoch": 0.6013617395124095, + "grad_norm": 2.44014048576355, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7171547412872314, + "num_tokens": 139980978.0, + "step": 5476 + }, + { + "epoch": 0.6014715572150231, + "grad_norm": 2.4040796756744385, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7064854502677917, + "num_tokens": 140003410.0, + "step": 5477 + }, + { + "epoch": 0.6015813749176367, + "grad_norm": 2.290086030960083, + "learning_rate": 1e-06, + "loss": 1.0558, + "mean_token_accuracy": 0.6834563612937927, + "num_tokens": 140026290.0, + "step": 5478 + }, + { + "epoch": 0.6016911926202504, + "grad_norm": 2.0637001991271973, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6932682991027832, + "num_tokens": 140055234.0, + "step": 5479 + }, + { + "epoch": 0.601801010322864, + "grad_norm": 2.391042709350586, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6918174028396606, + "num_tokens": 140076271.0, + "step": 5480 + }, + { + "epoch": 0.6019108280254777, + "grad_norm": 2.1587424278259277, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7104191780090332, + "num_tokens": 140100922.0, + "step": 5481 + }, + { + "epoch": 0.6020206457280913, + "grad_norm": 2.3211658000946045, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7191548347473145, + "num_tokens": 140122906.0, + "step": 5482 + }, + { + "epoch": 0.602130463430705, + "grad_norm": 2.4533627033233643, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7144806385040283, + "num_tokens": 140144458.0, + "step": 5483 + }, + { + "epoch": 0.6022402811333187, + "grad_norm": 2.8739702701568604, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7266561985015869, + "num_tokens": 140160214.0, + "step": 5484 + }, + { + "epoch": 0.6023500988359324, + "grad_norm": 2.265113592147827, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.6952717304229736, + "num_tokens": 140187160.0, + "step": 5485 + }, + { + "epoch": 0.602459916538546, + "grad_norm": 2.2163190841674805, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7008955478668213, + "num_tokens": 140211462.0, + "step": 5486 + }, + { + "epoch": 0.6025697342411597, + "grad_norm": 2.3473663330078125, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7144814729690552, + "num_tokens": 140233527.0, + "step": 5487 + }, + { + "epoch": 0.6026795519437733, + "grad_norm": 2.676182508468628, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7413653135299683, + "num_tokens": 140251068.0, + "step": 5488 + }, + { + "epoch": 0.602789369646387, + "grad_norm": 2.1481006145477295, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.698371410369873, + "num_tokens": 140277743.0, + "step": 5489 + }, + { + "epoch": 0.6028991873490006, + "grad_norm": 2.3189895153045654, + "learning_rate": 1e-06, + "loss": 1.0751, + "mean_token_accuracy": 0.6831545233726501, + "num_tokens": 140305324.0, + "step": 5490 + }, + { + "epoch": 0.6030090050516144, + "grad_norm": 2.377725839614868, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7342641949653625, + "num_tokens": 140326526.0, + "step": 5491 + }, + { + "epoch": 0.603118822754228, + "grad_norm": 2.1838467121124268, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7057381868362427, + "num_tokens": 140352609.0, + "step": 5492 + }, + { + "epoch": 0.6032286404568417, + "grad_norm": 2.2487456798553467, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7064677476882935, + "num_tokens": 140376352.0, + "step": 5493 + }, + { + "epoch": 0.6033384581594553, + "grad_norm": 2.055039405822754, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7188069820404053, + "num_tokens": 140404114.0, + "step": 5494 + }, + { + "epoch": 0.603448275862069, + "grad_norm": 1.9618712663650513, + "learning_rate": 1e-06, + "loss": 1.0788, + "mean_token_accuracy": 0.6848644018173218, + "num_tokens": 140436814.0, + "step": 5495 + }, + { + "epoch": 0.6035580935646826, + "grad_norm": 2.1862173080444336, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.6866172552108765, + "num_tokens": 140462180.0, + "step": 5496 + }, + { + "epoch": 0.6036679112672962, + "grad_norm": 2.2977142333984375, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7173616886138916, + "num_tokens": 140484409.0, + "step": 5497 + }, + { + "epoch": 0.60377772896991, + "grad_norm": 2.423539161682129, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7232407331466675, + "num_tokens": 140505284.0, + "step": 5498 + }, + { + "epoch": 0.6038875466725236, + "grad_norm": 2.185957193374634, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7000482082366943, + "num_tokens": 140531437.0, + "step": 5499 + }, + { + "epoch": 0.6039973643751373, + "grad_norm": 2.110915184020996, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6884382963180542, + "num_tokens": 140557486.0, + "step": 5500 + }, + { + "epoch": 0.6041071820777509, + "grad_norm": 2.0508475303649902, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7112554311752319, + "num_tokens": 140584338.0, + "step": 5501 + }, + { + "epoch": 0.6042169997803646, + "grad_norm": 1.9515842199325562, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7134189009666443, + "num_tokens": 140613606.0, + "step": 5502 + }, + { + "epoch": 0.6043268174829782, + "grad_norm": 2.1925740242004395, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.718012809753418, + "num_tokens": 140637463.0, + "step": 5503 + }, + { + "epoch": 0.6044366351855919, + "grad_norm": 2.192220449447632, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.70817631483078, + "num_tokens": 140662083.0, + "step": 5504 + }, + { + "epoch": 0.6045464528882056, + "grad_norm": 1.9721277952194214, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.6906291246414185, + "num_tokens": 140696052.0, + "step": 5505 + }, + { + "epoch": 0.6046562705908193, + "grad_norm": 2.266533851623535, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6974946856498718, + "num_tokens": 140722189.0, + "step": 5506 + }, + { + "epoch": 0.6047660882934329, + "grad_norm": 2.0967354774475098, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7031919360160828, + "num_tokens": 140749257.0, + "step": 5507 + }, + { + "epoch": 0.6048759059960466, + "grad_norm": 2.2015578746795654, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7199525237083435, + "num_tokens": 140774261.0, + "step": 5508 + }, + { + "epoch": 0.6049857236986602, + "grad_norm": 2.24394154548645, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7164318561553955, + "num_tokens": 140798437.0, + "step": 5509 + }, + { + "epoch": 0.6050955414012739, + "grad_norm": 2.085049629211426, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.6961609125137329, + "num_tokens": 140826190.0, + "step": 5510 + }, + { + "epoch": 0.6052053591038875, + "grad_norm": 2.017604112625122, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.724005937576294, + "num_tokens": 140856108.0, + "step": 5511 + }, + { + "epoch": 0.6053151768065012, + "grad_norm": 2.3279597759246826, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7073765397071838, + "num_tokens": 140877423.0, + "step": 5512 + }, + { + "epoch": 0.6054249945091149, + "grad_norm": 2.12092661857605, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.698147714138031, + "num_tokens": 140905788.0, + "step": 5513 + }, + { + "epoch": 0.6055348122117286, + "grad_norm": 2.574526071548462, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7114098072052002, + "num_tokens": 140926051.0, + "step": 5514 + }, + { + "epoch": 0.6056446299143422, + "grad_norm": 2.1193463802337646, + "learning_rate": 1e-06, + "loss": 1.0279, + "mean_token_accuracy": 0.6853336095809937, + "num_tokens": 140955234.0, + "step": 5515 + }, + { + "epoch": 0.6057544476169558, + "grad_norm": 2.406550884246826, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7144925594329834, + "num_tokens": 140976715.0, + "step": 5516 + }, + { + "epoch": 0.6058642653195695, + "grad_norm": 2.1578586101531982, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7006995677947998, + "num_tokens": 141003520.0, + "step": 5517 + }, + { + "epoch": 0.6059740830221831, + "grad_norm": 2.2829103469848633, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7069409489631653, + "num_tokens": 141031630.0, + "step": 5518 + }, + { + "epoch": 0.6060839007247968, + "grad_norm": 2.097001075744629, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6905512809753418, + "num_tokens": 141059818.0, + "step": 5519 + }, + { + "epoch": 0.6061937184274105, + "grad_norm": 2.2629528045654297, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6985193490982056, + "num_tokens": 141088360.0, + "step": 5520 + }, + { + "epoch": 0.6063035361300242, + "grad_norm": 2.3781845569610596, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.6993591785430908, + "num_tokens": 141109087.0, + "step": 5521 + }, + { + "epoch": 0.6064133538326378, + "grad_norm": 2.2221360206604004, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6926025152206421, + "num_tokens": 141133558.0, + "step": 5522 + }, + { + "epoch": 0.6065231715352515, + "grad_norm": 2.4088284969329834, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7216717004776001, + "num_tokens": 141155119.0, + "step": 5523 + }, + { + "epoch": 0.6066329892378651, + "grad_norm": 2.0144684314727783, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7071810960769653, + "num_tokens": 141184457.0, + "step": 5524 + }, + { + "epoch": 0.6067428069404788, + "grad_norm": 1.9688589572906494, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7052522897720337, + "num_tokens": 141216163.0, + "step": 5525 + }, + { + "epoch": 0.6068526246430924, + "grad_norm": 2.1974682807922363, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6877439618110657, + "num_tokens": 141242389.0, + "step": 5526 + }, + { + "epoch": 0.6069624423457062, + "grad_norm": 2.2192225456237793, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7067227959632874, + "num_tokens": 141267686.0, + "step": 5527 + }, + { + "epoch": 0.6070722600483198, + "grad_norm": 2.4233782291412354, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.6980222463607788, + "num_tokens": 141290740.0, + "step": 5528 + }, + { + "epoch": 0.6071820777509335, + "grad_norm": 2.4039127826690674, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7154417037963867, + "num_tokens": 141311679.0, + "step": 5529 + }, + { + "epoch": 0.6072918954535471, + "grad_norm": 1.9566493034362793, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6984497308731079, + "num_tokens": 141342410.0, + "step": 5530 + }, + { + "epoch": 0.6074017131561608, + "grad_norm": 2.3197505474090576, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7096267938613892, + "num_tokens": 141363837.0, + "step": 5531 + }, + { + "epoch": 0.6075115308587744, + "grad_norm": 2.2424702644348145, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.7004059553146362, + "num_tokens": 141388321.0, + "step": 5532 + }, + { + "epoch": 0.607621348561388, + "grad_norm": 2.1548197269439697, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.698276162147522, + "num_tokens": 141415523.0, + "step": 5533 + }, + { + "epoch": 0.6077311662640018, + "grad_norm": 2.2837002277374268, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7085293531417847, + "num_tokens": 141437784.0, + "step": 5534 + }, + { + "epoch": 0.6078409839666155, + "grad_norm": 2.1442768573760986, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7012842297554016, + "num_tokens": 141464057.0, + "step": 5535 + }, + { + "epoch": 0.6079508016692291, + "grad_norm": 2.1794657707214355, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7101831436157227, + "num_tokens": 141492167.0, + "step": 5536 + }, + { + "epoch": 0.6080606193718427, + "grad_norm": 2.0130178928375244, + "learning_rate": 1e-06, + "loss": 1.0195, + "mean_token_accuracy": 0.6914089918136597, + "num_tokens": 141520580.0, + "step": 5537 + }, + { + "epoch": 0.6081704370744564, + "grad_norm": 2.2948946952819824, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7247788906097412, + "num_tokens": 141543013.0, + "step": 5538 + }, + { + "epoch": 0.60828025477707, + "grad_norm": 2.1442995071411133, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7197014093399048, + "num_tokens": 141572724.0, + "step": 5539 + }, + { + "epoch": 0.6083900724796837, + "grad_norm": 2.4127001762390137, + "learning_rate": 1e-06, + "loss": 1.0721, + "mean_token_accuracy": 0.6779823303222656, + "num_tokens": 141596336.0, + "step": 5540 + }, + { + "epoch": 0.6084998901822973, + "grad_norm": 1.9220564365386963, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7021021842956543, + "num_tokens": 141627252.0, + "step": 5541 + }, + { + "epoch": 0.6086097078849111, + "grad_norm": 2.0363783836364746, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6942340731620789, + "num_tokens": 141657520.0, + "step": 5542 + }, + { + "epoch": 0.6087195255875247, + "grad_norm": 2.1139118671417236, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7068753242492676, + "num_tokens": 141686503.0, + "step": 5543 + }, + { + "epoch": 0.6088293432901384, + "grad_norm": 2.3533172607421875, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.715593695640564, + "num_tokens": 141709243.0, + "step": 5544 + }, + { + "epoch": 0.608939160992752, + "grad_norm": 1.9423805475234985, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.7032137513160706, + "num_tokens": 141740055.0, + "step": 5545 + }, + { + "epoch": 0.6090489786953657, + "grad_norm": 2.125810384750366, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.6940324306488037, + "num_tokens": 141767605.0, + "step": 5546 + }, + { + "epoch": 0.6091587963979793, + "grad_norm": 2.085108518600464, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.7032862901687622, + "num_tokens": 141795447.0, + "step": 5547 + }, + { + "epoch": 0.609268614100593, + "grad_norm": 2.241689443588257, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7209835052490234, + "num_tokens": 141818895.0, + "step": 5548 + }, + { + "epoch": 0.6093784318032067, + "grad_norm": 2.735106945037842, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7354146242141724, + "num_tokens": 141835283.0, + "step": 5549 + }, + { + "epoch": 0.6094882495058204, + "grad_norm": 2.2676734924316406, + "learning_rate": 1e-06, + "loss": 1.0575, + "mean_token_accuracy": 0.6866618990898132, + "num_tokens": 141861310.0, + "step": 5550 + }, + { + "epoch": 0.609598067208434, + "grad_norm": 1.9897745847702026, + "learning_rate": 1e-06, + "loss": 1.1016, + "mean_token_accuracy": 0.6693753004074097, + "num_tokens": 141893509.0, + "step": 5551 + }, + { + "epoch": 0.6097078849110477, + "grad_norm": 1.9143694639205933, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.696686327457428, + "num_tokens": 141927505.0, + "step": 5552 + }, + { + "epoch": 0.6098177026136613, + "grad_norm": 2.3598315715789795, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.7004372477531433, + "num_tokens": 141951199.0, + "step": 5553 + }, + { + "epoch": 0.609927520316275, + "grad_norm": 2.0410685539245605, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6950381994247437, + "num_tokens": 141981039.0, + "step": 5554 + }, + { + "epoch": 0.6100373380188886, + "grad_norm": 2.2452869415283203, + "learning_rate": 1e-06, + "loss": 1.0448, + "mean_token_accuracy": 0.687710165977478, + "num_tokens": 142007391.0, + "step": 5555 + }, + { + "epoch": 0.6101471557215024, + "grad_norm": 2.3731613159179688, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7085933685302734, + "num_tokens": 142030160.0, + "step": 5556 + }, + { + "epoch": 0.610256973424116, + "grad_norm": 2.4565489292144775, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7381277084350586, + "num_tokens": 142048550.0, + "step": 5557 + }, + { + "epoch": 0.6103667911267296, + "grad_norm": 2.2474026679992676, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6981908082962036, + "num_tokens": 142075904.0, + "step": 5558 + }, + { + "epoch": 0.6104766088293433, + "grad_norm": 2.084285020828247, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.691826581954956, + "num_tokens": 142105529.0, + "step": 5559 + }, + { + "epoch": 0.6105864265319569, + "grad_norm": 2.4913148880004883, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.7109134197235107, + "num_tokens": 142128432.0, + "step": 5560 + }, + { + "epoch": 0.6106962442345706, + "grad_norm": 2.068370819091797, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7083985805511475, + "num_tokens": 142156027.0, + "step": 5561 + }, + { + "epoch": 0.6108060619371842, + "grad_norm": 2.2639524936676025, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.6961469650268555, + "num_tokens": 142181717.0, + "step": 5562 + }, + { + "epoch": 0.610915879639798, + "grad_norm": 2.214768171310425, + "learning_rate": 1e-06, + "loss": 1.0465, + "mean_token_accuracy": 0.6815125942230225, + "num_tokens": 142208007.0, + "step": 5563 + }, + { + "epoch": 0.6110256973424116, + "grad_norm": 2.3924925327301025, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.6942453980445862, + "num_tokens": 142231330.0, + "step": 5564 + }, + { + "epoch": 0.6111355150450253, + "grad_norm": 2.1885852813720703, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6912425756454468, + "num_tokens": 142256829.0, + "step": 5565 + }, + { + "epoch": 0.6112453327476389, + "grad_norm": 2.209017515182495, + "learning_rate": 1e-06, + "loss": 1.0448, + "mean_token_accuracy": 0.6814030408859253, + "num_tokens": 142283365.0, + "step": 5566 + }, + { + "epoch": 0.6113551504502526, + "grad_norm": 2.059908866882324, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7204322814941406, + "num_tokens": 142310322.0, + "step": 5567 + }, + { + "epoch": 0.6114649681528662, + "grad_norm": 2.281255006790161, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7169466018676758, + "num_tokens": 142333768.0, + "step": 5568 + }, + { + "epoch": 0.6115747858554799, + "grad_norm": 2.4134678840637207, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7178150415420532, + "num_tokens": 142354997.0, + "step": 5569 + }, + { + "epoch": 0.6116846035580935, + "grad_norm": 1.922911524772644, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7051517367362976, + "num_tokens": 142387609.0, + "step": 5570 + }, + { + "epoch": 0.6117944212607073, + "grad_norm": 1.8322219848632812, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.6984938383102417, + "num_tokens": 142422144.0, + "step": 5571 + }, + { + "epoch": 0.6119042389633209, + "grad_norm": 2.036072254180908, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7309988141059875, + "num_tokens": 142450299.0, + "step": 5572 + }, + { + "epoch": 0.6120140566659346, + "grad_norm": 2.246371030807495, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7137981057167053, + "num_tokens": 142474826.0, + "step": 5573 + }, + { + "epoch": 0.6121238743685482, + "grad_norm": 2.152480125427246, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.702646791934967, + "num_tokens": 142501127.0, + "step": 5574 + }, + { + "epoch": 0.6122336920711619, + "grad_norm": 2.3621113300323486, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6916357278823853, + "num_tokens": 142525078.0, + "step": 5575 + }, + { + "epoch": 0.6123435097737755, + "grad_norm": 2.5759177207946777, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7121622562408447, + "num_tokens": 142544476.0, + "step": 5576 + }, + { + "epoch": 0.6124533274763891, + "grad_norm": 2.499511957168579, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.6998012661933899, + "num_tokens": 142564956.0, + "step": 5577 + }, + { + "epoch": 0.6125631451790029, + "grad_norm": 2.062917947769165, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7245840430259705, + "num_tokens": 142591281.0, + "step": 5578 + }, + { + "epoch": 0.6126729628816165, + "grad_norm": 2.050795078277588, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7163405418395996, + "num_tokens": 142620684.0, + "step": 5579 + }, + { + "epoch": 0.6127827805842302, + "grad_norm": 2.20650315284729, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.6997973918914795, + "num_tokens": 142647310.0, + "step": 5580 + }, + { + "epoch": 0.6128925982868438, + "grad_norm": 1.9099483489990234, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.6868754625320435, + "num_tokens": 142679815.0, + "step": 5581 + }, + { + "epoch": 0.6130024159894575, + "grad_norm": 2.1522467136383057, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.6993408203125, + "num_tokens": 142705563.0, + "step": 5582 + }, + { + "epoch": 0.6131122336920711, + "grad_norm": 2.229079484939575, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7210617661476135, + "num_tokens": 142729940.0, + "step": 5583 + }, + { + "epoch": 0.6132220513946848, + "grad_norm": 2.3153841495513916, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.6906788349151611, + "num_tokens": 142752771.0, + "step": 5584 + }, + { + "epoch": 0.6133318690972985, + "grad_norm": 2.15620493888855, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6914725303649902, + "num_tokens": 142780315.0, + "step": 5585 + }, + { + "epoch": 0.6134416867999122, + "grad_norm": 2.1547675132751465, + "learning_rate": 1e-06, + "loss": 1.044, + "mean_token_accuracy": 0.6851038932800293, + "num_tokens": 142806576.0, + "step": 5586 + }, + { + "epoch": 0.6135515045025258, + "grad_norm": 2.0802817344665527, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7142125368118286, + "num_tokens": 142834166.0, + "step": 5587 + }, + { + "epoch": 0.6136613222051395, + "grad_norm": 2.012143135070801, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7053877115249634, + "num_tokens": 142863478.0, + "step": 5588 + }, + { + "epoch": 0.6137711399077531, + "grad_norm": 2.146524667739868, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7069054245948792, + "num_tokens": 142888765.0, + "step": 5589 + }, + { + "epoch": 0.6138809576103668, + "grad_norm": 2.234400749206543, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7289602756500244, + "num_tokens": 142912889.0, + "step": 5590 + }, + { + "epoch": 0.6139907753129804, + "grad_norm": 2.3380532264709473, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7132037878036499, + "num_tokens": 142934357.0, + "step": 5591 + }, + { + "epoch": 0.6141005930155942, + "grad_norm": 2.1892428398132324, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7098652720451355, + "num_tokens": 142959007.0, + "step": 5592 + }, + { + "epoch": 0.6142104107182078, + "grad_norm": 2.1175713539123535, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.6941478252410889, + "num_tokens": 142985183.0, + "step": 5593 + }, + { + "epoch": 0.6143202284208215, + "grad_norm": 2.433908224105835, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7200595140457153, + "num_tokens": 143006345.0, + "step": 5594 + }, + { + "epoch": 0.6144300461234351, + "grad_norm": 2.3806629180908203, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6964619755744934, + "num_tokens": 143030937.0, + "step": 5595 + }, + { + "epoch": 0.6145398638260487, + "grad_norm": 2.1851108074188232, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6887437105178833, + "num_tokens": 143057322.0, + "step": 5596 + }, + { + "epoch": 0.6146496815286624, + "grad_norm": 2.286766290664673, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7195113897323608, + "num_tokens": 143079593.0, + "step": 5597 + }, + { + "epoch": 0.614759499231276, + "grad_norm": 2.2388365268707275, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7325499653816223, + "num_tokens": 143104709.0, + "step": 5598 + }, + { + "epoch": 0.6148693169338898, + "grad_norm": 2.6013946533203125, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7107983827590942, + "num_tokens": 143122752.0, + "step": 5599 + }, + { + "epoch": 0.6149791346365034, + "grad_norm": 2.047455072402954, + "learning_rate": 1e-06, + "loss": 1.0632, + "mean_token_accuracy": 0.6901888847351074, + "num_tokens": 143152305.0, + "step": 5600 + }, + { + "epoch": 0.6150889523391171, + "grad_norm": 2.061081647872925, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7214654684066772, + "num_tokens": 143179685.0, + "step": 5601 + }, + { + "epoch": 0.6151987700417307, + "grad_norm": 2.458331823348999, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7158650159835815, + "num_tokens": 143202904.0, + "step": 5602 + }, + { + "epoch": 0.6153085877443444, + "grad_norm": 2.604228973388672, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7184007167816162, + "num_tokens": 143221307.0, + "step": 5603 + }, + { + "epoch": 0.615418405446958, + "grad_norm": 2.5096206665039062, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.6922112703323364, + "num_tokens": 143243459.0, + "step": 5604 + }, + { + "epoch": 0.6155282231495717, + "grad_norm": 2.0629584789276123, + "learning_rate": 1e-06, + "loss": 1.0381, + "mean_token_accuracy": 0.6822583675384521, + "num_tokens": 143271342.0, + "step": 5605 + }, + { + "epoch": 0.6156380408521853, + "grad_norm": 2.0150585174560547, + "learning_rate": 1e-06, + "loss": 1.0574, + "mean_token_accuracy": 0.6865401268005371, + "num_tokens": 143301188.0, + "step": 5606 + }, + { + "epoch": 0.6157478585547991, + "grad_norm": 2.2890186309814453, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6914398074150085, + "num_tokens": 143325018.0, + "step": 5607 + }, + { + "epoch": 0.6158576762574127, + "grad_norm": 2.4762942790985107, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7102705240249634, + "num_tokens": 143346802.0, + "step": 5608 + }, + { + "epoch": 0.6159674939600264, + "grad_norm": 2.528261184692383, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7132259607315063, + "num_tokens": 143365353.0, + "step": 5609 + }, + { + "epoch": 0.61607731166264, + "grad_norm": 2.1586990356445312, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7186442017555237, + "num_tokens": 143391152.0, + "step": 5610 + }, + { + "epoch": 0.6161871293652537, + "grad_norm": 2.320603132247925, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7046759128570557, + "num_tokens": 143416319.0, + "step": 5611 + }, + { + "epoch": 0.6162969470678673, + "grad_norm": 2.310105800628662, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6950496435165405, + "num_tokens": 143439562.0, + "step": 5612 + }, + { + "epoch": 0.616406764770481, + "grad_norm": 2.3198130130767822, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7293728590011597, + "num_tokens": 143461901.0, + "step": 5613 + }, + { + "epoch": 0.6165165824730947, + "grad_norm": 2.1395065784454346, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7234693765640259, + "num_tokens": 143486394.0, + "step": 5614 + }, + { + "epoch": 0.6166264001757084, + "grad_norm": 2.2745649814605713, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7034157514572144, + "num_tokens": 143511624.0, + "step": 5615 + }, + { + "epoch": 0.616736217878322, + "grad_norm": 2.1436309814453125, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.6912828683853149, + "num_tokens": 143538135.0, + "step": 5616 + }, + { + "epoch": 0.6168460355809356, + "grad_norm": 2.0579493045806885, + "learning_rate": 1e-06, + "loss": 1.0698, + "mean_token_accuracy": 0.6768943071365356, + "num_tokens": 143567566.0, + "step": 5617 + }, + { + "epoch": 0.6169558532835493, + "grad_norm": 2.183696985244751, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7110767960548401, + "num_tokens": 143591596.0, + "step": 5618 + }, + { + "epoch": 0.6170656709861629, + "grad_norm": 1.9437428712844849, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6941966414451599, + "num_tokens": 143624667.0, + "step": 5619 + }, + { + "epoch": 0.6171754886887766, + "grad_norm": 2.347585439682007, + "learning_rate": 1e-06, + "loss": 1.0388, + "mean_token_accuracy": 0.6862552165985107, + "num_tokens": 143650237.0, + "step": 5620 + }, + { + "epoch": 0.6172853063913903, + "grad_norm": 1.9619734287261963, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7083792686462402, + "num_tokens": 143679870.0, + "step": 5621 + }, + { + "epoch": 0.617395124094004, + "grad_norm": 2.0592827796936035, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7026066184043884, + "num_tokens": 143708263.0, + "step": 5622 + }, + { + "epoch": 0.6175049417966176, + "grad_norm": 2.0845091342926025, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7085602283477783, + "num_tokens": 143735412.0, + "step": 5623 + }, + { + "epoch": 0.6176147594992313, + "grad_norm": 2.214384078979492, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7120040655136108, + "num_tokens": 143760426.0, + "step": 5624 + }, + { + "epoch": 0.6177245772018449, + "grad_norm": 2.012511730194092, + "learning_rate": 1e-06, + "loss": 1.073, + "mean_token_accuracy": 0.6770684719085693, + "num_tokens": 143790228.0, + "step": 5625 + }, + { + "epoch": 0.6178343949044586, + "grad_norm": 2.3757965564727783, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7059907913208008, + "num_tokens": 143810128.0, + "step": 5626 + }, + { + "epoch": 0.6179442126070722, + "grad_norm": 2.1966731548309326, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6972032785415649, + "num_tokens": 143836741.0, + "step": 5627 + }, + { + "epoch": 0.618054030309686, + "grad_norm": 2.314328908920288, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7252587080001831, + "num_tokens": 143861180.0, + "step": 5628 + }, + { + "epoch": 0.6181638480122996, + "grad_norm": 2.326490640640259, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7196073532104492, + "num_tokens": 143886477.0, + "step": 5629 + }, + { + "epoch": 0.6182736657149133, + "grad_norm": 2.1424760818481445, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7048385739326477, + "num_tokens": 143913109.0, + "step": 5630 + }, + { + "epoch": 0.6183834834175269, + "grad_norm": 2.2525341510772705, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7067790031433105, + "num_tokens": 143937227.0, + "step": 5631 + }, + { + "epoch": 0.6184933011201406, + "grad_norm": 1.9826691150665283, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6900686025619507, + "num_tokens": 143968354.0, + "step": 5632 + }, + { + "epoch": 0.6186031188227542, + "grad_norm": 1.938635230064392, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7096210718154907, + "num_tokens": 144002207.0, + "step": 5633 + }, + { + "epoch": 0.6187129365253679, + "grad_norm": 2.232991933822632, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.7029990553855896, + "num_tokens": 144026139.0, + "step": 5634 + }, + { + "epoch": 0.6188227542279815, + "grad_norm": 2.1492972373962402, + "learning_rate": 1e-06, + "loss": 1.0633, + "mean_token_accuracy": 0.6884124279022217, + "num_tokens": 144052358.0, + "step": 5635 + }, + { + "epoch": 0.6189325719305953, + "grad_norm": 2.3571255207061768, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7119188904762268, + "num_tokens": 144074685.0, + "step": 5636 + }, + { + "epoch": 0.6190423896332089, + "grad_norm": 2.1233768463134766, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7221560478210449, + "num_tokens": 144101366.0, + "step": 5637 + }, + { + "epoch": 0.6191522073358225, + "grad_norm": 2.1834938526153564, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.6995718479156494, + "num_tokens": 144127123.0, + "step": 5638 + }, + { + "epoch": 0.6192620250384362, + "grad_norm": 2.021801233291626, + "learning_rate": 1e-06, + "loss": 1.039, + "mean_token_accuracy": 0.683962345123291, + "num_tokens": 144156476.0, + "step": 5639 + }, + { + "epoch": 0.6193718427410498, + "grad_norm": 2.2545113563537598, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7054301500320435, + "num_tokens": 144181808.0, + "step": 5640 + }, + { + "epoch": 0.6194816604436635, + "grad_norm": 2.0972259044647217, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7061414122581482, + "num_tokens": 144208192.0, + "step": 5641 + }, + { + "epoch": 0.6195914781462771, + "grad_norm": 2.513796329498291, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6981121301651001, + "num_tokens": 144231812.0, + "step": 5642 + }, + { + "epoch": 0.6197012958488909, + "grad_norm": 2.707245349884033, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7116038799285889, + "num_tokens": 144250285.0, + "step": 5643 + }, + { + "epoch": 0.6198111135515045, + "grad_norm": 2.181880474090576, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6972962617874146, + "num_tokens": 144277433.0, + "step": 5644 + }, + { + "epoch": 0.6199209312541182, + "grad_norm": 2.278134346008301, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6862584948539734, + "num_tokens": 144301751.0, + "step": 5645 + }, + { + "epoch": 0.6200307489567318, + "grad_norm": 2.3714771270751953, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.703230082988739, + "num_tokens": 144325563.0, + "step": 5646 + }, + { + "epoch": 0.6201405666593455, + "grad_norm": 2.6737937927246094, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7313898205757141, + "num_tokens": 144343951.0, + "step": 5647 + }, + { + "epoch": 0.6202503843619591, + "grad_norm": 2.138547658920288, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.6996251344680786, + "num_tokens": 144370584.0, + "step": 5648 + }, + { + "epoch": 0.6203602020645728, + "grad_norm": 2.3786957263946533, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7143192887306213, + "num_tokens": 144392556.0, + "step": 5649 + }, + { + "epoch": 0.6204700197671865, + "grad_norm": 2.5564374923706055, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7379707098007202, + "num_tokens": 144412701.0, + "step": 5650 + }, + { + "epoch": 0.6205798374698002, + "grad_norm": 2.2232441902160645, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7226228713989258, + "num_tokens": 144436208.0, + "step": 5651 + }, + { + "epoch": 0.6206896551724138, + "grad_norm": 2.2436070442199707, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7201429009437561, + "num_tokens": 144459860.0, + "step": 5652 + }, + { + "epoch": 0.6207994728750275, + "grad_norm": 2.0742194652557373, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.7013576030731201, + "num_tokens": 144488913.0, + "step": 5653 + }, + { + "epoch": 0.6209092905776411, + "grad_norm": 2.1774656772613525, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.6877242922782898, + "num_tokens": 144515040.0, + "step": 5654 + }, + { + "epoch": 0.6210191082802548, + "grad_norm": 2.0503721237182617, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7067520022392273, + "num_tokens": 144543054.0, + "step": 5655 + }, + { + "epoch": 0.6211289259828684, + "grad_norm": 2.289492130279541, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.7065841555595398, + "num_tokens": 144566691.0, + "step": 5656 + }, + { + "epoch": 0.6212387436854822, + "grad_norm": 2.363588571548462, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6958383321762085, + "num_tokens": 144589962.0, + "step": 5657 + }, + { + "epoch": 0.6213485613880958, + "grad_norm": 2.4610533714294434, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7125037908554077, + "num_tokens": 144610588.0, + "step": 5658 + }, + { + "epoch": 0.6214583790907094, + "grad_norm": 2.072507381439209, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7143122553825378, + "num_tokens": 144638085.0, + "step": 5659 + }, + { + "epoch": 0.6215681967933231, + "grad_norm": 2.471461296081543, + "learning_rate": 1e-06, + "loss": 0.7976, + "mean_token_accuracy": 0.7474172115325928, + "num_tokens": 144657171.0, + "step": 5660 + }, + { + "epoch": 0.6216780144959367, + "grad_norm": 2.320291757583618, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7323740720748901, + "num_tokens": 144679182.0, + "step": 5661 + }, + { + "epoch": 0.6217878321985504, + "grad_norm": 2.634925127029419, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.721383273601532, + "num_tokens": 144697228.0, + "step": 5662 + }, + { + "epoch": 0.621897649901164, + "grad_norm": 2.226552963256836, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7128952741622925, + "num_tokens": 144720640.0, + "step": 5663 + }, + { + "epoch": 0.6220074676037777, + "grad_norm": 1.8728522062301636, + "learning_rate": 1e-06, + "loss": 1.0339, + "mean_token_accuracy": 0.6893487572669983, + "num_tokens": 144753979.0, + "step": 5664 + }, + { + "epoch": 0.6221172853063914, + "grad_norm": 2.439364194869995, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6968942880630493, + "num_tokens": 144776929.0, + "step": 5665 + }, + { + "epoch": 0.6222271030090051, + "grad_norm": 2.2438690662384033, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.7148301601409912, + "num_tokens": 144802112.0, + "step": 5666 + }, + { + "epoch": 0.6223369207116187, + "grad_norm": 2.0558722019195557, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7221630811691284, + "num_tokens": 144828674.0, + "step": 5667 + }, + { + "epoch": 0.6224467384142324, + "grad_norm": 2.383922576904297, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.711187481880188, + "num_tokens": 144849963.0, + "step": 5668 + }, + { + "epoch": 0.622556556116846, + "grad_norm": 2.0583550930023193, + "learning_rate": 1e-06, + "loss": 1.0732, + "mean_token_accuracy": 0.6869720816612244, + "num_tokens": 144879388.0, + "step": 5669 + }, + { + "epoch": 0.6226663738194597, + "grad_norm": 2.154392957687378, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7229801416397095, + "num_tokens": 144904806.0, + "step": 5670 + }, + { + "epoch": 0.6227761915220733, + "grad_norm": 2.185028314590454, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7262400388717651, + "num_tokens": 144928510.0, + "step": 5671 + }, + { + "epoch": 0.6228860092246871, + "grad_norm": 2.222562551498413, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7087079286575317, + "num_tokens": 144951329.0, + "step": 5672 + }, + { + "epoch": 0.6229958269273007, + "grad_norm": 2.3666064739227295, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.6918664574623108, + "num_tokens": 144975025.0, + "step": 5673 + }, + { + "epoch": 0.6231056446299144, + "grad_norm": 2.6327855587005615, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7290441989898682, + "num_tokens": 144993719.0, + "step": 5674 + }, + { + "epoch": 0.623215462332528, + "grad_norm": 2.160487651824951, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.699854850769043, + "num_tokens": 145018383.0, + "step": 5675 + }, + { + "epoch": 0.6233252800351416, + "grad_norm": 2.314009189605713, + "learning_rate": 1e-06, + "loss": 1.0369, + "mean_token_accuracy": 0.6813286542892456, + "num_tokens": 145044010.0, + "step": 5676 + }, + { + "epoch": 0.6234350977377553, + "grad_norm": 1.9837853908538818, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7076985239982605, + "num_tokens": 145075318.0, + "step": 5677 + }, + { + "epoch": 0.6235449154403689, + "grad_norm": 2.1506588459014893, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7244488596916199, + "num_tokens": 145101334.0, + "step": 5678 + }, + { + "epoch": 0.6236547331429827, + "grad_norm": 2.3624753952026367, + "learning_rate": 1e-06, + "loss": 1.05, + "mean_token_accuracy": 0.6979765892028809, + "num_tokens": 145126149.0, + "step": 5679 + }, + { + "epoch": 0.6237645508455963, + "grad_norm": 2.309649705886841, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7054707407951355, + "num_tokens": 145149670.0, + "step": 5680 + }, + { + "epoch": 0.62387436854821, + "grad_norm": 2.1447372436523438, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7121700048446655, + "num_tokens": 145177417.0, + "step": 5681 + }, + { + "epoch": 0.6239841862508236, + "grad_norm": 2.6725845336914062, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7204415202140808, + "num_tokens": 145198127.0, + "step": 5682 + }, + { + "epoch": 0.6240940039534373, + "grad_norm": 2.026033401489258, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7072129845619202, + "num_tokens": 145226984.0, + "step": 5683 + }, + { + "epoch": 0.6242038216560509, + "grad_norm": 1.9420372247695923, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7047913670539856, + "num_tokens": 145258493.0, + "step": 5684 + }, + { + "epoch": 0.6243136393586646, + "grad_norm": 2.0855660438537598, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7175612449645996, + "num_tokens": 145284562.0, + "step": 5685 + }, + { + "epoch": 0.6244234570612783, + "grad_norm": 2.2831292152404785, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7204722166061401, + "num_tokens": 145307109.0, + "step": 5686 + }, + { + "epoch": 0.624533274763892, + "grad_norm": 2.4010350704193115, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7086353898048401, + "num_tokens": 145330153.0, + "step": 5687 + }, + { + "epoch": 0.6246430924665056, + "grad_norm": 2.392836093902588, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6832282543182373, + "num_tokens": 145353186.0, + "step": 5688 + }, + { + "epoch": 0.6247529101691193, + "grad_norm": 2.3581643104553223, + "learning_rate": 1e-06, + "loss": 1.0574, + "mean_token_accuracy": 0.6794232130050659, + "num_tokens": 145378735.0, + "step": 5689 + }, + { + "epoch": 0.6248627278717329, + "grad_norm": 2.159444570541382, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7032437324523926, + "num_tokens": 145403278.0, + "step": 5690 + }, + { + "epoch": 0.6249725455743466, + "grad_norm": 2.329145908355713, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.6912104487419128, + "num_tokens": 145428095.0, + "step": 5691 + }, + { + "epoch": 0.6250823632769602, + "grad_norm": 2.2067177295684814, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7082035541534424, + "num_tokens": 145453614.0, + "step": 5692 + }, + { + "epoch": 0.6251921809795739, + "grad_norm": 2.0556981563568115, + "learning_rate": 1e-06, + "loss": 1.0565, + "mean_token_accuracy": 0.6971713900566101, + "num_tokens": 145484115.0, + "step": 5693 + }, + { + "epoch": 0.6253019986821876, + "grad_norm": 2.024407386779785, + "learning_rate": 1e-06, + "loss": 1.0334, + "mean_token_accuracy": 0.6936318874359131, + "num_tokens": 145514307.0, + "step": 5694 + }, + { + "epoch": 0.6254118163848013, + "grad_norm": 2.2807681560516357, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7281823754310608, + "num_tokens": 145538902.0, + "step": 5695 + }, + { + "epoch": 0.6255216340874149, + "grad_norm": 2.256615400314331, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7046434879302979, + "num_tokens": 145561364.0, + "step": 5696 + }, + { + "epoch": 0.6256314517900285, + "grad_norm": 2.18982195854187, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.70743727684021, + "num_tokens": 145587091.0, + "step": 5697 + }, + { + "epoch": 0.6257412694926422, + "grad_norm": 2.283841133117676, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7012878656387329, + "num_tokens": 145610909.0, + "step": 5698 + }, + { + "epoch": 0.6258510871952558, + "grad_norm": 2.571775436401367, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7145599722862244, + "num_tokens": 145631141.0, + "step": 5699 + }, + { + "epoch": 0.6259609048978695, + "grad_norm": 2.084914445877075, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7030338048934937, + "num_tokens": 145658724.0, + "step": 5700 + }, + { + "epoch": 0.6260707226004832, + "grad_norm": 2.173346519470215, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.6958016157150269, + "num_tokens": 145685656.0, + "step": 5701 + }, + { + "epoch": 0.6261805403030969, + "grad_norm": 2.2614595890045166, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.687820315361023, + "num_tokens": 145712350.0, + "step": 5702 + }, + { + "epoch": 0.6262903580057105, + "grad_norm": 1.8843183517456055, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.6879361867904663, + "num_tokens": 145746303.0, + "step": 5703 + }, + { + "epoch": 0.6264001757083242, + "grad_norm": 2.1931562423706055, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6921730041503906, + "num_tokens": 145777534.0, + "step": 5704 + }, + { + "epoch": 0.6265099934109378, + "grad_norm": 2.3063604831695557, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7090604305267334, + "num_tokens": 145803679.0, + "step": 5705 + }, + { + "epoch": 0.6266198111135515, + "grad_norm": 2.317486047744751, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7136857509613037, + "num_tokens": 145826427.0, + "step": 5706 + }, + { + "epoch": 0.6267296288161651, + "grad_norm": 2.121718168258667, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7353470325469971, + "num_tokens": 145851205.0, + "step": 5707 + }, + { + "epoch": 0.6268394465187789, + "grad_norm": 2.236107587814331, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6972975730895996, + "num_tokens": 145876493.0, + "step": 5708 + }, + { + "epoch": 0.6269492642213925, + "grad_norm": 2.1542389392852783, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6923706531524658, + "num_tokens": 145902332.0, + "step": 5709 + }, + { + "epoch": 0.6270590819240062, + "grad_norm": 2.2072463035583496, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7063242793083191, + "num_tokens": 145927760.0, + "step": 5710 + }, + { + "epoch": 0.6271688996266198, + "grad_norm": 2.2304978370666504, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.6988834142684937, + "num_tokens": 145950815.0, + "step": 5711 + }, + { + "epoch": 0.6272787173292335, + "grad_norm": 2.136725664138794, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.7043977975845337, + "num_tokens": 145976288.0, + "step": 5712 + }, + { + "epoch": 0.6273885350318471, + "grad_norm": 2.2725274562835693, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7082081437110901, + "num_tokens": 145999461.0, + "step": 5713 + }, + { + "epoch": 0.6274983527344608, + "grad_norm": 2.084482192993164, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7289775609970093, + "num_tokens": 146026770.0, + "step": 5714 + }, + { + "epoch": 0.6276081704370745, + "grad_norm": 2.420715093612671, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.6971712708473206, + "num_tokens": 146048668.0, + "step": 5715 + }, + { + "epoch": 0.6277179881396882, + "grad_norm": 2.155482053756714, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7222950458526611, + "num_tokens": 146073345.0, + "step": 5716 + }, + { + "epoch": 0.6278278058423018, + "grad_norm": 2.098680257797241, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.6958569288253784, + "num_tokens": 146100981.0, + "step": 5717 + }, + { + "epoch": 0.6279376235449154, + "grad_norm": 2.179931879043579, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7096620798110962, + "num_tokens": 146127052.0, + "step": 5718 + }, + { + "epoch": 0.6280474412475291, + "grad_norm": 2.501084327697754, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.6992798447608948, + "num_tokens": 146148203.0, + "step": 5719 + }, + { + "epoch": 0.6281572589501427, + "grad_norm": 2.1724321842193604, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6940509676933289, + "num_tokens": 146174638.0, + "step": 5720 + }, + { + "epoch": 0.6282670766527564, + "grad_norm": 2.3371171951293945, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6934698820114136, + "num_tokens": 146197612.0, + "step": 5721 + }, + { + "epoch": 0.62837689435537, + "grad_norm": 2.0909502506256104, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7295300364494324, + "num_tokens": 146220864.0, + "step": 5722 + }, + { + "epoch": 0.6284867120579838, + "grad_norm": 2.3458378314971924, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.7000116109848022, + "num_tokens": 146244196.0, + "step": 5723 + }, + { + "epoch": 0.6285965297605974, + "grad_norm": 1.997300624847412, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6916705369949341, + "num_tokens": 146274719.0, + "step": 5724 + }, + { + "epoch": 0.6287063474632111, + "grad_norm": 2.4680776596069336, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7205843925476074, + "num_tokens": 146298438.0, + "step": 5725 + }, + { + "epoch": 0.6288161651658247, + "grad_norm": 2.3771631717681885, + "learning_rate": 1e-06, + "loss": 1.0167, + "mean_token_accuracy": 0.6952430605888367, + "num_tokens": 146323408.0, + "step": 5726 + }, + { + "epoch": 0.6289259828684384, + "grad_norm": 2.298754930496216, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.712828516960144, + "num_tokens": 146345444.0, + "step": 5727 + }, + { + "epoch": 0.629035800571052, + "grad_norm": 1.9660277366638184, + "learning_rate": 1e-06, + "loss": 1.0569, + "mean_token_accuracy": 0.6748976111412048, + "num_tokens": 146377474.0, + "step": 5728 + }, + { + "epoch": 0.6291456182736657, + "grad_norm": 2.1985905170440674, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6925930976867676, + "num_tokens": 146402456.0, + "step": 5729 + }, + { + "epoch": 0.6292554359762794, + "grad_norm": 2.3895413875579834, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7190014719963074, + "num_tokens": 146423646.0, + "step": 5730 + }, + { + "epoch": 0.6293652536788931, + "grad_norm": 2.382437229156494, + "learning_rate": 1e-06, + "loss": 1.0658, + "mean_token_accuracy": 0.6792011260986328, + "num_tokens": 146448523.0, + "step": 5731 + }, + { + "epoch": 0.6294750713815067, + "grad_norm": 2.2340614795684814, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7192333936691284, + "num_tokens": 146472675.0, + "step": 5732 + }, + { + "epoch": 0.6295848890841204, + "grad_norm": 2.5238864421844482, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7114487886428833, + "num_tokens": 146493935.0, + "step": 5733 + }, + { + "epoch": 0.629694706786734, + "grad_norm": 2.18612003326416, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7066432237625122, + "num_tokens": 146520748.0, + "step": 5734 + }, + { + "epoch": 0.6298045244893477, + "grad_norm": 2.211134195327759, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7065480947494507, + "num_tokens": 146545927.0, + "step": 5735 + }, + { + "epoch": 0.6299143421919613, + "grad_norm": 1.946568250656128, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6950249671936035, + "num_tokens": 146577419.0, + "step": 5736 + }, + { + "epoch": 0.630024159894575, + "grad_norm": 2.3714194297790527, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7059489488601685, + "num_tokens": 146598751.0, + "step": 5737 + }, + { + "epoch": 0.6301339775971887, + "grad_norm": 2.365388870239258, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7154623866081238, + "num_tokens": 146620347.0, + "step": 5738 + }, + { + "epoch": 0.6302437952998023, + "grad_norm": 2.0442540645599365, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7133316993713379, + "num_tokens": 146647294.0, + "step": 5739 + }, + { + "epoch": 0.630353613002416, + "grad_norm": 2.391432523727417, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7357062101364136, + "num_tokens": 146667629.0, + "step": 5740 + }, + { + "epoch": 0.6304634307050296, + "grad_norm": 2.1321632862091064, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7136266231536865, + "num_tokens": 146693217.0, + "step": 5741 + }, + { + "epoch": 0.6305732484076433, + "grad_norm": 2.2500505447387695, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6967611312866211, + "num_tokens": 146716623.0, + "step": 5742 + }, + { + "epoch": 0.6306830661102569, + "grad_norm": 2.7256228923797607, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7117297053337097, + "num_tokens": 146733445.0, + "step": 5743 + }, + { + "epoch": 0.6307928838128707, + "grad_norm": 1.872353434562683, + "learning_rate": 1e-06, + "loss": 1.0417, + "mean_token_accuracy": 0.6843432188034058, + "num_tokens": 146765344.0, + "step": 5744 + }, + { + "epoch": 0.6309027015154843, + "grad_norm": 2.1820027828216553, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7034285068511963, + "num_tokens": 146791688.0, + "step": 5745 + }, + { + "epoch": 0.631012519218098, + "grad_norm": 2.1271989345550537, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.7036821246147156, + "num_tokens": 146816977.0, + "step": 5746 + }, + { + "epoch": 0.6311223369207116, + "grad_norm": 2.0422658920288086, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7063867449760437, + "num_tokens": 146844248.0, + "step": 5747 + }, + { + "epoch": 0.6312321546233253, + "grad_norm": 2.3649113178253174, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.6979684233665466, + "num_tokens": 146866618.0, + "step": 5748 + }, + { + "epoch": 0.6313419723259389, + "grad_norm": 2.1094977855682373, + "learning_rate": 1e-06, + "loss": 1.0594, + "mean_token_accuracy": 0.6837893724441528, + "num_tokens": 146895841.0, + "step": 5749 + }, + { + "epoch": 0.6314517900285526, + "grad_norm": 2.1158878803253174, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7137672901153564, + "num_tokens": 146922169.0, + "step": 5750 + }, + { + "epoch": 0.6315616077311663, + "grad_norm": 2.21662974357605, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7157498598098755, + "num_tokens": 146946384.0, + "step": 5751 + }, + { + "epoch": 0.63167142543378, + "grad_norm": 2.004607677459717, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7171578407287598, + "num_tokens": 146976343.0, + "step": 5752 + }, + { + "epoch": 0.6317812431363936, + "grad_norm": 2.5328121185302734, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7688606381416321, + "num_tokens": 146993891.0, + "step": 5753 + }, + { + "epoch": 0.6318910608390073, + "grad_norm": 2.425528049468994, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7130231857299805, + "num_tokens": 147016738.0, + "step": 5754 + }, + { + "epoch": 0.6320008785416209, + "grad_norm": 2.2895689010620117, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6885573863983154, + "num_tokens": 147040043.0, + "step": 5755 + }, + { + "epoch": 0.6321106962442345, + "grad_norm": 2.486985683441162, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6973549723625183, + "num_tokens": 147060516.0, + "step": 5756 + }, + { + "epoch": 0.6322205139468482, + "grad_norm": 2.1828508377075195, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.701494574546814, + "num_tokens": 147085124.0, + "step": 5757 + }, + { + "epoch": 0.6323303316494618, + "grad_norm": 2.286897659301758, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.7065664529800415, + "num_tokens": 147110126.0, + "step": 5758 + }, + { + "epoch": 0.6324401493520756, + "grad_norm": 2.173691987991333, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7124243974685669, + "num_tokens": 147132982.0, + "step": 5759 + }, + { + "epoch": 0.6325499670546892, + "grad_norm": 2.135305404663086, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.7006547451019287, + "num_tokens": 147159229.0, + "step": 5760 + }, + { + "epoch": 0.6326597847573029, + "grad_norm": 2.104891300201416, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7040523290634155, + "num_tokens": 147185400.0, + "step": 5761 + }, + { + "epoch": 0.6327696024599165, + "grad_norm": 2.5572948455810547, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7308813333511353, + "num_tokens": 147203929.0, + "step": 5762 + }, + { + "epoch": 0.6328794201625302, + "grad_norm": 2.1981050968170166, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7057123780250549, + "num_tokens": 147227746.0, + "step": 5763 + }, + { + "epoch": 0.6329892378651438, + "grad_norm": 2.449571371078491, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7319960594177246, + "num_tokens": 147247977.0, + "step": 5764 + }, + { + "epoch": 0.6330990555677575, + "grad_norm": 2.2025225162506104, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.716405987739563, + "num_tokens": 147272801.0, + "step": 5765 + }, + { + "epoch": 0.6332088732703712, + "grad_norm": 2.0432324409484863, + "learning_rate": 1e-06, + "loss": 1.0763, + "mean_token_accuracy": 0.6781789660453796, + "num_tokens": 147300892.0, + "step": 5766 + }, + { + "epoch": 0.6333186909729849, + "grad_norm": 2.156940221786499, + "learning_rate": 1e-06, + "loss": 1.0676, + "mean_token_accuracy": 0.6795058250427246, + "num_tokens": 147327989.0, + "step": 5767 + }, + { + "epoch": 0.6334285086755985, + "grad_norm": 2.0065555572509766, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7082720398902893, + "num_tokens": 147356578.0, + "step": 5768 + }, + { + "epoch": 0.6335383263782122, + "grad_norm": 1.9848122596740723, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7037001848220825, + "num_tokens": 147386247.0, + "step": 5769 + }, + { + "epoch": 0.6336481440808258, + "grad_norm": 2.238450527191162, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6922422647476196, + "num_tokens": 147411211.0, + "step": 5770 + }, + { + "epoch": 0.6337579617834395, + "grad_norm": 2.344379425048828, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7060659527778625, + "num_tokens": 147434552.0, + "step": 5771 + }, + { + "epoch": 0.6338677794860531, + "grad_norm": 2.2106003761291504, + "learning_rate": 1e-06, + "loss": 1.0546, + "mean_token_accuracy": 0.6839053630828857, + "num_tokens": 147459394.0, + "step": 5772 + }, + { + "epoch": 0.6339775971886669, + "grad_norm": 2.475370407104492, + "learning_rate": 1e-06, + "loss": 1.052, + "mean_token_accuracy": 0.6968674659729004, + "num_tokens": 147480409.0, + "step": 5773 + }, + { + "epoch": 0.6340874148912805, + "grad_norm": 2.218973398208618, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7180867195129395, + "num_tokens": 147503627.0, + "step": 5774 + }, + { + "epoch": 0.6341972325938942, + "grad_norm": 2.2669503688812256, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7005111575126648, + "num_tokens": 147527083.0, + "step": 5775 + }, + { + "epoch": 0.6343070502965078, + "grad_norm": 2.1048498153686523, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7050246000289917, + "num_tokens": 147554871.0, + "step": 5776 + }, + { + "epoch": 0.6344168679991214, + "grad_norm": 2.2752208709716797, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7063298225402832, + "num_tokens": 147577393.0, + "step": 5777 + }, + { + "epoch": 0.6345266857017351, + "grad_norm": 2.506927967071533, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7267808318138123, + "num_tokens": 147596872.0, + "step": 5778 + }, + { + "epoch": 0.6346365034043487, + "grad_norm": 2.3646600246429443, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7114197015762329, + "num_tokens": 147618955.0, + "step": 5779 + }, + { + "epoch": 0.6347463211069625, + "grad_norm": 2.263947010040283, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.7016713619232178, + "num_tokens": 147643346.0, + "step": 5780 + }, + { + "epoch": 0.6348561388095761, + "grad_norm": 2.4324655532836914, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7165395021438599, + "num_tokens": 147664568.0, + "step": 5781 + }, + { + "epoch": 0.6349659565121898, + "grad_norm": 2.255458354949951, + "learning_rate": 1e-06, + "loss": 1.028, + "mean_token_accuracy": 0.6919860243797302, + "num_tokens": 147689563.0, + "step": 5782 + }, + { + "epoch": 0.6350757742148034, + "grad_norm": 2.414536714553833, + "learning_rate": 1e-06, + "loss": 1.0557, + "mean_token_accuracy": 0.6865569949150085, + "num_tokens": 147712100.0, + "step": 5783 + }, + { + "epoch": 0.6351855919174171, + "grad_norm": 2.056114435195923, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.693501353263855, + "num_tokens": 147740819.0, + "step": 5784 + }, + { + "epoch": 0.6352954096200307, + "grad_norm": 2.2413487434387207, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7188805341720581, + "num_tokens": 147763269.0, + "step": 5785 + }, + { + "epoch": 0.6354052273226444, + "grad_norm": 2.1192626953125, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6920409202575684, + "num_tokens": 147792534.0, + "step": 5786 + }, + { + "epoch": 0.635515045025258, + "grad_norm": 1.9074538946151733, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.6965889930725098, + "num_tokens": 147828557.0, + "step": 5787 + }, + { + "epoch": 0.6356248627278718, + "grad_norm": 2.0921013355255127, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7031524181365967, + "num_tokens": 147856162.0, + "step": 5788 + }, + { + "epoch": 0.6357346804304854, + "grad_norm": 2.276249408721924, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7064696550369263, + "num_tokens": 147880735.0, + "step": 5789 + }, + { + "epoch": 0.6358444981330991, + "grad_norm": 2.0868592262268066, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7088344097137451, + "num_tokens": 147909324.0, + "step": 5790 + }, + { + "epoch": 0.6359543158357127, + "grad_norm": 2.3200433254241943, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.731367290019989, + "num_tokens": 147931103.0, + "step": 5791 + }, + { + "epoch": 0.6360641335383264, + "grad_norm": 2.260687828063965, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7204533815383911, + "num_tokens": 147954740.0, + "step": 5792 + }, + { + "epoch": 0.63617395124094, + "grad_norm": 2.231801748275757, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.710088312625885, + "num_tokens": 147979119.0, + "step": 5793 + }, + { + "epoch": 0.6362837689435537, + "grad_norm": 2.1749465465545654, + "learning_rate": 1e-06, + "loss": 1.0884, + "mean_token_accuracy": 0.6726467609405518, + "num_tokens": 148007217.0, + "step": 5794 + }, + { + "epoch": 0.6363935866461674, + "grad_norm": 2.2172651290893555, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.7019085884094238, + "num_tokens": 148031907.0, + "step": 5795 + }, + { + "epoch": 0.636503404348781, + "grad_norm": 1.923892855644226, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.6966472864151001, + "num_tokens": 148066561.0, + "step": 5796 + }, + { + "epoch": 0.6366132220513947, + "grad_norm": 1.9676141738891602, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7149377465248108, + "num_tokens": 148099250.0, + "step": 5797 + }, + { + "epoch": 0.6367230397540083, + "grad_norm": 2.1350111961364746, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6858363747596741, + "num_tokens": 148128800.0, + "step": 5798 + }, + { + "epoch": 0.636832857456622, + "grad_norm": 2.377854108810425, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7112866640090942, + "num_tokens": 148153599.0, + "step": 5799 + }, + { + "epoch": 0.6369426751592356, + "grad_norm": 1.932255744934082, + "learning_rate": 1e-06, + "loss": 1.052, + "mean_token_accuracy": 0.6785142421722412, + "num_tokens": 148184929.0, + "step": 5800 + }, + { + "epoch": 0.6370524928618493, + "grad_norm": 2.158134937286377, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7112509608268738, + "num_tokens": 148210819.0, + "step": 5801 + }, + { + "epoch": 0.637162310564463, + "grad_norm": 2.2372641563415527, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7023339867591858, + "num_tokens": 148234641.0, + "step": 5802 + }, + { + "epoch": 0.6372721282670767, + "grad_norm": 2.158546209335327, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7066254019737244, + "num_tokens": 148259425.0, + "step": 5803 + }, + { + "epoch": 0.6373819459696903, + "grad_norm": 2.4706106185913086, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7191661596298218, + "num_tokens": 148278342.0, + "step": 5804 + }, + { + "epoch": 0.637491763672304, + "grad_norm": 2.2155110836029053, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.695254921913147, + "num_tokens": 148303906.0, + "step": 5805 + }, + { + "epoch": 0.6376015813749176, + "grad_norm": 2.074415683746338, + "learning_rate": 1e-06, + "loss": 1.0393, + "mean_token_accuracy": 0.6892859935760498, + "num_tokens": 148333445.0, + "step": 5806 + }, + { + "epoch": 0.6377113990775313, + "grad_norm": 2.3945727348327637, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7028647065162659, + "num_tokens": 148355076.0, + "step": 5807 + }, + { + "epoch": 0.6378212167801449, + "grad_norm": 2.477658271789551, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7340836524963379, + "num_tokens": 148374584.0, + "step": 5808 + }, + { + "epoch": 0.6379310344827587, + "grad_norm": 2.443070411682129, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7196522355079651, + "num_tokens": 148395720.0, + "step": 5809 + }, + { + "epoch": 0.6380408521853723, + "grad_norm": 2.303668260574341, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6955139636993408, + "num_tokens": 148420616.0, + "step": 5810 + }, + { + "epoch": 0.638150669887986, + "grad_norm": 2.5365102291107178, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7097774147987366, + "num_tokens": 148440069.0, + "step": 5811 + }, + { + "epoch": 0.6382604875905996, + "grad_norm": 2.5335934162139893, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.7034590840339661, + "num_tokens": 148467396.0, + "step": 5812 + }, + { + "epoch": 0.6383703052932133, + "grad_norm": 2.351125955581665, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7160725593566895, + "num_tokens": 148490572.0, + "step": 5813 + }, + { + "epoch": 0.6384801229958269, + "grad_norm": 2.1524550914764404, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.6962716579437256, + "num_tokens": 148517671.0, + "step": 5814 + }, + { + "epoch": 0.6385899406984406, + "grad_norm": 2.0867366790771484, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7035370469093323, + "num_tokens": 148545819.0, + "step": 5815 + }, + { + "epoch": 0.6386997584010542, + "grad_norm": 2.160597324371338, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.7012478113174438, + "num_tokens": 148572169.0, + "step": 5816 + }, + { + "epoch": 0.638809576103668, + "grad_norm": 2.2187018394470215, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.716639518737793, + "num_tokens": 148595909.0, + "step": 5817 + }, + { + "epoch": 0.6389193938062816, + "grad_norm": 2.323230743408203, + "learning_rate": 1e-06, + "loss": 1.0696, + "mean_token_accuracy": 0.6875402927398682, + "num_tokens": 148618905.0, + "step": 5818 + }, + { + "epoch": 0.6390292115088952, + "grad_norm": 1.993025302886963, + "learning_rate": 1e-06, + "loss": 1.0475, + "mean_token_accuracy": 0.6825730204582214, + "num_tokens": 148650960.0, + "step": 5819 + }, + { + "epoch": 0.6391390292115089, + "grad_norm": 2.1846303939819336, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.7053742408752441, + "num_tokens": 148678155.0, + "step": 5820 + }, + { + "epoch": 0.6392488469141225, + "grad_norm": 2.2181124687194824, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7088653445243835, + "num_tokens": 148701754.0, + "step": 5821 + }, + { + "epoch": 0.6393586646167362, + "grad_norm": 2.2615246772766113, + "learning_rate": 1e-06, + "loss": 1.0289, + "mean_token_accuracy": 0.6965806484222412, + "num_tokens": 148726889.0, + "step": 5822 + }, + { + "epoch": 0.6394684823193498, + "grad_norm": 2.512413501739502, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6886402368545532, + "num_tokens": 148747811.0, + "step": 5823 + }, + { + "epoch": 0.6395783000219636, + "grad_norm": 2.2820794582366943, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.693139910697937, + "num_tokens": 148773411.0, + "step": 5824 + }, + { + "epoch": 0.6396881177245772, + "grad_norm": 2.5241620540618896, + "learning_rate": 1e-06, + "loss": 1.0413, + "mean_token_accuracy": 0.6922218799591064, + "num_tokens": 148800290.0, + "step": 5825 + }, + { + "epoch": 0.6397979354271909, + "grad_norm": 2.241306781768799, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7096266150474548, + "num_tokens": 148825027.0, + "step": 5826 + }, + { + "epoch": 0.6399077531298045, + "grad_norm": 2.192422866821289, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.685454249382019, + "num_tokens": 148852681.0, + "step": 5827 + }, + { + "epoch": 0.6400175708324182, + "grad_norm": 2.3386120796203613, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7181872129440308, + "num_tokens": 148877133.0, + "step": 5828 + }, + { + "epoch": 0.6401273885350318, + "grad_norm": 2.3771438598632812, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.716435432434082, + "num_tokens": 148898269.0, + "step": 5829 + }, + { + "epoch": 0.6402372062376455, + "grad_norm": 2.190352439880371, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7186722755432129, + "num_tokens": 148922901.0, + "step": 5830 + }, + { + "epoch": 0.6403470239402592, + "grad_norm": 2.1678519248962402, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.6988924741744995, + "num_tokens": 148949134.0, + "step": 5831 + }, + { + "epoch": 0.6404568416428729, + "grad_norm": 2.0455129146575928, + "learning_rate": 1e-06, + "loss": 0.7698, + "mean_token_accuracy": 0.7493873238563538, + "num_tokens": 148974264.0, + "step": 5832 + }, + { + "epoch": 0.6405666593454865, + "grad_norm": 2.2087905406951904, + "learning_rate": 1e-06, + "loss": 1.0684, + "mean_token_accuracy": 0.6838243007659912, + "num_tokens": 149002995.0, + "step": 5833 + }, + { + "epoch": 0.6406764770481002, + "grad_norm": 2.428969383239746, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6950589418411255, + "num_tokens": 149027797.0, + "step": 5834 + }, + { + "epoch": 0.6407862947507138, + "grad_norm": 2.374624252319336, + "learning_rate": 1e-06, + "loss": 1.0524, + "mean_token_accuracy": 0.6840032339096069, + "num_tokens": 149051552.0, + "step": 5835 + }, + { + "epoch": 0.6408961124533274, + "grad_norm": 2.3105111122131348, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7116581201553345, + "num_tokens": 149074418.0, + "step": 5836 + }, + { + "epoch": 0.6410059301559411, + "grad_norm": 2.1030654907226562, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7263841032981873, + "num_tokens": 149100330.0, + "step": 5837 + }, + { + "epoch": 0.6411157478585549, + "grad_norm": 1.8912229537963867, + "learning_rate": 1e-06, + "loss": 1.0476, + "mean_token_accuracy": 0.6874501705169678, + "num_tokens": 149134027.0, + "step": 5838 + }, + { + "epoch": 0.6412255655611685, + "grad_norm": 2.3820955753326416, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7036620378494263, + "num_tokens": 149155968.0, + "step": 5839 + }, + { + "epoch": 0.6413353832637821, + "grad_norm": 2.1546592712402344, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.708791196346283, + "num_tokens": 149181259.0, + "step": 5840 + }, + { + "epoch": 0.6414452009663958, + "grad_norm": 2.0288448333740234, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7244194746017456, + "num_tokens": 149209485.0, + "step": 5841 + }, + { + "epoch": 0.6415550186690094, + "grad_norm": 2.0087387561798096, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6948229670524597, + "num_tokens": 149238721.0, + "step": 5842 + }, + { + "epoch": 0.6416648363716231, + "grad_norm": 2.57080340385437, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6897575855255127, + "num_tokens": 149260445.0, + "step": 5843 + }, + { + "epoch": 0.6417746540742367, + "grad_norm": 2.204174757003784, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6966445446014404, + "num_tokens": 149286814.0, + "step": 5844 + }, + { + "epoch": 0.6418844717768504, + "grad_norm": 2.1543262004852295, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7005504965782166, + "num_tokens": 149313203.0, + "step": 5845 + }, + { + "epoch": 0.6419942894794641, + "grad_norm": 2.4540817737579346, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7002466320991516, + "num_tokens": 149336561.0, + "step": 5846 + }, + { + "epoch": 0.6421041071820778, + "grad_norm": 2.0377631187438965, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7316633462905884, + "num_tokens": 149362813.0, + "step": 5847 + }, + { + "epoch": 0.6422139248846914, + "grad_norm": 2.3598861694335938, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7045422792434692, + "num_tokens": 149386484.0, + "step": 5848 + }, + { + "epoch": 0.6423237425873051, + "grad_norm": 2.088501453399658, + "learning_rate": 1e-06, + "loss": 1.0846, + "mean_token_accuracy": 0.6736176013946533, + "num_tokens": 149421069.0, + "step": 5849 + }, + { + "epoch": 0.6424335602899187, + "grad_norm": 2.0429630279541016, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6912035942077637, + "num_tokens": 149449234.0, + "step": 5850 + }, + { + "epoch": 0.6425433779925324, + "grad_norm": 2.4163706302642822, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7130254507064819, + "num_tokens": 149470871.0, + "step": 5851 + }, + { + "epoch": 0.642653195695146, + "grad_norm": 1.9450575113296509, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7177658081054688, + "num_tokens": 149498914.0, + "step": 5852 + }, + { + "epoch": 0.6427630133977598, + "grad_norm": 2.08697509765625, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7078391313552856, + "num_tokens": 149525646.0, + "step": 5853 + }, + { + "epoch": 0.6428728311003734, + "grad_norm": 2.299675941467285, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7140907645225525, + "num_tokens": 149547910.0, + "step": 5854 + }, + { + "epoch": 0.6429826488029871, + "grad_norm": 2.0961129665374756, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7139145731925964, + "num_tokens": 149574473.0, + "step": 5855 + }, + { + "epoch": 0.6430924665056007, + "grad_norm": 2.083967685699463, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.7011522054672241, + "num_tokens": 149602968.0, + "step": 5856 + }, + { + "epoch": 0.6432022842082143, + "grad_norm": 1.9836971759796143, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7132573127746582, + "num_tokens": 149629990.0, + "step": 5857 + }, + { + "epoch": 0.643312101910828, + "grad_norm": 2.022045135498047, + "learning_rate": 1e-06, + "loss": 1.0274, + "mean_token_accuracy": 0.6889722347259521, + "num_tokens": 149658862.0, + "step": 5858 + }, + { + "epoch": 0.6434219196134416, + "grad_norm": 2.1123948097229004, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.693510890007019, + "num_tokens": 149686607.0, + "step": 5859 + }, + { + "epoch": 0.6435317373160554, + "grad_norm": 2.6322145462036133, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6919331550598145, + "num_tokens": 149706716.0, + "step": 5860 + }, + { + "epoch": 0.643641555018669, + "grad_norm": 2.2809641361236572, + "learning_rate": 1e-06, + "loss": 1.0481, + "mean_token_accuracy": 0.6896604299545288, + "num_tokens": 149731834.0, + "step": 5861 + }, + { + "epoch": 0.6437513727212827, + "grad_norm": 2.058702230453491, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6920689940452576, + "num_tokens": 149762342.0, + "step": 5862 + }, + { + "epoch": 0.6438611904238963, + "grad_norm": 2.2640814781188965, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7092093825340271, + "num_tokens": 149787542.0, + "step": 5863 + }, + { + "epoch": 0.64397100812651, + "grad_norm": 2.1078293323516846, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7085513472557068, + "num_tokens": 149815697.0, + "step": 5864 + }, + { + "epoch": 0.6440808258291236, + "grad_norm": 1.9577033519744873, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7144027352333069, + "num_tokens": 149846015.0, + "step": 5865 + }, + { + "epoch": 0.6441906435317373, + "grad_norm": 2.2700610160827637, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.693984866142273, + "num_tokens": 149871610.0, + "step": 5866 + }, + { + "epoch": 0.644300461234351, + "grad_norm": 1.9874000549316406, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.7048659324645996, + "num_tokens": 149902025.0, + "step": 5867 + }, + { + "epoch": 0.6444102789369647, + "grad_norm": 2.363041639328003, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7231106758117676, + "num_tokens": 149924588.0, + "step": 5868 + }, + { + "epoch": 0.6445200966395783, + "grad_norm": 2.2776453495025635, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7068284153938293, + "num_tokens": 149947479.0, + "step": 5869 + }, + { + "epoch": 0.644629914342192, + "grad_norm": 2.23445725440979, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6940184831619263, + "num_tokens": 149972801.0, + "step": 5870 + }, + { + "epoch": 0.6447397320448056, + "grad_norm": 2.2193470001220703, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.709039568901062, + "num_tokens": 149998110.0, + "step": 5871 + }, + { + "epoch": 0.6448495497474193, + "grad_norm": 2.2683041095733643, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7159879207611084, + "num_tokens": 150020516.0, + "step": 5872 + }, + { + "epoch": 0.6449593674500329, + "grad_norm": 2.4505653381347656, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7186706066131592, + "num_tokens": 150042565.0, + "step": 5873 + }, + { + "epoch": 0.6450691851526466, + "grad_norm": 2.3571014404296875, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7098037600517273, + "num_tokens": 150068256.0, + "step": 5874 + }, + { + "epoch": 0.6451790028552603, + "grad_norm": 2.128190755844116, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7193626165390015, + "num_tokens": 150095658.0, + "step": 5875 + }, + { + "epoch": 0.645288820557874, + "grad_norm": 2.2119383811950684, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7289321422576904, + "num_tokens": 150119019.0, + "step": 5876 + }, + { + "epoch": 0.6453986382604876, + "grad_norm": 2.12581729888916, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7025254964828491, + "num_tokens": 150145846.0, + "step": 5877 + }, + { + "epoch": 0.6455084559631012, + "grad_norm": 2.522170066833496, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7388933897018433, + "num_tokens": 150165132.0, + "step": 5878 + }, + { + "epoch": 0.6456182736657149, + "grad_norm": 1.8887406587600708, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.691855788230896, + "num_tokens": 150196419.0, + "step": 5879 + }, + { + "epoch": 0.6457280913683285, + "grad_norm": 2.4518778324127197, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.74228835105896, + "num_tokens": 150216275.0, + "step": 5880 + }, + { + "epoch": 0.6458379090709422, + "grad_norm": 2.0579378604888916, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6837036609649658, + "num_tokens": 150244365.0, + "step": 5881 + }, + { + "epoch": 0.6459477267735559, + "grad_norm": 2.1569137573242188, + "learning_rate": 1e-06, + "loss": 1.0376, + "mean_token_accuracy": 0.6861175298690796, + "num_tokens": 150272238.0, + "step": 5882 + }, + { + "epoch": 0.6460575444761696, + "grad_norm": 2.1728720664978027, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.6956290602684021, + "num_tokens": 150297625.0, + "step": 5883 + }, + { + "epoch": 0.6461673621787832, + "grad_norm": 2.083118438720703, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7044037580490112, + "num_tokens": 150326456.0, + "step": 5884 + }, + { + "epoch": 0.6462771798813969, + "grad_norm": 2.2164998054504395, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7336786985397339, + "num_tokens": 150350784.0, + "step": 5885 + }, + { + "epoch": 0.6463869975840105, + "grad_norm": 2.102498769760132, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7248939871788025, + "num_tokens": 150376378.0, + "step": 5886 + }, + { + "epoch": 0.6464968152866242, + "grad_norm": 2.153871536254883, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.6979496479034424, + "num_tokens": 150404718.0, + "step": 5887 + }, + { + "epoch": 0.6466066329892378, + "grad_norm": 2.199265241622925, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.7015067338943481, + "num_tokens": 150428905.0, + "step": 5888 + }, + { + "epoch": 0.6467164506918516, + "grad_norm": 2.1872966289520264, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7090059518814087, + "num_tokens": 150454369.0, + "step": 5889 + }, + { + "epoch": 0.6468262683944652, + "grad_norm": 2.4077584743499756, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.6908986568450928, + "num_tokens": 150477630.0, + "step": 5890 + }, + { + "epoch": 0.6469360860970789, + "grad_norm": 2.243312358856201, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7093065977096558, + "num_tokens": 150500578.0, + "step": 5891 + }, + { + "epoch": 0.6470459037996925, + "grad_norm": 1.9454649686813354, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7064743638038635, + "num_tokens": 150531502.0, + "step": 5892 + }, + { + "epoch": 0.6471557215023062, + "grad_norm": 2.0410356521606445, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6938732266426086, + "num_tokens": 150560371.0, + "step": 5893 + }, + { + "epoch": 0.6472655392049198, + "grad_norm": 2.469964027404785, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7174094915390015, + "num_tokens": 150579896.0, + "step": 5894 + }, + { + "epoch": 0.6473753569075335, + "grad_norm": 1.9935234785079956, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7175509333610535, + "num_tokens": 150608572.0, + "step": 5895 + }, + { + "epoch": 0.6474851746101472, + "grad_norm": 2.0830533504486084, + "learning_rate": 1e-06, + "loss": 1.0779, + "mean_token_accuracy": 0.6810572147369385, + "num_tokens": 150636420.0, + "step": 5896 + }, + { + "epoch": 0.6475949923127609, + "grad_norm": 2.3984971046447754, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7280300259590149, + "num_tokens": 150658569.0, + "step": 5897 + }, + { + "epoch": 0.6477048100153745, + "grad_norm": 2.150885581970215, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6966440081596375, + "num_tokens": 150684429.0, + "step": 5898 + }, + { + "epoch": 0.6478146277179881, + "grad_norm": 2.0388848781585693, + "learning_rate": 1e-06, + "loss": 1.0741, + "mean_token_accuracy": 0.678033709526062, + "num_tokens": 150714750.0, + "step": 5899 + }, + { + "epoch": 0.6479244454206018, + "grad_norm": 1.9596281051635742, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7097563743591309, + "num_tokens": 150744961.0, + "step": 5900 + }, + { + "epoch": 0.6480342631232154, + "grad_norm": 2.416628837585449, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7182320356369019, + "num_tokens": 150765354.0, + "step": 5901 + }, + { + "epoch": 0.6481440808258291, + "grad_norm": 2.119363784790039, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.7014191746711731, + "num_tokens": 150792219.0, + "step": 5902 + }, + { + "epoch": 0.6482538985284428, + "grad_norm": 2.170131206512451, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7163543701171875, + "num_tokens": 150818456.0, + "step": 5903 + }, + { + "epoch": 0.6483637162310565, + "grad_norm": 1.829807162284851, + "learning_rate": 1e-06, + "loss": 1.0598, + "mean_token_accuracy": 0.679022490978241, + "num_tokens": 150855590.0, + "step": 5904 + }, + { + "epoch": 0.6484735339336701, + "grad_norm": 2.1169047355651855, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6973933577537537, + "num_tokens": 150882399.0, + "step": 5905 + }, + { + "epoch": 0.6485833516362838, + "grad_norm": 2.460813045501709, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7199028730392456, + "num_tokens": 150901936.0, + "step": 5906 + }, + { + "epoch": 0.6486931693388974, + "grad_norm": 2.2824418544769287, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6928784847259521, + "num_tokens": 150928192.0, + "step": 5907 + }, + { + "epoch": 0.6488029870415111, + "grad_norm": 2.1973438262939453, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7166988253593445, + "num_tokens": 150952125.0, + "step": 5908 + }, + { + "epoch": 0.6489128047441247, + "grad_norm": 2.1564648151397705, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7096543908119202, + "num_tokens": 150978580.0, + "step": 5909 + }, + { + "epoch": 0.6490226224467384, + "grad_norm": 2.183302640914917, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7237579822540283, + "num_tokens": 151003808.0, + "step": 5910 + }, + { + "epoch": 0.6491324401493521, + "grad_norm": 2.2313811779022217, + "learning_rate": 1e-06, + "loss": 1.0395, + "mean_token_accuracy": 0.6818822622299194, + "num_tokens": 151030632.0, + "step": 5911 + }, + { + "epoch": 0.6492422578519658, + "grad_norm": 2.2445127964019775, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.6936566233634949, + "num_tokens": 151054718.0, + "step": 5912 + }, + { + "epoch": 0.6493520755545794, + "grad_norm": 1.7927604913711548, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7057604789733887, + "num_tokens": 151088498.0, + "step": 5913 + }, + { + "epoch": 0.6494618932571931, + "grad_norm": 1.917944312095642, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7149772644042969, + "num_tokens": 151118513.0, + "step": 5914 + }, + { + "epoch": 0.6495717109598067, + "grad_norm": 2.206993818283081, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7262881994247437, + "num_tokens": 151144624.0, + "step": 5915 + }, + { + "epoch": 0.6496815286624203, + "grad_norm": 2.1882412433624268, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7040632963180542, + "num_tokens": 151168944.0, + "step": 5916 + }, + { + "epoch": 0.649791346365034, + "grad_norm": 1.9885777235031128, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6985965967178345, + "num_tokens": 151198340.0, + "step": 5917 + }, + { + "epoch": 0.6499011640676478, + "grad_norm": 2.157113790512085, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.698495626449585, + "num_tokens": 151226375.0, + "step": 5918 + }, + { + "epoch": 0.6500109817702614, + "grad_norm": 2.5107932090759277, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7134559154510498, + "num_tokens": 151247518.0, + "step": 5919 + }, + { + "epoch": 0.650120799472875, + "grad_norm": 2.155686616897583, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.705106258392334, + "num_tokens": 151273272.0, + "step": 5920 + }, + { + "epoch": 0.6502306171754887, + "grad_norm": 2.287026882171631, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.71236252784729, + "num_tokens": 151298002.0, + "step": 5921 + }, + { + "epoch": 0.6503404348781023, + "grad_norm": 2.152944803237915, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6925559639930725, + "num_tokens": 151324372.0, + "step": 5922 + }, + { + "epoch": 0.650450252580716, + "grad_norm": 2.0678138732910156, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.701461672782898, + "num_tokens": 151353214.0, + "step": 5923 + }, + { + "epoch": 0.6505600702833296, + "grad_norm": 2.2685039043426514, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7349935173988342, + "num_tokens": 151375006.0, + "step": 5924 + }, + { + "epoch": 0.6506698879859434, + "grad_norm": 2.161036491394043, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6988421678543091, + "num_tokens": 151400526.0, + "step": 5925 + }, + { + "epoch": 0.650779705688557, + "grad_norm": 2.268106698989868, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.696394145488739, + "num_tokens": 151425866.0, + "step": 5926 + }, + { + "epoch": 0.6508895233911707, + "grad_norm": 1.938002109527588, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.707908570766449, + "num_tokens": 151455915.0, + "step": 5927 + }, + { + "epoch": 0.6509993410937843, + "grad_norm": 2.063324451446533, + "learning_rate": 1e-06, + "loss": 1.0477, + "mean_token_accuracy": 0.6890603303909302, + "num_tokens": 151484245.0, + "step": 5928 + }, + { + "epoch": 0.651109158796398, + "grad_norm": 2.020638942718506, + "learning_rate": 1e-06, + "loss": 1.0497, + "mean_token_accuracy": 0.6827914714813232, + "num_tokens": 151513884.0, + "step": 5929 + }, + { + "epoch": 0.6512189764990116, + "grad_norm": 2.0889203548431396, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6928184032440186, + "num_tokens": 151540783.0, + "step": 5930 + }, + { + "epoch": 0.6513287942016253, + "grad_norm": 2.1929097175598145, + "learning_rate": 1e-06, + "loss": 1.0979, + "mean_token_accuracy": 0.6699120402336121, + "num_tokens": 151568422.0, + "step": 5931 + }, + { + "epoch": 0.651438611904239, + "grad_norm": 1.9407713413238525, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6972042918205261, + "num_tokens": 151599051.0, + "step": 5932 + }, + { + "epoch": 0.6515484296068527, + "grad_norm": 2.553908586502075, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7148590683937073, + "num_tokens": 151617484.0, + "step": 5933 + }, + { + "epoch": 0.6516582473094663, + "grad_norm": 1.9769688844680786, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7062016129493713, + "num_tokens": 151646293.0, + "step": 5934 + }, + { + "epoch": 0.65176806501208, + "grad_norm": 2.1469342708587646, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.7031017541885376, + "num_tokens": 151672163.0, + "step": 5935 + }, + { + "epoch": 0.6518778827146936, + "grad_norm": 2.5247082710266113, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7318594455718994, + "num_tokens": 151690680.0, + "step": 5936 + }, + { + "epoch": 0.6519877004173072, + "grad_norm": 2.210663318634033, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7092429399490356, + "num_tokens": 151714132.0, + "step": 5937 + }, + { + "epoch": 0.6520975181199209, + "grad_norm": 2.1685614585876465, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.6869465112686157, + "num_tokens": 151741255.0, + "step": 5938 + }, + { + "epoch": 0.6522073358225345, + "grad_norm": 2.612447738647461, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.727107048034668, + "num_tokens": 151759370.0, + "step": 5939 + }, + { + "epoch": 0.6523171535251483, + "grad_norm": 2.2032551765441895, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6980903744697571, + "num_tokens": 151785532.0, + "step": 5940 + }, + { + "epoch": 0.6524269712277619, + "grad_norm": 2.234509229660034, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7122021913528442, + "num_tokens": 151812342.0, + "step": 5941 + }, + { + "epoch": 0.6525367889303756, + "grad_norm": 2.4683191776275635, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.6821488738059998, + "num_tokens": 151834252.0, + "step": 5942 + }, + { + "epoch": 0.6526466066329892, + "grad_norm": 2.202834129333496, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6886754035949707, + "num_tokens": 151861843.0, + "step": 5943 + }, + { + "epoch": 0.6527564243356029, + "grad_norm": 2.1010184288024902, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7064191102981567, + "num_tokens": 151889851.0, + "step": 5944 + }, + { + "epoch": 0.6528662420382165, + "grad_norm": 2.034092664718628, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7076223492622375, + "num_tokens": 151916662.0, + "step": 5945 + }, + { + "epoch": 0.6529760597408302, + "grad_norm": 2.1210520267486572, + "learning_rate": 1e-06, + "loss": 1.0638, + "mean_token_accuracy": 0.6766723990440369, + "num_tokens": 151943715.0, + "step": 5946 + }, + { + "epoch": 0.6530858774434439, + "grad_norm": 2.185335397720337, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7075099945068359, + "num_tokens": 151968624.0, + "step": 5947 + }, + { + "epoch": 0.6531956951460576, + "grad_norm": 2.427105188369751, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7159725427627563, + "num_tokens": 151990994.0, + "step": 5948 + }, + { + "epoch": 0.6533055128486712, + "grad_norm": 2.151240110397339, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7034449577331543, + "num_tokens": 152016141.0, + "step": 5949 + }, + { + "epoch": 0.6534153305512849, + "grad_norm": 2.4134719371795654, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.704824686050415, + "num_tokens": 152040653.0, + "step": 5950 + }, + { + "epoch": 0.6535251482538985, + "grad_norm": 2.533926010131836, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7201880216598511, + "num_tokens": 152061487.0, + "step": 5951 + }, + { + "epoch": 0.6536349659565122, + "grad_norm": 2.126344919204712, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7173068523406982, + "num_tokens": 152088626.0, + "step": 5952 + }, + { + "epoch": 0.6537447836591258, + "grad_norm": 2.1804895401000977, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7222346067428589, + "num_tokens": 152114151.0, + "step": 5953 + }, + { + "epoch": 0.6538546013617396, + "grad_norm": 2.1562819480895996, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7036293148994446, + "num_tokens": 152139879.0, + "step": 5954 + }, + { + "epoch": 0.6539644190643532, + "grad_norm": 2.1814463138580322, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7102134227752686, + "num_tokens": 152166286.0, + "step": 5955 + }, + { + "epoch": 0.6540742367669669, + "grad_norm": 2.491809368133545, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7145041227340698, + "num_tokens": 152186540.0, + "step": 5956 + }, + { + "epoch": 0.6541840544695805, + "grad_norm": 2.1528592109680176, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.731096625328064, + "num_tokens": 152210448.0, + "step": 5957 + }, + { + "epoch": 0.6542938721721941, + "grad_norm": 2.0578370094299316, + "learning_rate": 1e-06, + "loss": 1.0899, + "mean_token_accuracy": 0.6823695302009583, + "num_tokens": 152237649.0, + "step": 5958 + }, + { + "epoch": 0.6544036898748078, + "grad_norm": 2.2099077701568604, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7027732133865356, + "num_tokens": 152262956.0, + "step": 5959 + }, + { + "epoch": 0.6545135075774214, + "grad_norm": 2.530015230178833, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7354342341423035, + "num_tokens": 152281540.0, + "step": 5960 + }, + { + "epoch": 0.6546233252800352, + "grad_norm": 2.300812244415283, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6990972757339478, + "num_tokens": 152306922.0, + "step": 5961 + }, + { + "epoch": 0.6547331429826488, + "grad_norm": 2.093808650970459, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.6978399753570557, + "num_tokens": 152334840.0, + "step": 5962 + }, + { + "epoch": 0.6548429606852625, + "grad_norm": 2.1661646366119385, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7168624997138977, + "num_tokens": 152361321.0, + "step": 5963 + }, + { + "epoch": 0.6549527783878761, + "grad_norm": 2.1930763721466064, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6897377371788025, + "num_tokens": 152385645.0, + "step": 5964 + }, + { + "epoch": 0.6550625960904898, + "grad_norm": 1.9060391187667847, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7036212682723999, + "num_tokens": 152418444.0, + "step": 5965 + }, + { + "epoch": 0.6551724137931034, + "grad_norm": 2.131535768508911, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7161637544631958, + "num_tokens": 152444524.0, + "step": 5966 + }, + { + "epoch": 0.6552822314957171, + "grad_norm": 2.0143353939056396, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.705919086933136, + "num_tokens": 152471363.0, + "step": 5967 + }, + { + "epoch": 0.6553920491983307, + "grad_norm": 2.1902694702148438, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.7044155597686768, + "num_tokens": 152496702.0, + "step": 5968 + }, + { + "epoch": 0.6555018669009445, + "grad_norm": 2.3323845863342285, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7311123013496399, + "num_tokens": 152518843.0, + "step": 5969 + }, + { + "epoch": 0.6556116846035581, + "grad_norm": 2.459808588027954, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7152410745620728, + "num_tokens": 152537479.0, + "step": 5970 + }, + { + "epoch": 0.6557215023061718, + "grad_norm": 2.422618865966797, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.707720160484314, + "num_tokens": 152560458.0, + "step": 5971 + }, + { + "epoch": 0.6558313200087854, + "grad_norm": 2.1309139728546143, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7301751375198364, + "num_tokens": 152586360.0, + "step": 5972 + }, + { + "epoch": 0.6559411377113991, + "grad_norm": 2.2152228355407715, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.699884295463562, + "num_tokens": 152611887.0, + "step": 5973 + }, + { + "epoch": 0.6560509554140127, + "grad_norm": 2.2053234577178955, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.6999281048774719, + "num_tokens": 152637128.0, + "step": 5974 + }, + { + "epoch": 0.6561607731166264, + "grad_norm": 2.1810643672943115, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.710861086845398, + "num_tokens": 152661889.0, + "step": 5975 + }, + { + "epoch": 0.6562705908192401, + "grad_norm": 2.3436827659606934, + "learning_rate": 1e-06, + "loss": 1.0417, + "mean_token_accuracy": 0.6861920356750488, + "num_tokens": 152685954.0, + "step": 5976 + }, + { + "epoch": 0.6563804085218538, + "grad_norm": 2.0644733905792236, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7271587252616882, + "num_tokens": 152711560.0, + "step": 5977 + }, + { + "epoch": 0.6564902262244674, + "grad_norm": 2.0432229042053223, + "learning_rate": 1e-06, + "loss": 1.0364, + "mean_token_accuracy": 0.6860231161117554, + "num_tokens": 152739730.0, + "step": 5978 + }, + { + "epoch": 0.656600043927081, + "grad_norm": 2.077796220779419, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7053523659706116, + "num_tokens": 152767225.0, + "step": 5979 + }, + { + "epoch": 0.6567098616296947, + "grad_norm": 2.14888334274292, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.725077748298645, + "num_tokens": 152792075.0, + "step": 5980 + }, + { + "epoch": 0.6568196793323083, + "grad_norm": 2.467768907546997, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.6991146802902222, + "num_tokens": 152813156.0, + "step": 5981 + }, + { + "epoch": 0.656929497034922, + "grad_norm": 2.1481127738952637, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7230280637741089, + "num_tokens": 152837588.0, + "step": 5982 + }, + { + "epoch": 0.6570393147375357, + "grad_norm": 2.050286054611206, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7080379724502563, + "num_tokens": 152865545.0, + "step": 5983 + }, + { + "epoch": 0.6571491324401494, + "grad_norm": 2.34747314453125, + "learning_rate": 1e-06, + "loss": 1.0558, + "mean_token_accuracy": 0.694131076335907, + "num_tokens": 152888099.0, + "step": 5984 + }, + { + "epoch": 0.657258950142763, + "grad_norm": 2.3011844158172607, + "learning_rate": 1e-06, + "loss": 1.0624, + "mean_token_accuracy": 0.6793493628501892, + "num_tokens": 152912803.0, + "step": 5985 + }, + { + "epoch": 0.6573687678453767, + "grad_norm": 2.6443653106689453, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.733549952507019, + "num_tokens": 152931337.0, + "step": 5986 + }, + { + "epoch": 0.6574785855479903, + "grad_norm": 2.1684746742248535, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7240762114524841, + "num_tokens": 152956883.0, + "step": 5987 + }, + { + "epoch": 0.657588403250604, + "grad_norm": 2.2252845764160156, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7243447303771973, + "num_tokens": 152981070.0, + "step": 5988 + }, + { + "epoch": 0.6576982209532176, + "grad_norm": 2.377110719680786, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7087651491165161, + "num_tokens": 153002807.0, + "step": 5989 + }, + { + "epoch": 0.6578080386558314, + "grad_norm": 2.413799524307251, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7115853428840637, + "num_tokens": 153024255.0, + "step": 5990 + }, + { + "epoch": 0.657917856358445, + "grad_norm": 2.4092962741851807, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.6992840766906738, + "num_tokens": 153047909.0, + "step": 5991 + }, + { + "epoch": 0.6580276740610587, + "grad_norm": 2.1788384914398193, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6962651014328003, + "num_tokens": 153073052.0, + "step": 5992 + }, + { + "epoch": 0.6581374917636723, + "grad_norm": 2.1574723720550537, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7498881220817566, + "num_tokens": 153097621.0, + "step": 5993 + }, + { + "epoch": 0.658247309466286, + "grad_norm": 2.471769094467163, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7091560363769531, + "num_tokens": 153117066.0, + "step": 5994 + }, + { + "epoch": 0.6583571271688996, + "grad_norm": 2.1339221000671387, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7057725191116333, + "num_tokens": 153143046.0, + "step": 5995 + }, + { + "epoch": 0.6584669448715132, + "grad_norm": 2.0735023021698, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7015987634658813, + "num_tokens": 153170815.0, + "step": 5996 + }, + { + "epoch": 0.6585767625741269, + "grad_norm": 2.1590492725372314, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.726030707359314, + "num_tokens": 153196227.0, + "step": 5997 + }, + { + "epoch": 0.6586865802767407, + "grad_norm": 2.3535661697387695, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.7053321003913879, + "num_tokens": 153218605.0, + "step": 5998 + }, + { + "epoch": 0.6587963979793543, + "grad_norm": 1.9089939594268799, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7112763524055481, + "num_tokens": 153250639.0, + "step": 5999 + }, + { + "epoch": 0.6589062156819679, + "grad_norm": 1.9681202173233032, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7205221056938171, + "num_tokens": 153277278.0, + "step": 6000 + }, + { + "epoch": 0.6590160333845816, + "grad_norm": 2.1471810340881348, + "learning_rate": 1e-06, + "loss": 1.067, + "mean_token_accuracy": 0.6821860074996948, + "num_tokens": 153303388.0, + "step": 6001 + }, + { + "epoch": 0.6591258510871952, + "grad_norm": 2.271876335144043, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7083595991134644, + "num_tokens": 153326081.0, + "step": 6002 + }, + { + "epoch": 0.6592356687898089, + "grad_norm": 2.407618522644043, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7137668132781982, + "num_tokens": 153347239.0, + "step": 6003 + }, + { + "epoch": 0.6593454864924225, + "grad_norm": 2.4037373065948486, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7048197388648987, + "num_tokens": 153368004.0, + "step": 6004 + }, + { + "epoch": 0.6594553041950363, + "grad_norm": 2.206205129623413, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7168591022491455, + "num_tokens": 153391368.0, + "step": 6005 + }, + { + "epoch": 0.6595651218976499, + "grad_norm": 2.313328266143799, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.72687166929245, + "num_tokens": 153412993.0, + "step": 6006 + }, + { + "epoch": 0.6596749396002636, + "grad_norm": 2.181762933731079, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7040262222290039, + "num_tokens": 153438215.0, + "step": 6007 + }, + { + "epoch": 0.6597847573028772, + "grad_norm": 2.1271369457244873, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.6958696246147156, + "num_tokens": 153466044.0, + "step": 6008 + }, + { + "epoch": 0.6598945750054909, + "grad_norm": 2.464287042617798, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7110469341278076, + "num_tokens": 153487304.0, + "step": 6009 + }, + { + "epoch": 0.6600043927081045, + "grad_norm": 2.3260414600372314, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6973596811294556, + "num_tokens": 153510889.0, + "step": 6010 + }, + { + "epoch": 0.6601142104107182, + "grad_norm": 2.0643482208251953, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7225850820541382, + "num_tokens": 153536502.0, + "step": 6011 + }, + { + "epoch": 0.6602240281133319, + "grad_norm": 2.2589797973632812, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6890940070152283, + "num_tokens": 153560742.0, + "step": 6012 + }, + { + "epoch": 0.6603338458159456, + "grad_norm": 2.401618003845215, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7084052562713623, + "num_tokens": 153584179.0, + "step": 6013 + }, + { + "epoch": 0.6604436635185592, + "grad_norm": 2.0477688312530518, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6889115571975708, + "num_tokens": 153616324.0, + "step": 6014 + }, + { + "epoch": 0.6605534812211729, + "grad_norm": 2.225379705429077, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.6997841000556946, + "num_tokens": 153641935.0, + "step": 6015 + }, + { + "epoch": 0.6606632989237865, + "grad_norm": 2.2745556831359863, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6935884952545166, + "num_tokens": 153667442.0, + "step": 6016 + }, + { + "epoch": 0.6607731166264001, + "grad_norm": 2.3681938648223877, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.7044793367385864, + "num_tokens": 153689436.0, + "step": 6017 + }, + { + "epoch": 0.6608829343290138, + "grad_norm": 2.1346144676208496, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7096469402313232, + "num_tokens": 153713709.0, + "step": 6018 + }, + { + "epoch": 0.6609927520316276, + "grad_norm": 2.3379158973693848, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7187818288803101, + "num_tokens": 153735567.0, + "step": 6019 + }, + { + "epoch": 0.6611025697342412, + "grad_norm": 1.9309465885162354, + "learning_rate": 1e-06, + "loss": 1.1006, + "mean_token_accuracy": 0.6743876934051514, + "num_tokens": 153770327.0, + "step": 6020 + }, + { + "epoch": 0.6612123874368548, + "grad_norm": 2.013383626937866, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7062018513679504, + "num_tokens": 153798489.0, + "step": 6021 + }, + { + "epoch": 0.6613222051394685, + "grad_norm": 2.2171099185943604, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7242639064788818, + "num_tokens": 153822226.0, + "step": 6022 + }, + { + "epoch": 0.6614320228420821, + "grad_norm": 2.2735793590545654, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7100944519042969, + "num_tokens": 153846182.0, + "step": 6023 + }, + { + "epoch": 0.6615418405446958, + "grad_norm": 2.2266995906829834, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7008963823318481, + "num_tokens": 153873057.0, + "step": 6024 + }, + { + "epoch": 0.6616516582473094, + "grad_norm": 2.2498738765716553, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7272006869316101, + "num_tokens": 153896042.0, + "step": 6025 + }, + { + "epoch": 0.6617614759499231, + "grad_norm": 2.1313421726226807, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7180015444755554, + "num_tokens": 153921546.0, + "step": 6026 + }, + { + "epoch": 0.6618712936525368, + "grad_norm": 2.2030012607574463, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.7093170881271362, + "num_tokens": 153947147.0, + "step": 6027 + }, + { + "epoch": 0.6619811113551505, + "grad_norm": 2.403920888900757, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.695306658744812, + "num_tokens": 153969768.0, + "step": 6028 + }, + { + "epoch": 0.6620909290577641, + "grad_norm": 2.198991060256958, + "learning_rate": 1e-06, + "loss": 1.0919, + "mean_token_accuracy": 0.674858033657074, + "num_tokens": 153997183.0, + "step": 6029 + }, + { + "epoch": 0.6622007467603778, + "grad_norm": 2.124990701675415, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.6994251012802124, + "num_tokens": 154023684.0, + "step": 6030 + }, + { + "epoch": 0.6623105644629914, + "grad_norm": 2.434633731842041, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7206087112426758, + "num_tokens": 154043534.0, + "step": 6031 + }, + { + "epoch": 0.6624203821656051, + "grad_norm": 2.226276159286499, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7148098945617676, + "num_tokens": 154067368.0, + "step": 6032 + }, + { + "epoch": 0.6625301998682187, + "grad_norm": 2.245049238204956, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6895362138748169, + "num_tokens": 154090829.0, + "step": 6033 + }, + { + "epoch": 0.6626400175708325, + "grad_norm": 1.8400084972381592, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7111499309539795, + "num_tokens": 154124640.0, + "step": 6034 + }, + { + "epoch": 0.6627498352734461, + "grad_norm": 2.280217409133911, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6991257667541504, + "num_tokens": 154149208.0, + "step": 6035 + }, + { + "epoch": 0.6628596529760598, + "grad_norm": 2.1157586574554443, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7284021377563477, + "num_tokens": 154175292.0, + "step": 6036 + }, + { + "epoch": 0.6629694706786734, + "grad_norm": 2.2032227516174316, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.6966663599014282, + "num_tokens": 154201086.0, + "step": 6037 + }, + { + "epoch": 0.663079288381287, + "grad_norm": 1.9802970886230469, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7266022562980652, + "num_tokens": 154229422.0, + "step": 6038 + }, + { + "epoch": 0.6631891060839007, + "grad_norm": 2.343484401702881, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7360695004463196, + "num_tokens": 154250960.0, + "step": 6039 + }, + { + "epoch": 0.6632989237865143, + "grad_norm": 2.1291964054107666, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.6842184066772461, + "num_tokens": 154279864.0, + "step": 6040 + }, + { + "epoch": 0.6634087414891281, + "grad_norm": 2.4643962383270264, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7206820249557495, + "num_tokens": 154298713.0, + "step": 6041 + }, + { + "epoch": 0.6635185591917417, + "grad_norm": 2.073906183242798, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.712734043598175, + "num_tokens": 154328960.0, + "step": 6042 + }, + { + "epoch": 0.6636283768943554, + "grad_norm": 2.5940518379211426, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7267458438873291, + "num_tokens": 154347501.0, + "step": 6043 + }, + { + "epoch": 0.663738194596969, + "grad_norm": 2.43218994140625, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7061546444892883, + "num_tokens": 154368872.0, + "step": 6044 + }, + { + "epoch": 0.6638480122995827, + "grad_norm": 2.3969311714172363, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7325326204299927, + "num_tokens": 154390294.0, + "step": 6045 + }, + { + "epoch": 0.6639578300021963, + "grad_norm": 2.149315595626831, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.7101208567619324, + "num_tokens": 154417509.0, + "step": 6046 + }, + { + "epoch": 0.66406764770481, + "grad_norm": 2.268160820007324, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7191798686981201, + "num_tokens": 154440887.0, + "step": 6047 + }, + { + "epoch": 0.6641774654074237, + "grad_norm": 2.090155601501465, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7035605907440186, + "num_tokens": 154468917.0, + "step": 6048 + }, + { + "epoch": 0.6642872831100374, + "grad_norm": 2.851801633834839, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7231204509735107, + "num_tokens": 154486074.0, + "step": 6049 + }, + { + "epoch": 0.664397100812651, + "grad_norm": 2.1403841972351074, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6930480599403381, + "num_tokens": 154511938.0, + "step": 6050 + }, + { + "epoch": 0.6645069185152647, + "grad_norm": 2.186131000518799, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7148205041885376, + "num_tokens": 154537483.0, + "step": 6051 + }, + { + "epoch": 0.6646167362178783, + "grad_norm": 2.0205564498901367, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7018332481384277, + "num_tokens": 154566543.0, + "step": 6052 + }, + { + "epoch": 0.664726553920492, + "grad_norm": 2.102952003479004, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7150452136993408, + "num_tokens": 154592402.0, + "step": 6053 + }, + { + "epoch": 0.6648363716231056, + "grad_norm": 2.1503384113311768, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7222853899002075, + "num_tokens": 154615437.0, + "step": 6054 + }, + { + "epoch": 0.6649461893257194, + "grad_norm": 2.237823009490967, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.6842125058174133, + "num_tokens": 154642592.0, + "step": 6055 + }, + { + "epoch": 0.665056007028333, + "grad_norm": 1.8567218780517578, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6996757984161377, + "num_tokens": 154677881.0, + "step": 6056 + }, + { + "epoch": 0.6651658247309467, + "grad_norm": 2.5141265392303467, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7257390022277832, + "num_tokens": 154696451.0, + "step": 6057 + }, + { + "epoch": 0.6652756424335603, + "grad_norm": 2.297435760498047, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7249814867973328, + "num_tokens": 154717487.0, + "step": 6058 + }, + { + "epoch": 0.665385460136174, + "grad_norm": 2.2364823818206787, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.709377110004425, + "num_tokens": 154740863.0, + "step": 6059 + }, + { + "epoch": 0.6654952778387876, + "grad_norm": 2.230696201324463, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7269516587257385, + "num_tokens": 154763069.0, + "step": 6060 + }, + { + "epoch": 0.6656050955414012, + "grad_norm": 2.0607242584228516, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7248560786247253, + "num_tokens": 154789596.0, + "step": 6061 + }, + { + "epoch": 0.6657149132440149, + "grad_norm": 1.9621999263763428, + "learning_rate": 1e-06, + "loss": 1.0518, + "mean_token_accuracy": 0.6789760589599609, + "num_tokens": 154820076.0, + "step": 6062 + }, + { + "epoch": 0.6658247309466286, + "grad_norm": 2.141902446746826, + "learning_rate": 1e-06, + "loss": 1.0412, + "mean_token_accuracy": 0.6908255815505981, + "num_tokens": 154847298.0, + "step": 6063 + }, + { + "epoch": 0.6659345486492423, + "grad_norm": 2.3239665031433105, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7134606242179871, + "num_tokens": 154870117.0, + "step": 6064 + }, + { + "epoch": 0.6660443663518559, + "grad_norm": 2.407005786895752, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7313975095748901, + "num_tokens": 154892142.0, + "step": 6065 + }, + { + "epoch": 0.6661541840544696, + "grad_norm": 2.249483108520508, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7278064489364624, + "num_tokens": 154916270.0, + "step": 6066 + }, + { + "epoch": 0.6662640017570832, + "grad_norm": 2.3034770488739014, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.70904141664505, + "num_tokens": 154939590.0, + "step": 6067 + }, + { + "epoch": 0.6663738194596969, + "grad_norm": 2.4825708866119385, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7388451099395752, + "num_tokens": 154960078.0, + "step": 6068 + }, + { + "epoch": 0.6664836371623105, + "grad_norm": 2.379516363143921, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.6883935928344727, + "num_tokens": 154983054.0, + "step": 6069 + }, + { + "epoch": 0.6665934548649243, + "grad_norm": 1.9365719556808472, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.6986464262008667, + "num_tokens": 155012296.0, + "step": 6070 + }, + { + "epoch": 0.6667032725675379, + "grad_norm": 2.202033281326294, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7271700501441956, + "num_tokens": 155036438.0, + "step": 6071 + }, + { + "epoch": 0.6668130902701516, + "grad_norm": 1.9853324890136719, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.6966345906257629, + "num_tokens": 155067906.0, + "step": 6072 + }, + { + "epoch": 0.6669229079727652, + "grad_norm": 1.9178167581558228, + "learning_rate": 1e-06, + "loss": 1.1045, + "mean_token_accuracy": 0.6671674847602844, + "num_tokens": 155102249.0, + "step": 6073 + }, + { + "epoch": 0.6670327256753789, + "grad_norm": 2.1023378372192383, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7086280584335327, + "num_tokens": 155130437.0, + "step": 6074 + }, + { + "epoch": 0.6671425433779925, + "grad_norm": 2.1379892826080322, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.690761923789978, + "num_tokens": 155157993.0, + "step": 6075 + }, + { + "epoch": 0.6672523610806061, + "grad_norm": 2.2215209007263184, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7130696177482605, + "num_tokens": 155182774.0, + "step": 6076 + }, + { + "epoch": 0.6673621787832199, + "grad_norm": 2.0552053451538086, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7139804363250732, + "num_tokens": 155210967.0, + "step": 6077 + }, + { + "epoch": 0.6674719964858336, + "grad_norm": 2.4012460708618164, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6950278282165527, + "num_tokens": 155233769.0, + "step": 6078 + }, + { + "epoch": 0.6675818141884472, + "grad_norm": 2.222039222717285, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7074646949768066, + "num_tokens": 155257928.0, + "step": 6079 + }, + { + "epoch": 0.6676916318910608, + "grad_norm": 2.1342039108276367, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.6900697946548462, + "num_tokens": 155284411.0, + "step": 6080 + }, + { + "epoch": 0.6678014495936745, + "grad_norm": 2.2224082946777344, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7049641609191895, + "num_tokens": 155308691.0, + "step": 6081 + }, + { + "epoch": 0.6679112672962881, + "grad_norm": 2.318659543991089, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.6986161470413208, + "num_tokens": 155333316.0, + "step": 6082 + }, + { + "epoch": 0.6680210849989018, + "grad_norm": 2.394474506378174, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6971221566200256, + "num_tokens": 155356966.0, + "step": 6083 + }, + { + "epoch": 0.6681309027015155, + "grad_norm": 2.216352939605713, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.713273823261261, + "num_tokens": 155382500.0, + "step": 6084 + }, + { + "epoch": 0.6682407204041292, + "grad_norm": 2.1376516819000244, + "learning_rate": 1e-06, + "loss": 1.0278, + "mean_token_accuracy": 0.6941847801208496, + "num_tokens": 155410289.0, + "step": 6085 + }, + { + "epoch": 0.6683505381067428, + "grad_norm": 2.2890288829803467, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7155100107192993, + "num_tokens": 155434979.0, + "step": 6086 + }, + { + "epoch": 0.6684603558093565, + "grad_norm": 1.891710877418518, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.7011673450469971, + "num_tokens": 155466809.0, + "step": 6087 + }, + { + "epoch": 0.6685701735119701, + "grad_norm": 2.0548481941223145, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7143692970275879, + "num_tokens": 155494465.0, + "step": 6088 + }, + { + "epoch": 0.6686799912145838, + "grad_norm": 2.293729782104492, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7197211980819702, + "num_tokens": 155517530.0, + "step": 6089 + }, + { + "epoch": 0.6687898089171974, + "grad_norm": 2.005605936050415, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.6995545029640198, + "num_tokens": 155548591.0, + "step": 6090 + }, + { + "epoch": 0.6688996266198111, + "grad_norm": 2.042588472366333, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6929196119308472, + "num_tokens": 155579453.0, + "step": 6091 + }, + { + "epoch": 0.6690094443224248, + "grad_norm": 2.189714193344116, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7114464044570923, + "num_tokens": 155604753.0, + "step": 6092 + }, + { + "epoch": 0.6691192620250385, + "grad_norm": 2.322099208831787, + "learning_rate": 1e-06, + "loss": 1.0327, + "mean_token_accuracy": 0.6899117827415466, + "num_tokens": 155628737.0, + "step": 6093 + }, + { + "epoch": 0.6692290797276521, + "grad_norm": 2.0131757259368896, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7225783467292786, + "num_tokens": 155656265.0, + "step": 6094 + }, + { + "epoch": 0.6693388974302658, + "grad_norm": 2.4561359882354736, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7344862222671509, + "num_tokens": 155675951.0, + "step": 6095 + }, + { + "epoch": 0.6694487151328794, + "grad_norm": 2.1136021614074707, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.689940869808197, + "num_tokens": 155705690.0, + "step": 6096 + }, + { + "epoch": 0.669558532835493, + "grad_norm": 2.0553512573242188, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7098829746246338, + "num_tokens": 155733287.0, + "step": 6097 + }, + { + "epoch": 0.6696683505381067, + "grad_norm": 1.9939733743667603, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.6974586844444275, + "num_tokens": 155763126.0, + "step": 6098 + }, + { + "epoch": 0.6697781682407205, + "grad_norm": 2.0259740352630615, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.6886336207389832, + "num_tokens": 155793419.0, + "step": 6099 + }, + { + "epoch": 0.6698879859433341, + "grad_norm": 2.2098772525787354, + "learning_rate": 1e-06, + "loss": 1.0751, + "mean_token_accuracy": 0.6690502762794495, + "num_tokens": 155820601.0, + "step": 6100 + }, + { + "epoch": 0.6699978036459477, + "grad_norm": 2.0622670650482178, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6966657042503357, + "num_tokens": 155847813.0, + "step": 6101 + }, + { + "epoch": 0.6701076213485614, + "grad_norm": 2.3237640857696533, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7207838296890259, + "num_tokens": 155870204.0, + "step": 6102 + }, + { + "epoch": 0.670217439051175, + "grad_norm": 2.3382341861724854, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7174150347709656, + "num_tokens": 155892576.0, + "step": 6103 + }, + { + "epoch": 0.6703272567537887, + "grad_norm": 2.195349931716919, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.704937219619751, + "num_tokens": 155917704.0, + "step": 6104 + }, + { + "epoch": 0.6704370744564023, + "grad_norm": 2.5023181438446045, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7431040406227112, + "num_tokens": 155938130.0, + "step": 6105 + }, + { + "epoch": 0.6705468921590161, + "grad_norm": 2.3903911113739014, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.714271068572998, + "num_tokens": 155958290.0, + "step": 6106 + }, + { + "epoch": 0.6706567098616297, + "grad_norm": 2.410451889038086, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7201484441757202, + "num_tokens": 155979980.0, + "step": 6107 + }, + { + "epoch": 0.6707665275642434, + "grad_norm": 2.3255932331085205, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7032527923583984, + "num_tokens": 156002711.0, + "step": 6108 + }, + { + "epoch": 0.670876345266857, + "grad_norm": 2.229863166809082, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6955872774124146, + "num_tokens": 156027448.0, + "step": 6109 + }, + { + "epoch": 0.6709861629694707, + "grad_norm": 2.4153568744659424, + "learning_rate": 1e-06, + "loss": 1.0704, + "mean_token_accuracy": 0.6845367550849915, + "num_tokens": 156049633.0, + "step": 6110 + }, + { + "epoch": 0.6710959806720843, + "grad_norm": 2.0458555221557617, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.6961929202079773, + "num_tokens": 156078085.0, + "step": 6111 + }, + { + "epoch": 0.671205798374698, + "grad_norm": 2.1753616333007812, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7012243270874023, + "num_tokens": 156105704.0, + "step": 6112 + }, + { + "epoch": 0.6713156160773117, + "grad_norm": 2.1727395057678223, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6927002668380737, + "num_tokens": 156132041.0, + "step": 6113 + }, + { + "epoch": 0.6714254337799254, + "grad_norm": 2.3443820476531982, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7181453108787537, + "num_tokens": 156153599.0, + "step": 6114 + }, + { + "epoch": 0.671535251482539, + "grad_norm": 2.3931884765625, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7260537147521973, + "num_tokens": 156174874.0, + "step": 6115 + }, + { + "epoch": 0.6716450691851527, + "grad_norm": 2.139211416244507, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7203171849250793, + "num_tokens": 156199830.0, + "step": 6116 + }, + { + "epoch": 0.6717548868877663, + "grad_norm": 2.459322690963745, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7220394611358643, + "num_tokens": 156219613.0, + "step": 6117 + }, + { + "epoch": 0.67186470459038, + "grad_norm": 2.0917279720306396, + "learning_rate": 1e-06, + "loss": 1.0525, + "mean_token_accuracy": 0.6979391574859619, + "num_tokens": 156248736.0, + "step": 6118 + }, + { + "epoch": 0.6719745222929936, + "grad_norm": 2.1493849754333496, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7051529288291931, + "num_tokens": 156273541.0, + "step": 6119 + }, + { + "epoch": 0.6720843399956072, + "grad_norm": 1.9029488563537598, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.710950493812561, + "num_tokens": 156305356.0, + "step": 6120 + }, + { + "epoch": 0.672194157698221, + "grad_norm": 2.135772466659546, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7113159894943237, + "num_tokens": 156331327.0, + "step": 6121 + }, + { + "epoch": 0.6723039754008346, + "grad_norm": 2.1898181438446045, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7173061370849609, + "num_tokens": 156354900.0, + "step": 6122 + }, + { + "epoch": 0.6724137931034483, + "grad_norm": 2.2539784908294678, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.7016380429267883, + "num_tokens": 156380537.0, + "step": 6123 + }, + { + "epoch": 0.6725236108060619, + "grad_norm": 2.4263250827789307, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.739144504070282, + "num_tokens": 156399540.0, + "step": 6124 + }, + { + "epoch": 0.6726334285086756, + "grad_norm": 2.0954620838165283, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7028274536132812, + "num_tokens": 156426925.0, + "step": 6125 + }, + { + "epoch": 0.6727432462112892, + "grad_norm": 2.526057481765747, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7264702320098877, + "num_tokens": 156445531.0, + "step": 6126 + }, + { + "epoch": 0.6728530639139029, + "grad_norm": 2.1960437297821045, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7291796803474426, + "num_tokens": 156469199.0, + "step": 6127 + }, + { + "epoch": 0.6729628816165166, + "grad_norm": 2.390604257583618, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7324321866035461, + "num_tokens": 156490610.0, + "step": 6128 + }, + { + "epoch": 0.6730726993191303, + "grad_norm": 2.040073871612549, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7066606283187866, + "num_tokens": 156518469.0, + "step": 6129 + }, + { + "epoch": 0.6731825170217439, + "grad_norm": 2.2512547969818115, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.6956813335418701, + "num_tokens": 156543440.0, + "step": 6130 + }, + { + "epoch": 0.6732923347243576, + "grad_norm": 2.209367275238037, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7440721988677979, + "num_tokens": 156566157.0, + "step": 6131 + }, + { + "epoch": 0.6734021524269712, + "grad_norm": 2.301320791244507, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.7013406157493591, + "num_tokens": 156589963.0, + "step": 6132 + }, + { + "epoch": 0.6735119701295849, + "grad_norm": 2.611280918121338, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7201405763626099, + "num_tokens": 156608660.0, + "step": 6133 + }, + { + "epoch": 0.6736217878321985, + "grad_norm": 2.1433162689208984, + "learning_rate": 1e-06, + "loss": 1.0426, + "mean_token_accuracy": 0.6928200125694275, + "num_tokens": 156637288.0, + "step": 6134 + }, + { + "epoch": 0.6737316055348123, + "grad_norm": 2.1143689155578613, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7198056578636169, + "num_tokens": 156664703.0, + "step": 6135 + }, + { + "epoch": 0.6738414232374259, + "grad_norm": 2.4030537605285645, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7072447538375854, + "num_tokens": 156688222.0, + "step": 6136 + }, + { + "epoch": 0.6739512409400396, + "grad_norm": 2.461172103881836, + "learning_rate": 1e-06, + "loss": 1.0223, + "mean_token_accuracy": 0.6842929720878601, + "num_tokens": 156709366.0, + "step": 6137 + }, + { + "epoch": 0.6740610586426532, + "grad_norm": 2.4126322269439697, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7220926284790039, + "num_tokens": 156730532.0, + "step": 6138 + }, + { + "epoch": 0.6741708763452668, + "grad_norm": 2.253305196762085, + "learning_rate": 1e-06, + "loss": 1.0773, + "mean_token_accuracy": 0.680733323097229, + "num_tokens": 156754359.0, + "step": 6139 + }, + { + "epoch": 0.6742806940478805, + "grad_norm": 2.251363754272461, + "learning_rate": 1e-06, + "loss": 1.0281, + "mean_token_accuracy": 0.6929035186767578, + "num_tokens": 156779608.0, + "step": 6140 + }, + { + "epoch": 0.6743905117504941, + "grad_norm": 2.1802916526794434, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7310937643051147, + "num_tokens": 156805486.0, + "step": 6141 + }, + { + "epoch": 0.6745003294531079, + "grad_norm": 2.121047019958496, + "learning_rate": 1e-06, + "loss": 1.0464, + "mean_token_accuracy": 0.6834834814071655, + "num_tokens": 156835411.0, + "step": 6142 + }, + { + "epoch": 0.6746101471557215, + "grad_norm": 2.1882872581481934, + "learning_rate": 1e-06, + "loss": 1.0328, + "mean_token_accuracy": 0.68500155210495, + "num_tokens": 156860897.0, + "step": 6143 + }, + { + "epoch": 0.6747199648583352, + "grad_norm": 2.405862808227539, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7030220031738281, + "num_tokens": 156883230.0, + "step": 6144 + }, + { + "epoch": 0.6748297825609488, + "grad_norm": 2.460848093032837, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7183225750923157, + "num_tokens": 156902770.0, + "step": 6145 + }, + { + "epoch": 0.6749396002635625, + "grad_norm": 2.4071948528289795, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7101339101791382, + "num_tokens": 156925284.0, + "step": 6146 + }, + { + "epoch": 0.6750494179661761, + "grad_norm": 2.001967668533325, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.7006272673606873, + "num_tokens": 156956041.0, + "step": 6147 + }, + { + "epoch": 0.6751592356687898, + "grad_norm": 2.327752113342285, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7032364010810852, + "num_tokens": 156978346.0, + "step": 6148 + }, + { + "epoch": 0.6752690533714034, + "grad_norm": 1.9954603910446167, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7142705321311951, + "num_tokens": 157008339.0, + "step": 6149 + }, + { + "epoch": 0.6753788710740172, + "grad_norm": 2.358996868133545, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7294178009033203, + "num_tokens": 157030678.0, + "step": 6150 + }, + { + "epoch": 0.6754886887766308, + "grad_norm": 2.3048298358917236, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6913741230964661, + "num_tokens": 157054476.0, + "step": 6151 + }, + { + "epoch": 0.6755985064792445, + "grad_norm": 1.8911455869674683, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.715161144733429, + "num_tokens": 157086000.0, + "step": 6152 + }, + { + "epoch": 0.6757083241818581, + "grad_norm": 2.2729921340942383, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7053560018539429, + "num_tokens": 157109522.0, + "step": 6153 + }, + { + "epoch": 0.6758181418844718, + "grad_norm": 1.9207935333251953, + "learning_rate": 1e-06, + "loss": 1.0663, + "mean_token_accuracy": 0.6814523935317993, + "num_tokens": 157141643.0, + "step": 6154 + }, + { + "epoch": 0.6759279595870854, + "grad_norm": 2.5398330688476562, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7155652046203613, + "num_tokens": 157160949.0, + "step": 6155 + }, + { + "epoch": 0.676037777289699, + "grad_norm": 2.0915234088897705, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7061172723770142, + "num_tokens": 157187668.0, + "step": 6156 + }, + { + "epoch": 0.6761475949923128, + "grad_norm": 2.3615431785583496, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.700840950012207, + "num_tokens": 157210668.0, + "step": 6157 + }, + { + "epoch": 0.6762574126949265, + "grad_norm": 1.9472107887268066, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6901847124099731, + "num_tokens": 157243942.0, + "step": 6158 + }, + { + "epoch": 0.6763672303975401, + "grad_norm": 2.4085617065429688, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7169795036315918, + "num_tokens": 157264089.0, + "step": 6159 + }, + { + "epoch": 0.6764770481001537, + "grad_norm": 2.42520809173584, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7156368494033813, + "num_tokens": 157285308.0, + "step": 6160 + }, + { + "epoch": 0.6765868658027674, + "grad_norm": 2.195142984390259, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7018039226531982, + "num_tokens": 157310173.0, + "step": 6161 + }, + { + "epoch": 0.676696683505381, + "grad_norm": 1.9505763053894043, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7108907699584961, + "num_tokens": 157340261.0, + "step": 6162 + }, + { + "epoch": 0.6768065012079947, + "grad_norm": 2.340672492980957, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.706784725189209, + "num_tokens": 157362862.0, + "step": 6163 + }, + { + "epoch": 0.6769163189106084, + "grad_norm": 2.3357040882110596, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.718773365020752, + "num_tokens": 157386085.0, + "step": 6164 + }, + { + "epoch": 0.6770261366132221, + "grad_norm": 2.246349334716797, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6975584030151367, + "num_tokens": 157410732.0, + "step": 6165 + }, + { + "epoch": 0.6771359543158357, + "grad_norm": 2.2772574424743652, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7034890651702881, + "num_tokens": 157435218.0, + "step": 6166 + }, + { + "epoch": 0.6772457720184494, + "grad_norm": 1.8901067972183228, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7175304889678955, + "num_tokens": 157468123.0, + "step": 6167 + }, + { + "epoch": 0.677355589721063, + "grad_norm": 2.0543084144592285, + "learning_rate": 1e-06, + "loss": 1.0731, + "mean_token_accuracy": 0.6807045936584473, + "num_tokens": 157498440.0, + "step": 6168 + }, + { + "epoch": 0.6774654074236767, + "grad_norm": 1.926806926727295, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6897251605987549, + "num_tokens": 157530798.0, + "step": 6169 + }, + { + "epoch": 0.6775752251262903, + "grad_norm": 2.267841100692749, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7256671786308289, + "num_tokens": 157553636.0, + "step": 6170 + }, + { + "epoch": 0.6776850428289041, + "grad_norm": 2.151787757873535, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7169265747070312, + "num_tokens": 157578219.0, + "step": 6171 + }, + { + "epoch": 0.6777948605315177, + "grad_norm": 1.8733855485916138, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7020570039749146, + "num_tokens": 157611368.0, + "step": 6172 + }, + { + "epoch": 0.6779046782341314, + "grad_norm": 2.271700382232666, + "learning_rate": 1e-06, + "loss": 1.0485, + "mean_token_accuracy": 0.6860637664794922, + "num_tokens": 157636207.0, + "step": 6173 + }, + { + "epoch": 0.678014495936745, + "grad_norm": 2.030236005783081, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.7021253108978271, + "num_tokens": 157666098.0, + "step": 6174 + }, + { + "epoch": 0.6781243136393587, + "grad_norm": 1.9976917505264282, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.7106467485427856, + "num_tokens": 157697195.0, + "step": 6175 + }, + { + "epoch": 0.6782341313419723, + "grad_norm": 2.1950674057006836, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7256832718849182, + "num_tokens": 157721088.0, + "step": 6176 + }, + { + "epoch": 0.678343949044586, + "grad_norm": 1.904357671737671, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.705384373664856, + "num_tokens": 157752414.0, + "step": 6177 + }, + { + "epoch": 0.6784537667471996, + "grad_norm": 2.1552724838256836, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7134716510772705, + "num_tokens": 157776593.0, + "step": 6178 + }, + { + "epoch": 0.6785635844498134, + "grad_norm": 2.2069602012634277, + "learning_rate": 1e-06, + "loss": 1.0507, + "mean_token_accuracy": 0.683495819568634, + "num_tokens": 157804021.0, + "step": 6179 + }, + { + "epoch": 0.678673402152427, + "grad_norm": 2.483391523361206, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7307934761047363, + "num_tokens": 157824305.0, + "step": 6180 + }, + { + "epoch": 0.6787832198550406, + "grad_norm": 1.9207676649093628, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7108919620513916, + "num_tokens": 157855037.0, + "step": 6181 + }, + { + "epoch": 0.6788930375576543, + "grad_norm": 2.246454954147339, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7078866958618164, + "num_tokens": 157878570.0, + "step": 6182 + }, + { + "epoch": 0.6790028552602679, + "grad_norm": 2.1759095191955566, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7076231241226196, + "num_tokens": 157904393.0, + "step": 6183 + }, + { + "epoch": 0.6791126729628816, + "grad_norm": 2.3691797256469727, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7278558611869812, + "num_tokens": 157925960.0, + "step": 6184 + }, + { + "epoch": 0.6792224906654952, + "grad_norm": 2.0299072265625, + "learning_rate": 1e-06, + "loss": 1.0463, + "mean_token_accuracy": 0.6931804418563843, + "num_tokens": 157956511.0, + "step": 6185 + }, + { + "epoch": 0.679332308368109, + "grad_norm": 2.152865171432495, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.6847381591796875, + "num_tokens": 157985302.0, + "step": 6186 + }, + { + "epoch": 0.6794421260707226, + "grad_norm": 2.1824026107788086, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7162489891052246, + "num_tokens": 158008761.0, + "step": 6187 + }, + { + "epoch": 0.6795519437733363, + "grad_norm": 2.220219612121582, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.7528481483459473, + "num_tokens": 158030627.0, + "step": 6188 + }, + { + "epoch": 0.6796617614759499, + "grad_norm": 2.1286416053771973, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7165423631668091, + "num_tokens": 158057379.0, + "step": 6189 + }, + { + "epoch": 0.6797715791785636, + "grad_norm": 2.3466367721557617, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7200442552566528, + "num_tokens": 158080391.0, + "step": 6190 + }, + { + "epoch": 0.6798813968811772, + "grad_norm": 2.4425714015960693, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7228282690048218, + "num_tokens": 158101773.0, + "step": 6191 + }, + { + "epoch": 0.6799912145837909, + "grad_norm": 2.3623902797698975, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7087885141372681, + "num_tokens": 158123980.0, + "step": 6192 + }, + { + "epoch": 0.6801010322864046, + "grad_norm": 2.263943672180176, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7168522477149963, + "num_tokens": 158147107.0, + "step": 6193 + }, + { + "epoch": 0.6802108499890183, + "grad_norm": 2.155280590057373, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.6954400539398193, + "num_tokens": 158173103.0, + "step": 6194 + }, + { + "epoch": 0.6803206676916319, + "grad_norm": 2.0729727745056152, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.6977558135986328, + "num_tokens": 158201181.0, + "step": 6195 + }, + { + "epoch": 0.6804304853942456, + "grad_norm": 2.0932374000549316, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.6980381011962891, + "num_tokens": 158230246.0, + "step": 6196 + }, + { + "epoch": 0.6805403030968592, + "grad_norm": 2.151632070541382, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.697242259979248, + "num_tokens": 158255486.0, + "step": 6197 + }, + { + "epoch": 0.6806501207994728, + "grad_norm": 2.1218767166137695, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7097504734992981, + "num_tokens": 158281698.0, + "step": 6198 + }, + { + "epoch": 0.6807599385020865, + "grad_norm": 2.6438441276550293, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6915399432182312, + "num_tokens": 158302764.0, + "step": 6199 + }, + { + "epoch": 0.6808697562047002, + "grad_norm": 2.258073329925537, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6971728801727295, + "num_tokens": 158327514.0, + "step": 6200 + }, + { + "epoch": 0.6809795739073139, + "grad_norm": 1.9546735286712646, + "learning_rate": 1e-06, + "loss": 1.0256, + "mean_token_accuracy": 0.6893452405929565, + "num_tokens": 158358474.0, + "step": 6201 + }, + { + "epoch": 0.6810893916099275, + "grad_norm": 2.34787917137146, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7296155691146851, + "num_tokens": 158380581.0, + "step": 6202 + }, + { + "epoch": 0.6811992093125412, + "grad_norm": 2.1187236309051514, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.6900397539138794, + "num_tokens": 158409167.0, + "step": 6203 + }, + { + "epoch": 0.6813090270151548, + "grad_norm": 2.1955411434173584, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7107966542243958, + "num_tokens": 158433642.0, + "step": 6204 + }, + { + "epoch": 0.6814188447177685, + "grad_norm": 2.171638011932373, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7029359340667725, + "num_tokens": 158457675.0, + "step": 6205 + }, + { + "epoch": 0.6815286624203821, + "grad_norm": 1.9657310247421265, + "learning_rate": 1e-06, + "loss": 1.0576, + "mean_token_accuracy": 0.6827880144119263, + "num_tokens": 158487285.0, + "step": 6206 + }, + { + "epoch": 0.6816384801229959, + "grad_norm": 2.182095766067505, + "learning_rate": 1e-06, + "loss": 1.0388, + "mean_token_accuracy": 0.6850160956382751, + "num_tokens": 158513135.0, + "step": 6207 + }, + { + "epoch": 0.6817482978256095, + "grad_norm": 2.1592915058135986, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7116313576698303, + "num_tokens": 158540248.0, + "step": 6208 + }, + { + "epoch": 0.6818581155282232, + "grad_norm": 2.1871562004089355, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7378398180007935, + "num_tokens": 158564598.0, + "step": 6209 + }, + { + "epoch": 0.6819679332308368, + "grad_norm": 2.110182523727417, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6944357752799988, + "num_tokens": 158591151.0, + "step": 6210 + }, + { + "epoch": 0.6820777509334505, + "grad_norm": 2.172346830368042, + "learning_rate": 1e-06, + "loss": 1.0462, + "mean_token_accuracy": 0.6835298538208008, + "num_tokens": 158617427.0, + "step": 6211 + }, + { + "epoch": 0.6821875686360641, + "grad_norm": 2.2727468013763428, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6886447668075562, + "num_tokens": 158642955.0, + "step": 6212 + }, + { + "epoch": 0.6822973863386778, + "grad_norm": 2.2652053833007812, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.690528929233551, + "num_tokens": 158668482.0, + "step": 6213 + }, + { + "epoch": 0.6824072040412914, + "grad_norm": 2.317228078842163, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7074785232543945, + "num_tokens": 158691853.0, + "step": 6214 + }, + { + "epoch": 0.6825170217439052, + "grad_norm": 2.404158115386963, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7313977479934692, + "num_tokens": 158712358.0, + "step": 6215 + }, + { + "epoch": 0.6826268394465188, + "grad_norm": 2.4504854679107666, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7275794744491577, + "num_tokens": 158732244.0, + "step": 6216 + }, + { + "epoch": 0.6827366571491325, + "grad_norm": 2.527627944946289, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7154128551483154, + "num_tokens": 158753505.0, + "step": 6217 + }, + { + "epoch": 0.6828464748517461, + "grad_norm": 2.02323579788208, + "learning_rate": 1e-06, + "loss": 1.0691, + "mean_token_accuracy": 0.6815009713172913, + "num_tokens": 158784843.0, + "step": 6218 + }, + { + "epoch": 0.6829562925543597, + "grad_norm": 2.2978434562683105, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7124649882316589, + "num_tokens": 158808188.0, + "step": 6219 + }, + { + "epoch": 0.6830661102569734, + "grad_norm": 2.3555524349212646, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7030774354934692, + "num_tokens": 158830744.0, + "step": 6220 + }, + { + "epoch": 0.683175927959587, + "grad_norm": 1.9156588315963745, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.695911705493927, + "num_tokens": 158862283.0, + "step": 6221 + }, + { + "epoch": 0.6832857456622008, + "grad_norm": 2.1436455249786377, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7178415060043335, + "num_tokens": 158889982.0, + "step": 6222 + }, + { + "epoch": 0.6833955633648144, + "grad_norm": 2.4003286361694336, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7194616794586182, + "num_tokens": 158911697.0, + "step": 6223 + }, + { + "epoch": 0.6835053810674281, + "grad_norm": 2.291411876678467, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.7075328230857849, + "num_tokens": 158937459.0, + "step": 6224 + }, + { + "epoch": 0.6836151987700417, + "grad_norm": 2.3675692081451416, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7117664813995361, + "num_tokens": 158959083.0, + "step": 6225 + }, + { + "epoch": 0.6837250164726554, + "grad_norm": 2.186971664428711, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7031266689300537, + "num_tokens": 158984450.0, + "step": 6226 + }, + { + "epoch": 0.683834834175269, + "grad_norm": 2.1537740230560303, + "learning_rate": 1e-06, + "loss": 1.0325, + "mean_token_accuracy": 0.684526801109314, + "num_tokens": 159013050.0, + "step": 6227 + }, + { + "epoch": 0.6839446518778827, + "grad_norm": 2.240109443664551, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7204551100730896, + "num_tokens": 159036167.0, + "step": 6228 + }, + { + "epoch": 0.6840544695804964, + "grad_norm": 2.137204647064209, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7118807435035706, + "num_tokens": 159063002.0, + "step": 6229 + }, + { + "epoch": 0.6841642872831101, + "grad_norm": 1.996463418006897, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7185692191123962, + "num_tokens": 159091503.0, + "step": 6230 + }, + { + "epoch": 0.6842741049857237, + "grad_norm": 2.1360085010528564, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7344896197319031, + "num_tokens": 159118873.0, + "step": 6231 + }, + { + "epoch": 0.6843839226883374, + "grad_norm": 1.9655793905258179, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6914528608322144, + "num_tokens": 159151430.0, + "step": 6232 + }, + { + "epoch": 0.684493740390951, + "grad_norm": 2.536114454269409, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7306500673294067, + "num_tokens": 159170759.0, + "step": 6233 + }, + { + "epoch": 0.6846035580935647, + "grad_norm": 2.2450239658355713, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7172874808311462, + "num_tokens": 159194015.0, + "step": 6234 + }, + { + "epoch": 0.6847133757961783, + "grad_norm": 2.431337594985962, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7024955749511719, + "num_tokens": 159216490.0, + "step": 6235 + }, + { + "epoch": 0.6848231934987921, + "grad_norm": 2.0830540657043457, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.714780330657959, + "num_tokens": 159242578.0, + "step": 6236 + }, + { + "epoch": 0.6849330112014057, + "grad_norm": 2.323903799057007, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7260407209396362, + "num_tokens": 159264583.0, + "step": 6237 + }, + { + "epoch": 0.6850428289040194, + "grad_norm": 2.215820789337158, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7265181541442871, + "num_tokens": 159288821.0, + "step": 6238 + }, + { + "epoch": 0.685152646606633, + "grad_norm": 2.4465763568878174, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7096184492111206, + "num_tokens": 159310873.0, + "step": 6239 + }, + { + "epoch": 0.6852624643092466, + "grad_norm": 2.258875846862793, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7097473740577698, + "num_tokens": 159335578.0, + "step": 6240 + }, + { + "epoch": 0.6853722820118603, + "grad_norm": 2.0518627166748047, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7119290828704834, + "num_tokens": 159363178.0, + "step": 6241 + }, + { + "epoch": 0.6854820997144739, + "grad_norm": 2.1242220401763916, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7083305716514587, + "num_tokens": 159388922.0, + "step": 6242 + }, + { + "epoch": 0.6855919174170876, + "grad_norm": 1.9568912982940674, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.6939507126808167, + "num_tokens": 159420319.0, + "step": 6243 + }, + { + "epoch": 0.6857017351197013, + "grad_norm": 2.2012457847595215, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7271296381950378, + "num_tokens": 159444324.0, + "step": 6244 + }, + { + "epoch": 0.685811552822315, + "grad_norm": 2.1593093872070312, + "learning_rate": 1e-06, + "loss": 1.0619, + "mean_token_accuracy": 0.6785043478012085, + "num_tokens": 159471086.0, + "step": 6245 + }, + { + "epoch": 0.6859213705249286, + "grad_norm": 2.0544192790985107, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7075493335723877, + "num_tokens": 159501125.0, + "step": 6246 + }, + { + "epoch": 0.6860311882275423, + "grad_norm": 2.371699571609497, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7205874919891357, + "num_tokens": 159522973.0, + "step": 6247 + }, + { + "epoch": 0.6861410059301559, + "grad_norm": 2.1265034675598145, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7324409484863281, + "num_tokens": 159547348.0, + "step": 6248 + }, + { + "epoch": 0.6862508236327696, + "grad_norm": 2.0657715797424316, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7038202881813049, + "num_tokens": 159574670.0, + "step": 6249 + }, + { + "epoch": 0.6863606413353832, + "grad_norm": 2.1689453125, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7123364806175232, + "num_tokens": 159599050.0, + "step": 6250 + }, + { + "epoch": 0.686470459037997, + "grad_norm": 1.968375563621521, + "learning_rate": 1e-06, + "loss": 1.059, + "mean_token_accuracy": 0.6793856024742126, + "num_tokens": 159633125.0, + "step": 6251 + }, + { + "epoch": 0.6865802767406106, + "grad_norm": 2.0396082401275635, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6958298683166504, + "num_tokens": 159662430.0, + "step": 6252 + }, + { + "epoch": 0.6866900944432243, + "grad_norm": 2.226262092590332, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.702890157699585, + "num_tokens": 159686798.0, + "step": 6253 + }, + { + "epoch": 0.6867999121458379, + "grad_norm": 2.214250087738037, + "learning_rate": 1e-06, + "loss": 1.0549, + "mean_token_accuracy": 0.6791458129882812, + "num_tokens": 159713757.0, + "step": 6254 + }, + { + "epoch": 0.6869097298484516, + "grad_norm": 2.20572829246521, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.6983447074890137, + "num_tokens": 159740597.0, + "step": 6255 + }, + { + "epoch": 0.6870195475510652, + "grad_norm": 2.2862279415130615, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7225722074508667, + "num_tokens": 159763055.0, + "step": 6256 + }, + { + "epoch": 0.6871293652536788, + "grad_norm": 2.1048684120178223, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.6925852298736572, + "num_tokens": 159791403.0, + "step": 6257 + }, + { + "epoch": 0.6872391829562926, + "grad_norm": 2.0035769939422607, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6845847368240356, + "num_tokens": 159821284.0, + "step": 6258 + }, + { + "epoch": 0.6873490006589063, + "grad_norm": 1.9780700206756592, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.6958751082420349, + "num_tokens": 159852009.0, + "step": 6259 + }, + { + "epoch": 0.6874588183615199, + "grad_norm": 2.0940351486206055, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7305803298950195, + "num_tokens": 159877976.0, + "step": 6260 + }, + { + "epoch": 0.6875686360641335, + "grad_norm": 2.086521625518799, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.692708432674408, + "num_tokens": 159906461.0, + "step": 6261 + }, + { + "epoch": 0.6876784537667472, + "grad_norm": 2.3636457920074463, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7127515077590942, + "num_tokens": 159929158.0, + "step": 6262 + }, + { + "epoch": 0.6877882714693608, + "grad_norm": 2.1134071350097656, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7148877382278442, + "num_tokens": 159955284.0, + "step": 6263 + }, + { + "epoch": 0.6878980891719745, + "grad_norm": 2.8164191246032715, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7381558418273926, + "num_tokens": 159972206.0, + "step": 6264 + }, + { + "epoch": 0.6880079068745882, + "grad_norm": 2.4117214679718018, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7093298435211182, + "num_tokens": 159993016.0, + "step": 6265 + }, + { + "epoch": 0.6881177245772019, + "grad_norm": 1.9379466772079468, + "learning_rate": 1e-06, + "loss": 1.0494, + "mean_token_accuracy": 0.6916463375091553, + "num_tokens": 160024792.0, + "step": 6266 + }, + { + "epoch": 0.6882275422798155, + "grad_norm": 2.1745645999908447, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7234973311424255, + "num_tokens": 160049527.0, + "step": 6267 + }, + { + "epoch": 0.6883373599824292, + "grad_norm": 2.2263407707214355, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.708818793296814, + "num_tokens": 160075250.0, + "step": 6268 + }, + { + "epoch": 0.6884471776850428, + "grad_norm": 2.236302614212036, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.7017184495925903, + "num_tokens": 160100072.0, + "step": 6269 + }, + { + "epoch": 0.6885569953876565, + "grad_norm": 2.2582476139068604, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.6872249245643616, + "num_tokens": 160124284.0, + "step": 6270 + }, + { + "epoch": 0.6886668130902701, + "grad_norm": 2.355543851852417, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7158929705619812, + "num_tokens": 160146604.0, + "step": 6271 + }, + { + "epoch": 0.6887766307928838, + "grad_norm": 2.309394359588623, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7230954766273499, + "num_tokens": 160169343.0, + "step": 6272 + }, + { + "epoch": 0.6888864484954975, + "grad_norm": 1.8990998268127441, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6963977813720703, + "num_tokens": 160200720.0, + "step": 6273 + }, + { + "epoch": 0.6889962661981112, + "grad_norm": 2.219021797180176, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.7014355659484863, + "num_tokens": 160225206.0, + "step": 6274 + }, + { + "epoch": 0.6891060839007248, + "grad_norm": 2.2268424034118652, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7291911840438843, + "num_tokens": 160250670.0, + "step": 6275 + }, + { + "epoch": 0.6892159016033385, + "grad_norm": 1.9351484775543213, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.718792200088501, + "num_tokens": 160280001.0, + "step": 6276 + }, + { + "epoch": 0.6893257193059521, + "grad_norm": 2.0607359409332275, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6935042142868042, + "num_tokens": 160308405.0, + "step": 6277 + }, + { + "epoch": 0.6894355370085657, + "grad_norm": 2.0048792362213135, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7058900594711304, + "num_tokens": 160338375.0, + "step": 6278 + }, + { + "epoch": 0.6895453547111794, + "grad_norm": 2.041348457336426, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.6998458504676819, + "num_tokens": 160369104.0, + "step": 6279 + }, + { + "epoch": 0.6896551724137931, + "grad_norm": 2.3003790378570557, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7156379818916321, + "num_tokens": 160391156.0, + "step": 6280 + }, + { + "epoch": 0.6897649901164068, + "grad_norm": 2.456740379333496, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7034755945205688, + "num_tokens": 160413373.0, + "step": 6281 + }, + { + "epoch": 0.6898748078190204, + "grad_norm": 2.123586893081665, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7059451937675476, + "num_tokens": 160438546.0, + "step": 6282 + }, + { + "epoch": 0.6899846255216341, + "grad_norm": 1.7911972999572754, + "learning_rate": 1e-06, + "loss": 1.0559, + "mean_token_accuracy": 0.6827784776687622, + "num_tokens": 160477786.0, + "step": 6283 + }, + { + "epoch": 0.6900944432242477, + "grad_norm": 2.4449241161346436, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7296180129051208, + "num_tokens": 160498731.0, + "step": 6284 + }, + { + "epoch": 0.6902042609268614, + "grad_norm": 2.4713339805603027, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7237600088119507, + "num_tokens": 160519019.0, + "step": 6285 + }, + { + "epoch": 0.690314078629475, + "grad_norm": 2.176816940307617, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.709492027759552, + "num_tokens": 160542954.0, + "step": 6286 + }, + { + "epoch": 0.6904238963320888, + "grad_norm": 2.1093924045562744, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7239699363708496, + "num_tokens": 160570637.0, + "step": 6287 + }, + { + "epoch": 0.6905337140347024, + "grad_norm": 2.2590699195861816, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.709600567817688, + "num_tokens": 160592513.0, + "step": 6288 + }, + { + "epoch": 0.6906435317373161, + "grad_norm": 2.3113551139831543, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.712537407875061, + "num_tokens": 160617113.0, + "step": 6289 + }, + { + "epoch": 0.6907533494399297, + "grad_norm": 1.8310205936431885, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7125555872917175, + "num_tokens": 160649439.0, + "step": 6290 + }, + { + "epoch": 0.6908631671425434, + "grad_norm": 2.1199162006378174, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7050905227661133, + "num_tokens": 160678040.0, + "step": 6291 + }, + { + "epoch": 0.690972984845157, + "grad_norm": 1.9826984405517578, + "learning_rate": 1e-06, + "loss": 1.0286, + "mean_token_accuracy": 0.6868367195129395, + "num_tokens": 160709532.0, + "step": 6292 + }, + { + "epoch": 0.6910828025477707, + "grad_norm": 2.2521631717681885, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7292897701263428, + "num_tokens": 160732802.0, + "step": 6293 + }, + { + "epoch": 0.6911926202503844, + "grad_norm": 2.245008945465088, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7088438272476196, + "num_tokens": 160758961.0, + "step": 6294 + }, + { + "epoch": 0.6913024379529981, + "grad_norm": 2.3631980419158936, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.719077467918396, + "num_tokens": 160779860.0, + "step": 6295 + }, + { + "epoch": 0.6914122556556117, + "grad_norm": 2.238960027694702, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7102588415145874, + "num_tokens": 160805715.0, + "step": 6296 + }, + { + "epoch": 0.6915220733582254, + "grad_norm": 2.2831318378448486, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7094452977180481, + "num_tokens": 160829028.0, + "step": 6297 + }, + { + "epoch": 0.691631891060839, + "grad_norm": 2.2566845417022705, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.7070229053497314, + "num_tokens": 160853523.0, + "step": 6298 + }, + { + "epoch": 0.6917417087634526, + "grad_norm": 2.0982937812805176, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7381402850151062, + "num_tokens": 160879869.0, + "step": 6299 + }, + { + "epoch": 0.6918515264660663, + "grad_norm": 2.2091610431671143, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.6986038684844971, + "num_tokens": 160905699.0, + "step": 6300 + }, + { + "epoch": 0.6919613441686799, + "grad_norm": 1.7293174266815186, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7143592238426208, + "num_tokens": 160944475.0, + "step": 6301 + }, + { + "epoch": 0.6920711618712937, + "grad_norm": 2.5139095783233643, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.732461154460907, + "num_tokens": 160965001.0, + "step": 6302 + }, + { + "epoch": 0.6921809795739073, + "grad_norm": 2.205378532409668, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7177870273590088, + "num_tokens": 160989612.0, + "step": 6303 + }, + { + "epoch": 0.692290797276521, + "grad_norm": 1.8058912754058838, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.6928426623344421, + "num_tokens": 161023657.0, + "step": 6304 + }, + { + "epoch": 0.6924006149791346, + "grad_norm": 1.9847733974456787, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7233261466026306, + "num_tokens": 161052145.0, + "step": 6305 + }, + { + "epoch": 0.6925104326817483, + "grad_norm": 2.1805686950683594, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7213925123214722, + "num_tokens": 161077931.0, + "step": 6306 + }, + { + "epoch": 0.6926202503843619, + "grad_norm": 2.113083600997925, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7119369506835938, + "num_tokens": 161107346.0, + "step": 6307 + }, + { + "epoch": 0.6927300680869756, + "grad_norm": 2.394951581954956, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7090355157852173, + "num_tokens": 161129168.0, + "step": 6308 + }, + { + "epoch": 0.6928398857895893, + "grad_norm": 2.35213565826416, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.7018811702728271, + "num_tokens": 161157684.0, + "step": 6309 + }, + { + "epoch": 0.692949703492203, + "grad_norm": 2.453094482421875, + "learning_rate": 1e-06, + "loss": 1.0442, + "mean_token_accuracy": 0.6829589009284973, + "num_tokens": 161181553.0, + "step": 6310 + }, + { + "epoch": 0.6930595211948166, + "grad_norm": 2.5373330116271973, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7235827445983887, + "num_tokens": 161200411.0, + "step": 6311 + }, + { + "epoch": 0.6931693388974303, + "grad_norm": 2.1623475551605225, + "learning_rate": 1e-06, + "loss": 1.0969, + "mean_token_accuracy": 0.6827707290649414, + "num_tokens": 161227964.0, + "step": 6312 + }, + { + "epoch": 0.6932791566000439, + "grad_norm": 2.2496023178100586, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7058776617050171, + "num_tokens": 161251674.0, + "step": 6313 + }, + { + "epoch": 0.6933889743026576, + "grad_norm": 2.2501485347747803, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7109763622283936, + "num_tokens": 161276829.0, + "step": 6314 + }, + { + "epoch": 0.6934987920052712, + "grad_norm": 2.1905269622802734, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.6971436738967896, + "num_tokens": 161301924.0, + "step": 6315 + }, + { + "epoch": 0.693608609707885, + "grad_norm": 2.1972033977508545, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.6995803117752075, + "num_tokens": 161328277.0, + "step": 6316 + }, + { + "epoch": 0.6937184274104986, + "grad_norm": 2.2511160373687744, + "learning_rate": 1e-06, + "loss": 1.0327, + "mean_token_accuracy": 0.6865798830986023, + "num_tokens": 161352581.0, + "step": 6317 + }, + { + "epoch": 0.6938282451131123, + "grad_norm": 2.118009090423584, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.7010616064071655, + "num_tokens": 161378980.0, + "step": 6318 + }, + { + "epoch": 0.6939380628157259, + "grad_norm": 2.007037878036499, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7044745087623596, + "num_tokens": 161410535.0, + "step": 6319 + }, + { + "epoch": 0.6940478805183395, + "grad_norm": 2.357048273086548, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.6996300220489502, + "num_tokens": 161433852.0, + "step": 6320 + }, + { + "epoch": 0.6941576982209532, + "grad_norm": 2.2518646717071533, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7090324759483337, + "num_tokens": 161458590.0, + "step": 6321 + }, + { + "epoch": 0.6942675159235668, + "grad_norm": 2.1997926235198975, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.706118106842041, + "num_tokens": 161484187.0, + "step": 6322 + }, + { + "epoch": 0.6943773336261806, + "grad_norm": 2.1514692306518555, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.732526421546936, + "num_tokens": 161507377.0, + "step": 6323 + }, + { + "epoch": 0.6944871513287942, + "grad_norm": 2.4439785480499268, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.714873194694519, + "num_tokens": 161529554.0, + "step": 6324 + }, + { + "epoch": 0.6945969690314079, + "grad_norm": 2.089261770248413, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7410666942596436, + "num_tokens": 161555748.0, + "step": 6325 + }, + { + "epoch": 0.6947067867340215, + "grad_norm": 2.342592477798462, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7098715901374817, + "num_tokens": 161579552.0, + "step": 6326 + }, + { + "epoch": 0.6948166044366352, + "grad_norm": 2.1972970962524414, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.7005494236946106, + "num_tokens": 161605269.0, + "step": 6327 + }, + { + "epoch": 0.6949264221392488, + "grad_norm": 2.2870495319366455, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7123150825500488, + "num_tokens": 161629581.0, + "step": 6328 + }, + { + "epoch": 0.6950362398418625, + "grad_norm": 2.1231863498687744, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6947693824768066, + "num_tokens": 161656604.0, + "step": 6329 + }, + { + "epoch": 0.6951460575444761, + "grad_norm": 2.0131568908691406, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.6963650584220886, + "num_tokens": 161685245.0, + "step": 6330 + }, + { + "epoch": 0.6952558752470899, + "grad_norm": 2.041102170944214, + "learning_rate": 1e-06, + "loss": 1.1322, + "mean_token_accuracy": 0.6659142971038818, + "num_tokens": 161716615.0, + "step": 6331 + }, + { + "epoch": 0.6953656929497035, + "grad_norm": 2.2057907581329346, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7058019042015076, + "num_tokens": 161741855.0, + "step": 6332 + }, + { + "epoch": 0.6954755106523172, + "grad_norm": 2.090205430984497, + "learning_rate": 1e-06, + "loss": 1.0377, + "mean_token_accuracy": 0.6882715225219727, + "num_tokens": 161770690.0, + "step": 6333 + }, + { + "epoch": 0.6955853283549308, + "grad_norm": 2.1165144443511963, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7142671346664429, + "num_tokens": 161798315.0, + "step": 6334 + }, + { + "epoch": 0.6956951460575445, + "grad_norm": 2.2166807651519775, + "learning_rate": 1e-06, + "loss": 1.0382, + "mean_token_accuracy": 0.6898202896118164, + "num_tokens": 161825261.0, + "step": 6335 + }, + { + "epoch": 0.6958049637601581, + "grad_norm": 2.280651092529297, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7168702483177185, + "num_tokens": 161847872.0, + "step": 6336 + }, + { + "epoch": 0.6959147814627717, + "grad_norm": 2.1541144847869873, + "learning_rate": 1e-06, + "loss": 1.063, + "mean_token_accuracy": 0.6846659779548645, + "num_tokens": 161875845.0, + "step": 6337 + }, + { + "epoch": 0.6960245991653855, + "grad_norm": 2.3464195728302, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7254653573036194, + "num_tokens": 161897928.0, + "step": 6338 + }, + { + "epoch": 0.6961344168679992, + "grad_norm": 2.2636241912841797, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7298429012298584, + "num_tokens": 161919175.0, + "step": 6339 + }, + { + "epoch": 0.6962442345706128, + "grad_norm": 2.1209676265716553, + "learning_rate": 1e-06, + "loss": 1.071, + "mean_token_accuracy": 0.6765591502189636, + "num_tokens": 161948359.0, + "step": 6340 + }, + { + "epoch": 0.6963540522732264, + "grad_norm": 1.9999775886535645, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7209181785583496, + "num_tokens": 161977654.0, + "step": 6341 + }, + { + "epoch": 0.6964638699758401, + "grad_norm": 2.1903884410858154, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.693569540977478, + "num_tokens": 162002220.0, + "step": 6342 + }, + { + "epoch": 0.6965736876784537, + "grad_norm": 2.1324028968811035, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7081407308578491, + "num_tokens": 162028234.0, + "step": 6343 + }, + { + "epoch": 0.6966835053810674, + "grad_norm": 2.5646705627441406, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.716971755027771, + "num_tokens": 162047997.0, + "step": 6344 + }, + { + "epoch": 0.6967933230836811, + "grad_norm": 2.1132664680480957, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6976434588432312, + "num_tokens": 162074720.0, + "step": 6345 + }, + { + "epoch": 0.6969031407862948, + "grad_norm": 2.4229421615600586, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7264189720153809, + "num_tokens": 162096169.0, + "step": 6346 + }, + { + "epoch": 0.6970129584889084, + "grad_norm": 2.2414748668670654, + "learning_rate": 1e-06, + "loss": 1.0242, + "mean_token_accuracy": 0.6857913732528687, + "num_tokens": 162121877.0, + "step": 6347 + }, + { + "epoch": 0.6971227761915221, + "grad_norm": 2.245086431503296, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.7066947221755981, + "num_tokens": 162147552.0, + "step": 6348 + }, + { + "epoch": 0.6972325938941357, + "grad_norm": 2.1505444049835205, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7274195551872253, + "num_tokens": 162172842.0, + "step": 6349 + }, + { + "epoch": 0.6973424115967494, + "grad_norm": 2.4058709144592285, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.6976983547210693, + "num_tokens": 162194684.0, + "step": 6350 + }, + { + "epoch": 0.697452229299363, + "grad_norm": 2.262280225753784, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.7017403841018677, + "num_tokens": 162218483.0, + "step": 6351 + }, + { + "epoch": 0.6975620470019768, + "grad_norm": 2.0997488498687744, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6925431489944458, + "num_tokens": 162246738.0, + "step": 6352 + }, + { + "epoch": 0.6976718647045904, + "grad_norm": 2.161024570465088, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7277392745018005, + "num_tokens": 162272160.0, + "step": 6353 + }, + { + "epoch": 0.6977816824072041, + "grad_norm": 2.461174488067627, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7091563940048218, + "num_tokens": 162292047.0, + "step": 6354 + }, + { + "epoch": 0.6978915001098177, + "grad_norm": 2.226862907409668, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.7032247185707092, + "num_tokens": 162319352.0, + "step": 6355 + }, + { + "epoch": 0.6980013178124314, + "grad_norm": 2.3143723011016846, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7186523675918579, + "num_tokens": 162342127.0, + "step": 6356 + }, + { + "epoch": 0.698111135515045, + "grad_norm": 2.2147021293640137, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7174625396728516, + "num_tokens": 162367077.0, + "step": 6357 + }, + { + "epoch": 0.6982209532176586, + "grad_norm": 2.1544547080993652, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.7026911973953247, + "num_tokens": 162395078.0, + "step": 6358 + }, + { + "epoch": 0.6983307709202724, + "grad_norm": 2.2497451305389404, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.6980552077293396, + "num_tokens": 162420828.0, + "step": 6359 + }, + { + "epoch": 0.698440588622886, + "grad_norm": 2.2342374324798584, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7036316394805908, + "num_tokens": 162445719.0, + "step": 6360 + }, + { + "epoch": 0.6985504063254997, + "grad_norm": 2.375174045562744, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7205930948257446, + "num_tokens": 162468413.0, + "step": 6361 + }, + { + "epoch": 0.6986602240281133, + "grad_norm": 2.1074883937835693, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7089076042175293, + "num_tokens": 162497562.0, + "step": 6362 + }, + { + "epoch": 0.698770041730727, + "grad_norm": 2.2797796726226807, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7208689451217651, + "num_tokens": 162522084.0, + "step": 6363 + }, + { + "epoch": 0.6988798594333406, + "grad_norm": 2.1438488960266113, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6945013999938965, + "num_tokens": 162549475.0, + "step": 6364 + }, + { + "epoch": 0.6989896771359543, + "grad_norm": 2.159820079803467, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7100033760070801, + "num_tokens": 162573963.0, + "step": 6365 + }, + { + "epoch": 0.6990994948385679, + "grad_norm": 2.4769201278686523, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7168135046958923, + "num_tokens": 162595162.0, + "step": 6366 + }, + { + "epoch": 0.6992093125411817, + "grad_norm": 2.3300647735595703, + "learning_rate": 1e-06, + "loss": 1.0351, + "mean_token_accuracy": 0.699815034866333, + "num_tokens": 162618814.0, + "step": 6367 + }, + { + "epoch": 0.6993191302437953, + "grad_norm": 2.09967303276062, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7169378995895386, + "num_tokens": 162645195.0, + "step": 6368 + }, + { + "epoch": 0.699428947946409, + "grad_norm": 2.20938777923584, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7145267724990845, + "num_tokens": 162672297.0, + "step": 6369 + }, + { + "epoch": 0.6995387656490226, + "grad_norm": 2.142361640930176, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7101010084152222, + "num_tokens": 162700268.0, + "step": 6370 + }, + { + "epoch": 0.6996485833516363, + "grad_norm": 1.9606513977050781, + "learning_rate": 1e-06, + "loss": 1.0407, + "mean_token_accuracy": 0.682086169719696, + "num_tokens": 162732285.0, + "step": 6371 + }, + { + "epoch": 0.6997584010542499, + "grad_norm": 2.259345531463623, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7211834192276001, + "num_tokens": 162756123.0, + "step": 6372 + }, + { + "epoch": 0.6998682187568636, + "grad_norm": 2.65970516204834, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7280965447425842, + "num_tokens": 162774524.0, + "step": 6373 + }, + { + "epoch": 0.6999780364594773, + "grad_norm": 2.1408469676971436, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7124357223510742, + "num_tokens": 162799212.0, + "step": 6374 + }, + { + "epoch": 0.700087854162091, + "grad_norm": 2.0600783824920654, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.7009001970291138, + "num_tokens": 162825936.0, + "step": 6375 + }, + { + "epoch": 0.7001976718647046, + "grad_norm": 2.328251838684082, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7071413993835449, + "num_tokens": 162847559.0, + "step": 6376 + }, + { + "epoch": 0.7003074895673183, + "grad_norm": 2.7407045364379883, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7117499113082886, + "num_tokens": 162865559.0, + "step": 6377 + }, + { + "epoch": 0.7004173072699319, + "grad_norm": 2.235212802886963, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.712567925453186, + "num_tokens": 162888994.0, + "step": 6378 + }, + { + "epoch": 0.7005271249725455, + "grad_norm": 2.0028340816497803, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7068995237350464, + "num_tokens": 162921957.0, + "step": 6379 + }, + { + "epoch": 0.7006369426751592, + "grad_norm": 2.1713192462921143, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.6896969079971313, + "num_tokens": 162948914.0, + "step": 6380 + }, + { + "epoch": 0.700746760377773, + "grad_norm": 1.9947720766067505, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7397967576980591, + "num_tokens": 162979924.0, + "step": 6381 + }, + { + "epoch": 0.7008565780803866, + "grad_norm": 2.393617630004883, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7068547606468201, + "num_tokens": 163001984.0, + "step": 6382 + }, + { + "epoch": 0.7009663957830002, + "grad_norm": 1.9868955612182617, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.693157434463501, + "num_tokens": 163031559.0, + "step": 6383 + }, + { + "epoch": 0.7010762134856139, + "grad_norm": 2.113757848739624, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.7028769254684448, + "num_tokens": 163059387.0, + "step": 6384 + }, + { + "epoch": 0.7011860311882275, + "grad_norm": 2.2693686485290527, + "learning_rate": 1e-06, + "loss": 1.0225, + "mean_token_accuracy": 0.6926047205924988, + "num_tokens": 163087289.0, + "step": 6385 + }, + { + "epoch": 0.7012958488908412, + "grad_norm": 2.0082170963287354, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7000464200973511, + "num_tokens": 163115808.0, + "step": 6386 + }, + { + "epoch": 0.7014056665934548, + "grad_norm": 2.1576881408691406, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7135074138641357, + "num_tokens": 163140242.0, + "step": 6387 + }, + { + "epoch": 0.7015154842960686, + "grad_norm": 2.1291615962982178, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7058314681053162, + "num_tokens": 163165804.0, + "step": 6388 + }, + { + "epoch": 0.7016253019986822, + "grad_norm": 2.44486665725708, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7153528928756714, + "num_tokens": 163187697.0, + "step": 6389 + }, + { + "epoch": 0.7017351197012959, + "grad_norm": 2.3247153759002686, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7371031641960144, + "num_tokens": 163210352.0, + "step": 6390 + }, + { + "epoch": 0.7018449374039095, + "grad_norm": 2.01570200920105, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.724208414554596, + "num_tokens": 163239298.0, + "step": 6391 + }, + { + "epoch": 0.7019547551065232, + "grad_norm": 2.0114059448242188, + "learning_rate": 1e-06, + "loss": 1.0029, + "mean_token_accuracy": 0.694726824760437, + "num_tokens": 163270066.0, + "step": 6392 + }, + { + "epoch": 0.7020645728091368, + "grad_norm": 2.17423677444458, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7079952955245972, + "num_tokens": 163296396.0, + "step": 6393 + }, + { + "epoch": 0.7021743905117505, + "grad_norm": 2.1257662773132324, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.6994407176971436, + "num_tokens": 163324362.0, + "step": 6394 + }, + { + "epoch": 0.7022842082143641, + "grad_norm": 2.1014366149902344, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.700458824634552, + "num_tokens": 163351034.0, + "step": 6395 + }, + { + "epoch": 0.7023940259169779, + "grad_norm": 2.175612211227417, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6939904689788818, + "num_tokens": 163377932.0, + "step": 6396 + }, + { + "epoch": 0.7025038436195915, + "grad_norm": 2.0454862117767334, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7079846262931824, + "num_tokens": 163404792.0, + "step": 6397 + }, + { + "epoch": 0.7026136613222052, + "grad_norm": 1.9100455045700073, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6990265846252441, + "num_tokens": 163438808.0, + "step": 6398 + }, + { + "epoch": 0.7027234790248188, + "grad_norm": 2.284960985183716, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7079721689224243, + "num_tokens": 163462337.0, + "step": 6399 + }, + { + "epoch": 0.7028332967274324, + "grad_norm": 1.9892841577529907, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7078119516372681, + "num_tokens": 163490659.0, + "step": 6400 + }, + { + "epoch": 0.7029431144300461, + "grad_norm": 2.333732843399048, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6895077228546143, + "num_tokens": 163513731.0, + "step": 6401 + }, + { + "epoch": 0.7030529321326597, + "grad_norm": 2.06842303276062, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.7081696391105652, + "num_tokens": 163541231.0, + "step": 6402 + }, + { + "epoch": 0.7031627498352735, + "grad_norm": 2.697465658187866, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7266560196876526, + "num_tokens": 163558994.0, + "step": 6403 + }, + { + "epoch": 0.7032725675378871, + "grad_norm": 2.2940216064453125, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.695318341255188, + "num_tokens": 163584711.0, + "step": 6404 + }, + { + "epoch": 0.7033823852405008, + "grad_norm": 2.2324953079223633, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7156537771224976, + "num_tokens": 163609538.0, + "step": 6405 + }, + { + "epoch": 0.7034922029431144, + "grad_norm": 2.4873931407928467, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7128095030784607, + "num_tokens": 163629069.0, + "step": 6406 + }, + { + "epoch": 0.7036020206457281, + "grad_norm": 2.2344789505004883, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7052707672119141, + "num_tokens": 163652881.0, + "step": 6407 + }, + { + "epoch": 0.7037118383483417, + "grad_norm": 2.3468449115753174, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7133498787879944, + "num_tokens": 163674660.0, + "step": 6408 + }, + { + "epoch": 0.7038216560509554, + "grad_norm": 2.1805200576782227, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.6908982396125793, + "num_tokens": 163701506.0, + "step": 6409 + }, + { + "epoch": 0.7039314737535691, + "grad_norm": 2.111457586288452, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7053706049919128, + "num_tokens": 163727818.0, + "step": 6410 + }, + { + "epoch": 0.7040412914561828, + "grad_norm": 2.4455811977386475, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7113417983055115, + "num_tokens": 163749844.0, + "step": 6411 + }, + { + "epoch": 0.7041511091587964, + "grad_norm": 2.174266815185547, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7234265208244324, + "num_tokens": 163775045.0, + "step": 6412 + }, + { + "epoch": 0.7042609268614101, + "grad_norm": 2.268876791000366, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7363615036010742, + "num_tokens": 163797507.0, + "step": 6413 + }, + { + "epoch": 0.7043707445640237, + "grad_norm": 2.1024768352508545, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.718682587146759, + "num_tokens": 163822736.0, + "step": 6414 + }, + { + "epoch": 0.7044805622666374, + "grad_norm": 2.115114212036133, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7197284698486328, + "num_tokens": 163849110.0, + "step": 6415 + }, + { + "epoch": 0.704590379969251, + "grad_norm": 2.2096197605133057, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6992743015289307, + "num_tokens": 163873858.0, + "step": 6416 + }, + { + "epoch": 0.7047001976718648, + "grad_norm": 1.836201548576355, + "learning_rate": 1e-06, + "loss": 1.1133, + "mean_token_accuracy": 0.670554518699646, + "num_tokens": 163911439.0, + "step": 6417 + }, + { + "epoch": 0.7048100153744784, + "grad_norm": 2.288123607635498, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7014394998550415, + "num_tokens": 163935152.0, + "step": 6418 + }, + { + "epoch": 0.704919833077092, + "grad_norm": 2.3691132068634033, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7149587273597717, + "num_tokens": 163956140.0, + "step": 6419 + }, + { + "epoch": 0.7050296507797057, + "grad_norm": 2.478949785232544, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6881081461906433, + "num_tokens": 163979431.0, + "step": 6420 + }, + { + "epoch": 0.7051394684823193, + "grad_norm": 2.2489593029022217, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7001148462295532, + "num_tokens": 164004826.0, + "step": 6421 + }, + { + "epoch": 0.705249286184933, + "grad_norm": 2.0185203552246094, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7186398506164551, + "num_tokens": 164034561.0, + "step": 6422 + }, + { + "epoch": 0.7053591038875466, + "grad_norm": 2.161339521408081, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7132403254508972, + "num_tokens": 164061608.0, + "step": 6423 + }, + { + "epoch": 0.7054689215901603, + "grad_norm": 2.4410042762756348, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7104395627975464, + "num_tokens": 164083521.0, + "step": 6424 + }, + { + "epoch": 0.705578739292774, + "grad_norm": 2.503006935119629, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6980715990066528, + "num_tokens": 164104262.0, + "step": 6425 + }, + { + "epoch": 0.7056885569953877, + "grad_norm": 2.3600502014160156, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7203531861305237, + "num_tokens": 164125431.0, + "step": 6426 + }, + { + "epoch": 0.7057983746980013, + "grad_norm": 2.0729453563690186, + "learning_rate": 1e-06, + "loss": 1.0641, + "mean_token_accuracy": 0.677943229675293, + "num_tokens": 164154989.0, + "step": 6427 + }, + { + "epoch": 0.705908192400615, + "grad_norm": 2.0705413818359375, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.7042847871780396, + "num_tokens": 164184325.0, + "step": 6428 + }, + { + "epoch": 0.7060180101032286, + "grad_norm": 2.188805103302002, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7032017707824707, + "num_tokens": 164207981.0, + "step": 6429 + }, + { + "epoch": 0.7061278278058423, + "grad_norm": 2.208544969558716, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7307319045066833, + "num_tokens": 164231889.0, + "step": 6430 + }, + { + "epoch": 0.7062376455084559, + "grad_norm": 2.0740907192230225, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7263839244842529, + "num_tokens": 164258593.0, + "step": 6431 + }, + { + "epoch": 0.7063474632110697, + "grad_norm": 2.1196606159210205, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.706499457359314, + "num_tokens": 164287157.0, + "step": 6432 + }, + { + "epoch": 0.7064572809136833, + "grad_norm": 2.2075095176696777, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.6994069814682007, + "num_tokens": 164311679.0, + "step": 6433 + }, + { + "epoch": 0.706567098616297, + "grad_norm": 2.066943407058716, + "learning_rate": 1e-06, + "loss": 1.0371, + "mean_token_accuracy": 0.7094550132751465, + "num_tokens": 164339491.0, + "step": 6434 + }, + { + "epoch": 0.7066769163189106, + "grad_norm": 2.291175603866577, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.705534815788269, + "num_tokens": 164362068.0, + "step": 6435 + }, + { + "epoch": 0.7067867340215243, + "grad_norm": 1.9985997676849365, + "learning_rate": 1e-06, + "loss": 1.025, + "mean_token_accuracy": 0.6859040856361389, + "num_tokens": 164394784.0, + "step": 6436 + }, + { + "epoch": 0.7068965517241379, + "grad_norm": 2.051715612411499, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6979950666427612, + "num_tokens": 164425300.0, + "step": 6437 + }, + { + "epoch": 0.7070063694267515, + "grad_norm": 2.3923559188842773, + "learning_rate": 1e-06, + "loss": 1.0541, + "mean_token_accuracy": 0.6813509464263916, + "num_tokens": 164450374.0, + "step": 6438 + }, + { + "epoch": 0.7071161871293653, + "grad_norm": 2.2206833362579346, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6902644634246826, + "num_tokens": 164476428.0, + "step": 6439 + }, + { + "epoch": 0.707226004831979, + "grad_norm": 1.917110562324524, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7074451446533203, + "num_tokens": 164507137.0, + "step": 6440 + }, + { + "epoch": 0.7073358225345926, + "grad_norm": 2.2043464183807373, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7367275953292847, + "num_tokens": 164528771.0, + "step": 6441 + }, + { + "epoch": 0.7074456402372062, + "grad_norm": 2.089224338531494, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7129830121994019, + "num_tokens": 164554778.0, + "step": 6442 + }, + { + "epoch": 0.7075554579398199, + "grad_norm": 2.1703951358795166, + "learning_rate": 1e-06, + "loss": 1.007, + "mean_token_accuracy": 0.7019292116165161, + "num_tokens": 164581211.0, + "step": 6443 + }, + { + "epoch": 0.7076652756424335, + "grad_norm": 2.1298906803131104, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6976405382156372, + "num_tokens": 164607638.0, + "step": 6444 + }, + { + "epoch": 0.7077750933450472, + "grad_norm": 2.377636194229126, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6896327137947083, + "num_tokens": 164629689.0, + "step": 6445 + }, + { + "epoch": 0.7078849110476609, + "grad_norm": 2.2452855110168457, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.6987004280090332, + "num_tokens": 164653764.0, + "step": 6446 + }, + { + "epoch": 0.7079947287502746, + "grad_norm": 1.9733706712722778, + "learning_rate": 1e-06, + "loss": 1.0644, + "mean_token_accuracy": 0.6818358898162842, + "num_tokens": 164687602.0, + "step": 6447 + }, + { + "epoch": 0.7081045464528882, + "grad_norm": 1.981452465057373, + "learning_rate": 1e-06, + "loss": 1.0525, + "mean_token_accuracy": 0.6843647360801697, + "num_tokens": 164720706.0, + "step": 6448 + }, + { + "epoch": 0.7082143641555019, + "grad_norm": 2.345886707305908, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7381086349487305, + "num_tokens": 164742255.0, + "step": 6449 + }, + { + "epoch": 0.7083241818581155, + "grad_norm": 2.132861614227295, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.7085322141647339, + "num_tokens": 164766577.0, + "step": 6450 + }, + { + "epoch": 0.7084339995607292, + "grad_norm": 2.4049715995788574, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.715988278388977, + "num_tokens": 164787040.0, + "step": 6451 + }, + { + "epoch": 0.7085438172633428, + "grad_norm": 2.063602924346924, + "learning_rate": 1e-06, + "loss": 1.0406, + "mean_token_accuracy": 0.6997270584106445, + "num_tokens": 164816382.0, + "step": 6452 + }, + { + "epoch": 0.7086536349659565, + "grad_norm": 2.347825050354004, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7040706276893616, + "num_tokens": 164840610.0, + "step": 6453 + }, + { + "epoch": 0.7087634526685702, + "grad_norm": 2.630156993865967, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7087210416793823, + "num_tokens": 164860856.0, + "step": 6454 + }, + { + "epoch": 0.7088732703711839, + "grad_norm": 2.6889281272888184, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7281712293624878, + "num_tokens": 164878412.0, + "step": 6455 + }, + { + "epoch": 0.7089830880737975, + "grad_norm": 2.2021472454071045, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7031633853912354, + "num_tokens": 164903219.0, + "step": 6456 + }, + { + "epoch": 0.7090929057764112, + "grad_norm": 2.1780667304992676, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7239129543304443, + "num_tokens": 164928199.0, + "step": 6457 + }, + { + "epoch": 0.7092027234790248, + "grad_norm": 2.085627794265747, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7127014994621277, + "num_tokens": 164956056.0, + "step": 6458 + }, + { + "epoch": 0.7093125411816384, + "grad_norm": 2.1370489597320557, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6987764835357666, + "num_tokens": 164982791.0, + "step": 6459 + }, + { + "epoch": 0.7094223588842521, + "grad_norm": 2.2492735385894775, + "learning_rate": 1e-06, + "loss": 1.073, + "mean_token_accuracy": 0.6762462258338928, + "num_tokens": 165008052.0, + "step": 6460 + }, + { + "epoch": 0.7095321765868658, + "grad_norm": 2.1308929920196533, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7075533866882324, + "num_tokens": 165032316.0, + "step": 6461 + }, + { + "epoch": 0.7096419942894795, + "grad_norm": 2.047173261642456, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.694568932056427, + "num_tokens": 165059357.0, + "step": 6462 + }, + { + "epoch": 0.7097518119920931, + "grad_norm": 2.4057281017303467, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.706278383731842, + "num_tokens": 165082313.0, + "step": 6463 + }, + { + "epoch": 0.7098616296947068, + "grad_norm": 2.023515462875366, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6952130794525146, + "num_tokens": 165112284.0, + "step": 6464 + }, + { + "epoch": 0.7099714473973204, + "grad_norm": 2.1632354259490967, + "learning_rate": 1e-06, + "loss": 1.0544, + "mean_token_accuracy": 0.6731860041618347, + "num_tokens": 165137477.0, + "step": 6465 + }, + { + "epoch": 0.7100812650999341, + "grad_norm": 2.086925983428955, + "learning_rate": 1e-06, + "loss": 1.0448, + "mean_token_accuracy": 0.690534770488739, + "num_tokens": 165165692.0, + "step": 6466 + }, + { + "epoch": 0.7101910828025477, + "grad_norm": 2.18351411819458, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6948575377464294, + "num_tokens": 165192672.0, + "step": 6467 + }, + { + "epoch": 0.7103009005051615, + "grad_norm": 3.45125675201416, + "learning_rate": 1e-06, + "loss": 1.0891, + "mean_token_accuracy": 0.678306519985199, + "num_tokens": 165223514.0, + "step": 6468 + }, + { + "epoch": 0.7104107182077751, + "grad_norm": 2.020885705947876, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.6886826753616333, + "num_tokens": 165255331.0, + "step": 6469 + }, + { + "epoch": 0.7105205359103888, + "grad_norm": 1.9120159149169922, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7011874318122864, + "num_tokens": 165289070.0, + "step": 6470 + }, + { + "epoch": 0.7106303536130024, + "grad_norm": 2.209561347961426, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7030687928199768, + "num_tokens": 165314326.0, + "step": 6471 + }, + { + "epoch": 0.7107401713156161, + "grad_norm": 2.3926613330841064, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.705100417137146, + "num_tokens": 165336035.0, + "step": 6472 + }, + { + "epoch": 0.7108499890182297, + "grad_norm": 2.0044748783111572, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7217378616333008, + "num_tokens": 165363050.0, + "step": 6473 + }, + { + "epoch": 0.7109598067208434, + "grad_norm": 1.9979803562164307, + "learning_rate": 1e-06, + "loss": 1.0555, + "mean_token_accuracy": 0.6875229477882385, + "num_tokens": 165394051.0, + "step": 6474 + }, + { + "epoch": 0.7110696244234571, + "grad_norm": 2.2463161945343018, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7079447507858276, + "num_tokens": 165417286.0, + "step": 6475 + }, + { + "epoch": 0.7111794421260708, + "grad_norm": 2.044257164001465, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7126815319061279, + "num_tokens": 165444641.0, + "step": 6476 + }, + { + "epoch": 0.7112892598286844, + "grad_norm": 2.0592024326324463, + "learning_rate": 1e-06, + "loss": 1.0877, + "mean_token_accuracy": 0.6825616955757141, + "num_tokens": 165474819.0, + "step": 6477 + }, + { + "epoch": 0.711399077531298, + "grad_norm": 2.3343114852905273, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7003850340843201, + "num_tokens": 165499786.0, + "step": 6478 + }, + { + "epoch": 0.7115088952339117, + "grad_norm": 1.9257293939590454, + "learning_rate": 1e-06, + "loss": 1.0249, + "mean_token_accuracy": 0.6860222220420837, + "num_tokens": 165531874.0, + "step": 6479 + }, + { + "epoch": 0.7116187129365253, + "grad_norm": 2.2335188388824463, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7105407118797302, + "num_tokens": 165558146.0, + "step": 6480 + }, + { + "epoch": 0.711728530639139, + "grad_norm": 1.9545207023620605, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.6953070759773254, + "num_tokens": 165588932.0, + "step": 6481 + }, + { + "epoch": 0.7118383483417526, + "grad_norm": 2.0733866691589355, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7004303932189941, + "num_tokens": 165616261.0, + "step": 6482 + }, + { + "epoch": 0.7119481660443664, + "grad_norm": 2.2765822410583496, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7045267224311829, + "num_tokens": 165639721.0, + "step": 6483 + }, + { + "epoch": 0.71205798374698, + "grad_norm": 2.2828872203826904, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.6975790858268738, + "num_tokens": 165662513.0, + "step": 6484 + }, + { + "epoch": 0.7121678014495937, + "grad_norm": 2.1351218223571777, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.6985158920288086, + "num_tokens": 165691021.0, + "step": 6485 + }, + { + "epoch": 0.7122776191522073, + "grad_norm": 2.0764000415802, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7008771896362305, + "num_tokens": 165716964.0, + "step": 6486 + }, + { + "epoch": 0.712387436854821, + "grad_norm": 2.353065252304077, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7186572551727295, + "num_tokens": 165739560.0, + "step": 6487 + }, + { + "epoch": 0.7124972545574346, + "grad_norm": 2.406020402908325, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7165974378585815, + "num_tokens": 165761888.0, + "step": 6488 + }, + { + "epoch": 0.7126070722600483, + "grad_norm": 2.568243980407715, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7266442775726318, + "num_tokens": 165781196.0, + "step": 6489 + }, + { + "epoch": 0.712716889962662, + "grad_norm": 2.168158769607544, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.709627091884613, + "num_tokens": 165805127.0, + "step": 6490 + }, + { + "epoch": 0.7128267076652757, + "grad_norm": 2.1958906650543213, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7184732556343079, + "num_tokens": 165832010.0, + "step": 6491 + }, + { + "epoch": 0.7129365253678893, + "grad_norm": 2.456857681274414, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7197962403297424, + "num_tokens": 165851713.0, + "step": 6492 + }, + { + "epoch": 0.713046343070503, + "grad_norm": 2.0350561141967773, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.7032518982887268, + "num_tokens": 165883617.0, + "step": 6493 + }, + { + "epoch": 0.7131561607731166, + "grad_norm": 2.1621170043945312, + "learning_rate": 1e-06, + "loss": 1.0896, + "mean_token_accuracy": 0.6751837134361267, + "num_tokens": 165910892.0, + "step": 6494 + }, + { + "epoch": 0.7132659784757303, + "grad_norm": 2.0573554039001465, + "learning_rate": 1e-06, + "loss": 1.0406, + "mean_token_accuracy": 0.6842668652534485, + "num_tokens": 165939783.0, + "step": 6495 + }, + { + "epoch": 0.7133757961783439, + "grad_norm": 2.025139093399048, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6912363767623901, + "num_tokens": 165970129.0, + "step": 6496 + }, + { + "epoch": 0.7134856138809577, + "grad_norm": 2.204866409301758, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7257002592086792, + "num_tokens": 165995726.0, + "step": 6497 + }, + { + "epoch": 0.7135954315835713, + "grad_norm": 2.3464512825012207, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7061240673065186, + "num_tokens": 166020750.0, + "step": 6498 + }, + { + "epoch": 0.713705249286185, + "grad_norm": 2.2967793941497803, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.6988520622253418, + "num_tokens": 166045320.0, + "step": 6499 + }, + { + "epoch": 0.7138150669887986, + "grad_norm": 1.924999475479126, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7069196701049805, + "num_tokens": 166075637.0, + "step": 6500 + }, + { + "epoch": 0.7139248846914122, + "grad_norm": 2.1186022758483887, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6979837417602539, + "num_tokens": 166103024.0, + "step": 6501 + }, + { + "epoch": 0.7140347023940259, + "grad_norm": 2.142996072769165, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.6969772577285767, + "num_tokens": 166129781.0, + "step": 6502 + }, + { + "epoch": 0.7141445200966395, + "grad_norm": 2.1762285232543945, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7084766030311584, + "num_tokens": 166154132.0, + "step": 6503 + }, + { + "epoch": 0.7142543377992533, + "grad_norm": 2.3013477325439453, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7121838331222534, + "num_tokens": 166178120.0, + "step": 6504 + }, + { + "epoch": 0.7143641555018669, + "grad_norm": 1.8824199438095093, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7093285322189331, + "num_tokens": 166209521.0, + "step": 6505 + }, + { + "epoch": 0.7144739732044806, + "grad_norm": 2.447396755218506, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7377341985702515, + "num_tokens": 166229817.0, + "step": 6506 + }, + { + "epoch": 0.7145837909070942, + "grad_norm": 2.14850115776062, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7054668664932251, + "num_tokens": 166255968.0, + "step": 6507 + }, + { + "epoch": 0.7146936086097079, + "grad_norm": 2.0007362365722656, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.69477379322052, + "num_tokens": 166284504.0, + "step": 6508 + }, + { + "epoch": 0.7148034263123215, + "grad_norm": 2.1315674781799316, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7007845640182495, + "num_tokens": 166311847.0, + "step": 6509 + }, + { + "epoch": 0.7149132440149352, + "grad_norm": 1.8737661838531494, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.705341100692749, + "num_tokens": 166344867.0, + "step": 6510 + }, + { + "epoch": 0.7150230617175488, + "grad_norm": 2.3390443325042725, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7022144794464111, + "num_tokens": 166368954.0, + "step": 6511 + }, + { + "epoch": 0.7151328794201626, + "grad_norm": 2.1887457370758057, + "learning_rate": 1e-06, + "loss": 1.0723, + "mean_token_accuracy": 0.6800045371055603, + "num_tokens": 166398233.0, + "step": 6512 + }, + { + "epoch": 0.7152426971227762, + "grad_norm": 2.2785301208496094, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6913977265357971, + "num_tokens": 166424697.0, + "step": 6513 + }, + { + "epoch": 0.7153525148253899, + "grad_norm": 2.3102002143859863, + "learning_rate": 1e-06, + "loss": 1.0309, + "mean_token_accuracy": 0.6850275993347168, + "num_tokens": 166450325.0, + "step": 6514 + }, + { + "epoch": 0.7154623325280035, + "grad_norm": 2.1690099239349365, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.6923770904541016, + "num_tokens": 166476871.0, + "step": 6515 + }, + { + "epoch": 0.7155721502306172, + "grad_norm": 2.2454965114593506, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7180432677268982, + "num_tokens": 166499501.0, + "step": 6516 + }, + { + "epoch": 0.7156819679332308, + "grad_norm": 2.078648090362549, + "learning_rate": 1e-06, + "loss": 1.1213, + "mean_token_accuracy": 0.6606951951980591, + "num_tokens": 166528704.0, + "step": 6517 + }, + { + "epoch": 0.7157917856358444, + "grad_norm": 2.232349395751953, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7102149128913879, + "num_tokens": 166552071.0, + "step": 6518 + }, + { + "epoch": 0.7159016033384582, + "grad_norm": 2.3522167205810547, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.7174291610717773, + "num_tokens": 166574740.0, + "step": 6519 + }, + { + "epoch": 0.7160114210410718, + "grad_norm": 2.3575119972229004, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7200202941894531, + "num_tokens": 166597424.0, + "step": 6520 + }, + { + "epoch": 0.7161212387436855, + "grad_norm": 1.9054869413375854, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7236236333847046, + "num_tokens": 166628441.0, + "step": 6521 + }, + { + "epoch": 0.7162310564462991, + "grad_norm": 2.326632261276245, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7158005237579346, + "num_tokens": 166651304.0, + "step": 6522 + }, + { + "epoch": 0.7163408741489128, + "grad_norm": 2.0250422954559326, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7187851071357727, + "num_tokens": 166679028.0, + "step": 6523 + }, + { + "epoch": 0.7164506918515264, + "grad_norm": 2.2144739627838135, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7094797492027283, + "num_tokens": 166704466.0, + "step": 6524 + }, + { + "epoch": 0.7165605095541401, + "grad_norm": 2.464679718017578, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.716696560382843, + "num_tokens": 166725327.0, + "step": 6525 + }, + { + "epoch": 0.7166703272567538, + "grad_norm": 2.5523226261138916, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7222847938537598, + "num_tokens": 166744450.0, + "step": 6526 + }, + { + "epoch": 0.7167801449593675, + "grad_norm": 2.17032790184021, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.6994573473930359, + "num_tokens": 166768728.0, + "step": 6527 + }, + { + "epoch": 0.7168899626619811, + "grad_norm": 2.29557204246521, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7078250646591187, + "num_tokens": 166791364.0, + "step": 6528 + }, + { + "epoch": 0.7169997803645948, + "grad_norm": 1.9162484407424927, + "learning_rate": 1e-06, + "loss": 1.0498, + "mean_token_accuracy": 0.6822890043258667, + "num_tokens": 166824041.0, + "step": 6529 + }, + { + "epoch": 0.7171095980672084, + "grad_norm": 2.081862449645996, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6917417049407959, + "num_tokens": 166852055.0, + "step": 6530 + }, + { + "epoch": 0.7172194157698221, + "grad_norm": 2.2017760276794434, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7128459811210632, + "num_tokens": 166875720.0, + "step": 6531 + }, + { + "epoch": 0.7173292334724357, + "grad_norm": 2.2391116619110107, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6907281875610352, + "num_tokens": 166900954.0, + "step": 6532 + }, + { + "epoch": 0.7174390511750495, + "grad_norm": 2.4278414249420166, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7150005102157593, + "num_tokens": 166924121.0, + "step": 6533 + }, + { + "epoch": 0.7175488688776631, + "grad_norm": 2.2925310134887695, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7042592763900757, + "num_tokens": 166948511.0, + "step": 6534 + }, + { + "epoch": 0.7176586865802768, + "grad_norm": 2.240213632583618, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7397880554199219, + "num_tokens": 166971395.0, + "step": 6535 + }, + { + "epoch": 0.7177685042828904, + "grad_norm": 2.295483112335205, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7120667099952698, + "num_tokens": 166995631.0, + "step": 6536 + }, + { + "epoch": 0.717878321985504, + "grad_norm": 2.27588152885437, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7083296179771423, + "num_tokens": 167019687.0, + "step": 6537 + }, + { + "epoch": 0.7179881396881177, + "grad_norm": 2.5687954425811768, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7168639302253723, + "num_tokens": 167037821.0, + "step": 6538 + }, + { + "epoch": 0.7180979573907313, + "grad_norm": 2.3095507621765137, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.7001419067382812, + "num_tokens": 167061303.0, + "step": 6539 + }, + { + "epoch": 0.7182077750933451, + "grad_norm": 2.4823408126831055, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.726304292678833, + "num_tokens": 167079832.0, + "step": 6540 + }, + { + "epoch": 0.7183175927959587, + "grad_norm": 2.2390453815460205, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6885086297988892, + "num_tokens": 167105331.0, + "step": 6541 + }, + { + "epoch": 0.7184274104985724, + "grad_norm": 2.2358195781707764, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.715572714805603, + "num_tokens": 167129654.0, + "step": 6542 + }, + { + "epoch": 0.718537228201186, + "grad_norm": 2.12221622467041, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7344759106636047, + "num_tokens": 167153754.0, + "step": 6543 + }, + { + "epoch": 0.7186470459037997, + "grad_norm": 2.5346577167510986, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7111213207244873, + "num_tokens": 167175390.0, + "step": 6544 + }, + { + "epoch": 0.7187568636064133, + "grad_norm": 2.3666274547576904, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7335221767425537, + "num_tokens": 167196247.0, + "step": 6545 + }, + { + "epoch": 0.718866681309027, + "grad_norm": 2.148432970046997, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6818736791610718, + "num_tokens": 167222199.0, + "step": 6546 + }, + { + "epoch": 0.7189764990116406, + "grad_norm": 2.6004104614257812, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7224549055099487, + "num_tokens": 167242839.0, + "step": 6547 + }, + { + "epoch": 0.7190863167142544, + "grad_norm": 2.195857286453247, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6904602646827698, + "num_tokens": 167270395.0, + "step": 6548 + }, + { + "epoch": 0.719196134416868, + "grad_norm": 2.3971147537231445, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7190099954605103, + "num_tokens": 167292924.0, + "step": 6549 + }, + { + "epoch": 0.7193059521194817, + "grad_norm": 2.077439546585083, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7000786066055298, + "num_tokens": 167321578.0, + "step": 6550 + }, + { + "epoch": 0.7194157698220953, + "grad_norm": 2.136925458908081, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7142090797424316, + "num_tokens": 167347656.0, + "step": 6551 + }, + { + "epoch": 0.719525587524709, + "grad_norm": 2.241116762161255, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7047643065452576, + "num_tokens": 167372483.0, + "step": 6552 + }, + { + "epoch": 0.7196354052273226, + "grad_norm": 2.089782476425171, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.6971247792243958, + "num_tokens": 167400769.0, + "step": 6553 + }, + { + "epoch": 0.7197452229299363, + "grad_norm": 2.1974027156829834, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.6948577165603638, + "num_tokens": 167426414.0, + "step": 6554 + }, + { + "epoch": 0.71985504063255, + "grad_norm": 2.4214956760406494, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7003908157348633, + "num_tokens": 167448910.0, + "step": 6555 + }, + { + "epoch": 0.7199648583351637, + "grad_norm": 2.1940858364105225, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7155857086181641, + "num_tokens": 167473387.0, + "step": 6556 + }, + { + "epoch": 0.7200746760377773, + "grad_norm": 2.04011869430542, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7178676724433899, + "num_tokens": 167500455.0, + "step": 6557 + }, + { + "epoch": 0.720184493740391, + "grad_norm": 2.4069433212280273, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7133681774139404, + "num_tokens": 167522831.0, + "step": 6558 + }, + { + "epoch": 0.7202943114430046, + "grad_norm": 2.5286576747894287, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7195756435394287, + "num_tokens": 167541669.0, + "step": 6559 + }, + { + "epoch": 0.7204041291456182, + "grad_norm": 2.2972524166107178, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7203911542892456, + "num_tokens": 167565447.0, + "step": 6560 + }, + { + "epoch": 0.7205139468482319, + "grad_norm": 2.16296648979187, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7181299924850464, + "num_tokens": 167589317.0, + "step": 6561 + }, + { + "epoch": 0.7206237645508456, + "grad_norm": 2.091930866241455, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7199075222015381, + "num_tokens": 167614318.0, + "step": 6562 + }, + { + "epoch": 0.7207335822534593, + "grad_norm": 2.4063520431518555, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7040470838546753, + "num_tokens": 167635678.0, + "step": 6563 + }, + { + "epoch": 0.7208433999560729, + "grad_norm": 2.036130666732788, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7204554080963135, + "num_tokens": 167664559.0, + "step": 6564 + }, + { + "epoch": 0.7209532176586866, + "grad_norm": 2.0946879386901855, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7261850833892822, + "num_tokens": 167692089.0, + "step": 6565 + }, + { + "epoch": 0.7210630353613002, + "grad_norm": 2.2735483646392822, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.6987799406051636, + "num_tokens": 167717022.0, + "step": 6566 + }, + { + "epoch": 0.7211728530639139, + "grad_norm": 2.029452323913574, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7084723711013794, + "num_tokens": 167746130.0, + "step": 6567 + }, + { + "epoch": 0.7212826707665275, + "grad_norm": 2.148413896560669, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6920669674873352, + "num_tokens": 167775235.0, + "step": 6568 + }, + { + "epoch": 0.7213924884691413, + "grad_norm": 2.409397602081299, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7262705564498901, + "num_tokens": 167796132.0, + "step": 6569 + }, + { + "epoch": 0.7215023061717549, + "grad_norm": 2.123555898666382, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6937485933303833, + "num_tokens": 167825075.0, + "step": 6570 + }, + { + "epoch": 0.7216121238743686, + "grad_norm": 2.1144137382507324, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7154814600944519, + "num_tokens": 167852126.0, + "step": 6571 + }, + { + "epoch": 0.7217219415769822, + "grad_norm": 2.135500431060791, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7083311080932617, + "num_tokens": 167877129.0, + "step": 6572 + }, + { + "epoch": 0.7218317592795959, + "grad_norm": 2.320291042327881, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7123281955718994, + "num_tokens": 167899858.0, + "step": 6573 + }, + { + "epoch": 0.7219415769822095, + "grad_norm": 1.9654135704040527, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7099294662475586, + "num_tokens": 167930602.0, + "step": 6574 + }, + { + "epoch": 0.7220513946848232, + "grad_norm": 2.1901376247406006, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.6975574493408203, + "num_tokens": 167955456.0, + "step": 6575 + }, + { + "epoch": 0.7221612123874368, + "grad_norm": 2.3335418701171875, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.708695113658905, + "num_tokens": 167980147.0, + "step": 6576 + }, + { + "epoch": 0.7222710300900506, + "grad_norm": 2.0940909385681152, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.6991730332374573, + "num_tokens": 168008246.0, + "step": 6577 + }, + { + "epoch": 0.7223808477926642, + "grad_norm": 2.323474168777466, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7134947776794434, + "num_tokens": 168031928.0, + "step": 6578 + }, + { + "epoch": 0.7224906654952779, + "grad_norm": 2.3002076148986816, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6967952251434326, + "num_tokens": 168057268.0, + "step": 6579 + }, + { + "epoch": 0.7226004831978915, + "grad_norm": 2.100659132003784, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7143502235412598, + "num_tokens": 168085095.0, + "step": 6580 + }, + { + "epoch": 0.7227103009005051, + "grad_norm": 2.199991464614868, + "learning_rate": 1e-06, + "loss": 1.0529, + "mean_token_accuracy": 0.6791256070137024, + "num_tokens": 168112057.0, + "step": 6581 + }, + { + "epoch": 0.7228201186031188, + "grad_norm": 2.325019598007202, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7043939828872681, + "num_tokens": 168134707.0, + "step": 6582 + }, + { + "epoch": 0.7229299363057324, + "grad_norm": 2.4462363719940186, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7041988968849182, + "num_tokens": 168154306.0, + "step": 6583 + }, + { + "epoch": 0.7230397540083462, + "grad_norm": 2.17608904838562, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.7023366093635559, + "num_tokens": 168180697.0, + "step": 6584 + }, + { + "epoch": 0.7231495717109598, + "grad_norm": 2.0166049003601074, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6881808042526245, + "num_tokens": 168210361.0, + "step": 6585 + }, + { + "epoch": 0.7232593894135735, + "grad_norm": 2.3664302825927734, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6984085440635681, + "num_tokens": 168233001.0, + "step": 6586 + }, + { + "epoch": 0.7233692071161871, + "grad_norm": 2.0195631980895996, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6952337622642517, + "num_tokens": 168262114.0, + "step": 6587 + }, + { + "epoch": 0.7234790248188008, + "grad_norm": 2.260937213897705, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7210211753845215, + "num_tokens": 168285995.0, + "step": 6588 + }, + { + "epoch": 0.7235888425214144, + "grad_norm": 2.1835529804229736, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7104482650756836, + "num_tokens": 168312067.0, + "step": 6589 + }, + { + "epoch": 0.7236986602240281, + "grad_norm": 2.241346597671509, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.711459755897522, + "num_tokens": 168334610.0, + "step": 6590 + }, + { + "epoch": 0.7238084779266418, + "grad_norm": 2.3438241481781006, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7023717164993286, + "num_tokens": 168358040.0, + "step": 6591 + }, + { + "epoch": 0.7239182956292555, + "grad_norm": 2.219790458679199, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7017540335655212, + "num_tokens": 168382684.0, + "step": 6592 + }, + { + "epoch": 0.7240281133318691, + "grad_norm": 2.303819179534912, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7248802185058594, + "num_tokens": 168406429.0, + "step": 6593 + }, + { + "epoch": 0.7241379310344828, + "grad_norm": 2.0693776607513428, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7057932019233704, + "num_tokens": 168434571.0, + "step": 6594 + }, + { + "epoch": 0.7242477487370964, + "grad_norm": 2.5733585357666016, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7214589715003967, + "num_tokens": 168453749.0, + "step": 6595 + }, + { + "epoch": 0.72435756643971, + "grad_norm": 2.2753257751464844, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7028486728668213, + "num_tokens": 168476413.0, + "step": 6596 + }, + { + "epoch": 0.7244673841423237, + "grad_norm": 2.2285444736480713, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7149226665496826, + "num_tokens": 168502122.0, + "step": 6597 + }, + { + "epoch": 0.7245772018449375, + "grad_norm": 2.0536928176879883, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.709267795085907, + "num_tokens": 168530038.0, + "step": 6598 + }, + { + "epoch": 0.7246870195475511, + "grad_norm": 2.0100419521331787, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.692513644695282, + "num_tokens": 168559559.0, + "step": 6599 + }, + { + "epoch": 0.7247968372501647, + "grad_norm": 2.2662034034729004, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7208808660507202, + "num_tokens": 168583671.0, + "step": 6600 + }, + { + "epoch": 0.7249066549527784, + "grad_norm": 2.335505723953247, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.7009838819503784, + "num_tokens": 168609098.0, + "step": 6601 + }, + { + "epoch": 0.725016472655392, + "grad_norm": 2.2670629024505615, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7085080146789551, + "num_tokens": 168632879.0, + "step": 6602 + }, + { + "epoch": 0.7251262903580057, + "grad_norm": 2.134585380554199, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6866640448570251, + "num_tokens": 168659334.0, + "step": 6603 + }, + { + "epoch": 0.7252361080606193, + "grad_norm": 2.16355562210083, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7356417775154114, + "num_tokens": 168681964.0, + "step": 6604 + }, + { + "epoch": 0.725345925763233, + "grad_norm": 2.34515118598938, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7290735244750977, + "num_tokens": 168702957.0, + "step": 6605 + }, + { + "epoch": 0.7254557434658467, + "grad_norm": 2.072897434234619, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7089154720306396, + "num_tokens": 168732312.0, + "step": 6606 + }, + { + "epoch": 0.7255655611684604, + "grad_norm": 2.3584017753601074, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7229568362236023, + "num_tokens": 168754514.0, + "step": 6607 + }, + { + "epoch": 0.725675378871074, + "grad_norm": 2.163959264755249, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7194387912750244, + "num_tokens": 168780666.0, + "step": 6608 + }, + { + "epoch": 0.7257851965736877, + "grad_norm": 2.053670883178711, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6941443681716919, + "num_tokens": 168810988.0, + "step": 6609 + }, + { + "epoch": 0.7258950142763013, + "grad_norm": 2.4740891456604004, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.729422926902771, + "num_tokens": 168829061.0, + "step": 6610 + }, + { + "epoch": 0.726004831978915, + "grad_norm": 2.217958450317383, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7274643778800964, + "num_tokens": 168853307.0, + "step": 6611 + }, + { + "epoch": 0.7261146496815286, + "grad_norm": 2.235811471939087, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.705135703086853, + "num_tokens": 168876876.0, + "step": 6612 + }, + { + "epoch": 0.7262244673841424, + "grad_norm": 2.196350574493408, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6950769424438477, + "num_tokens": 168903141.0, + "step": 6613 + }, + { + "epoch": 0.726334285086756, + "grad_norm": 2.1354827880859375, + "learning_rate": 1e-06, + "loss": 1.0347, + "mean_token_accuracy": 0.6806291341781616, + "num_tokens": 168929211.0, + "step": 6614 + }, + { + "epoch": 0.7264441027893697, + "grad_norm": 2.6135170459747314, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7068318724632263, + "num_tokens": 168949588.0, + "step": 6615 + }, + { + "epoch": 0.7265539204919833, + "grad_norm": 1.9737194776535034, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7230901122093201, + "num_tokens": 168979922.0, + "step": 6616 + }, + { + "epoch": 0.726663738194597, + "grad_norm": 2.5325660705566406, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7201030254364014, + "num_tokens": 169001118.0, + "step": 6617 + }, + { + "epoch": 0.7267735558972106, + "grad_norm": 2.290893077850342, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6893245577812195, + "num_tokens": 169024345.0, + "step": 6618 + }, + { + "epoch": 0.7268833735998242, + "grad_norm": 2.3321282863616943, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6964080333709717, + "num_tokens": 169048160.0, + "step": 6619 + }, + { + "epoch": 0.726993191302438, + "grad_norm": 1.9951363801956177, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7200882434844971, + "num_tokens": 169076531.0, + "step": 6620 + }, + { + "epoch": 0.7271030090050516, + "grad_norm": 1.9884706735610962, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7175359129905701, + "num_tokens": 169107888.0, + "step": 6621 + }, + { + "epoch": 0.7272128267076653, + "grad_norm": 1.9456186294555664, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.7004824876785278, + "num_tokens": 169138331.0, + "step": 6622 + }, + { + "epoch": 0.7273226444102789, + "grad_norm": 2.2572543621063232, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.6949574947357178, + "num_tokens": 169165118.0, + "step": 6623 + }, + { + "epoch": 0.7274324621128926, + "grad_norm": 2.1261308193206787, + "learning_rate": 1e-06, + "loss": 1.0632, + "mean_token_accuracy": 0.6786578297615051, + "num_tokens": 169191503.0, + "step": 6624 + }, + { + "epoch": 0.7275422798155062, + "grad_norm": 2.3948557376861572, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6929107308387756, + "num_tokens": 169213488.0, + "step": 6625 + }, + { + "epoch": 0.7276520975181199, + "grad_norm": 2.3699872493743896, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7325634956359863, + "num_tokens": 169234493.0, + "step": 6626 + }, + { + "epoch": 0.7277619152207336, + "grad_norm": 1.8668582439422607, + "learning_rate": 1e-06, + "loss": 1.0218, + "mean_token_accuracy": 0.696233868598938, + "num_tokens": 169267790.0, + "step": 6627 + }, + { + "epoch": 0.7278717329233473, + "grad_norm": 2.43251371383667, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7224711775779724, + "num_tokens": 169289959.0, + "step": 6628 + }, + { + "epoch": 0.7279815506259609, + "grad_norm": 2.0630505084991455, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.6978812217712402, + "num_tokens": 169318567.0, + "step": 6629 + }, + { + "epoch": 0.7280913683285746, + "grad_norm": 2.3503246307373047, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7122563123703003, + "num_tokens": 169340187.0, + "step": 6630 + }, + { + "epoch": 0.7282011860311882, + "grad_norm": 1.885705590248108, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7099802494049072, + "num_tokens": 169371852.0, + "step": 6631 + }, + { + "epoch": 0.7283110037338019, + "grad_norm": 2.1488542556762695, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.6985834836959839, + "num_tokens": 169395838.0, + "step": 6632 + }, + { + "epoch": 0.7284208214364155, + "grad_norm": 2.3052937984466553, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7448431253433228, + "num_tokens": 169416707.0, + "step": 6633 + }, + { + "epoch": 0.7285306391390292, + "grad_norm": 2.3684446811676025, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7155101895332336, + "num_tokens": 169439337.0, + "step": 6634 + }, + { + "epoch": 0.7286404568416429, + "grad_norm": 2.241312265396118, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7154764533042908, + "num_tokens": 169463633.0, + "step": 6635 + }, + { + "epoch": 0.7287502745442566, + "grad_norm": 1.9529778957366943, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.6948935985565186, + "num_tokens": 169495212.0, + "step": 6636 + }, + { + "epoch": 0.7288600922468702, + "grad_norm": 2.314171075820923, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7072259187698364, + "num_tokens": 169518396.0, + "step": 6637 + }, + { + "epoch": 0.7289699099494839, + "grad_norm": 1.9896314144134521, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7141691446304321, + "num_tokens": 169547231.0, + "step": 6638 + }, + { + "epoch": 0.7290797276520975, + "grad_norm": 2.170328378677368, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7233569622039795, + "num_tokens": 169570927.0, + "step": 6639 + }, + { + "epoch": 0.7291895453547111, + "grad_norm": 2.1884050369262695, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7016357183456421, + "num_tokens": 169596099.0, + "step": 6640 + }, + { + "epoch": 0.7292993630573248, + "grad_norm": 2.322205066680908, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.7120645046234131, + "num_tokens": 169619536.0, + "step": 6641 + }, + { + "epoch": 0.7294091807599385, + "grad_norm": 2.1509203910827637, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.6901477575302124, + "num_tokens": 169645553.0, + "step": 6642 + }, + { + "epoch": 0.7295189984625522, + "grad_norm": 2.0056333541870117, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6978133916854858, + "num_tokens": 169675169.0, + "step": 6643 + }, + { + "epoch": 0.7296288161651658, + "grad_norm": 2.063354253768921, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7016423940658569, + "num_tokens": 169702300.0, + "step": 6644 + }, + { + "epoch": 0.7297386338677795, + "grad_norm": 2.102370500564575, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7051463723182678, + "num_tokens": 169730886.0, + "step": 6645 + }, + { + "epoch": 0.7298484515703931, + "grad_norm": 1.8881199359893799, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7062208652496338, + "num_tokens": 169763012.0, + "step": 6646 + }, + { + "epoch": 0.7299582692730068, + "grad_norm": 2.1274807453155518, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7041673064231873, + "num_tokens": 169786602.0, + "step": 6647 + }, + { + "epoch": 0.7300680869756204, + "grad_norm": 2.440537452697754, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7149666547775269, + "num_tokens": 169807680.0, + "step": 6648 + }, + { + "epoch": 0.7301779046782342, + "grad_norm": 2.478581666946411, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7423630952835083, + "num_tokens": 169826188.0, + "step": 6649 + }, + { + "epoch": 0.7302877223808478, + "grad_norm": 2.141087293624878, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7289336323738098, + "num_tokens": 169850041.0, + "step": 6650 + }, + { + "epoch": 0.7303975400834615, + "grad_norm": 2.015869379043579, + "learning_rate": 1e-06, + "loss": 1.0739, + "mean_token_accuracy": 0.6793984174728394, + "num_tokens": 169882638.0, + "step": 6651 + }, + { + "epoch": 0.7305073577860751, + "grad_norm": 2.192739963531494, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7211424708366394, + "num_tokens": 169905729.0, + "step": 6652 + }, + { + "epoch": 0.7306171754886888, + "grad_norm": 2.321079730987549, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7124115824699402, + "num_tokens": 169927303.0, + "step": 6653 + }, + { + "epoch": 0.7307269931913024, + "grad_norm": 2.1306076049804688, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7261172533035278, + "num_tokens": 169951696.0, + "step": 6654 + }, + { + "epoch": 0.7308368108939161, + "grad_norm": 2.318937301635742, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7018424272537231, + "num_tokens": 169975393.0, + "step": 6655 + }, + { + "epoch": 0.7309466285965298, + "grad_norm": 2.297147035598755, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7091328501701355, + "num_tokens": 169998486.0, + "step": 6656 + }, + { + "epoch": 0.7310564462991435, + "grad_norm": 2.326580762863159, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6921843886375427, + "num_tokens": 170023642.0, + "step": 6657 + }, + { + "epoch": 0.7311662640017571, + "grad_norm": 2.2617993354797363, + "learning_rate": 1e-06, + "loss": 1.0503, + "mean_token_accuracy": 0.6900471448898315, + "num_tokens": 170049625.0, + "step": 6658 + }, + { + "epoch": 0.7312760817043708, + "grad_norm": 2.0667195320129395, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7031692266464233, + "num_tokens": 170079323.0, + "step": 6659 + }, + { + "epoch": 0.7313858994069844, + "grad_norm": 2.122952699661255, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.697158694267273, + "num_tokens": 170107125.0, + "step": 6660 + }, + { + "epoch": 0.731495717109598, + "grad_norm": 2.0988335609436035, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.6993369460105896, + "num_tokens": 170136604.0, + "step": 6661 + }, + { + "epoch": 0.7316055348122117, + "grad_norm": 2.3951964378356934, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7241882085800171, + "num_tokens": 170158398.0, + "step": 6662 + }, + { + "epoch": 0.7317153525148253, + "grad_norm": 2.068316698074341, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6941769123077393, + "num_tokens": 170186278.0, + "step": 6663 + }, + { + "epoch": 0.7318251702174391, + "grad_norm": 2.209784746170044, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7288556098937988, + "num_tokens": 170210684.0, + "step": 6664 + }, + { + "epoch": 0.7319349879200527, + "grad_norm": 2.1871538162231445, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6982841491699219, + "num_tokens": 170238935.0, + "step": 6665 + }, + { + "epoch": 0.7320448056226664, + "grad_norm": 2.5393362045288086, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.739044189453125, + "num_tokens": 170257674.0, + "step": 6666 + }, + { + "epoch": 0.73215462332528, + "grad_norm": 2.7358639240264893, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7295068502426147, + "num_tokens": 170273975.0, + "step": 6667 + }, + { + "epoch": 0.7322644410278937, + "grad_norm": 1.9636077880859375, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.6993826031684875, + "num_tokens": 170303568.0, + "step": 6668 + }, + { + "epoch": 0.7323742587305073, + "grad_norm": 2.136028289794922, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6965494751930237, + "num_tokens": 170331783.0, + "step": 6669 + }, + { + "epoch": 0.732484076433121, + "grad_norm": 2.2139039039611816, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.717660665512085, + "num_tokens": 170355368.0, + "step": 6670 + }, + { + "epoch": 0.7325938941357347, + "grad_norm": 2.1143105030059814, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7048116326332092, + "num_tokens": 170381836.0, + "step": 6671 + }, + { + "epoch": 0.7327037118383484, + "grad_norm": 2.2729194164276123, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7149926424026489, + "num_tokens": 170405478.0, + "step": 6672 + }, + { + "epoch": 0.732813529540962, + "grad_norm": 1.9943667650222778, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7290912866592407, + "num_tokens": 170434294.0, + "step": 6673 + }, + { + "epoch": 0.7329233472435757, + "grad_norm": 1.939892053604126, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.6851552128791809, + "num_tokens": 170468122.0, + "step": 6674 + }, + { + "epoch": 0.7330331649461893, + "grad_norm": 2.4386050701141357, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7215496301651001, + "num_tokens": 170490738.0, + "step": 6675 + }, + { + "epoch": 0.733142982648803, + "grad_norm": 2.3625168800354004, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7166693806648254, + "num_tokens": 170513169.0, + "step": 6676 + }, + { + "epoch": 0.7332528003514166, + "grad_norm": 2.09454607963562, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7143393158912659, + "num_tokens": 170539421.0, + "step": 6677 + }, + { + "epoch": 0.7333626180540304, + "grad_norm": 2.109215021133423, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7265310287475586, + "num_tokens": 170565335.0, + "step": 6678 + }, + { + "epoch": 0.733472435756644, + "grad_norm": 2.424405813217163, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7133047580718994, + "num_tokens": 170585228.0, + "step": 6679 + }, + { + "epoch": 0.7335822534592576, + "grad_norm": 2.10375714302063, + "learning_rate": 1e-06, + "loss": 1.0581, + "mean_token_accuracy": 0.6814203262329102, + "num_tokens": 170612774.0, + "step": 6680 + }, + { + "epoch": 0.7336920711618713, + "grad_norm": 2.09295654296875, + "learning_rate": 1e-06, + "loss": 1.0532, + "mean_token_accuracy": 0.6826814413070679, + "num_tokens": 170641792.0, + "step": 6681 + }, + { + "epoch": 0.7338018888644849, + "grad_norm": 1.9054487943649292, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7136187553405762, + "num_tokens": 170673863.0, + "step": 6682 + }, + { + "epoch": 0.7339117065670986, + "grad_norm": 1.7974114418029785, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6917558908462524, + "num_tokens": 170712129.0, + "step": 6683 + }, + { + "epoch": 0.7340215242697122, + "grad_norm": 2.184168815612793, + "learning_rate": 1e-06, + "loss": 1.0509, + "mean_token_accuracy": 0.6923324465751648, + "num_tokens": 170738921.0, + "step": 6684 + }, + { + "epoch": 0.734131341972326, + "grad_norm": 2.229048252105713, + "learning_rate": 1e-06, + "loss": 1.0485, + "mean_token_accuracy": 0.6956989765167236, + "num_tokens": 170764274.0, + "step": 6685 + }, + { + "epoch": 0.7342411596749396, + "grad_norm": 2.0415561199188232, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7235676050186157, + "num_tokens": 170793500.0, + "step": 6686 + }, + { + "epoch": 0.7343509773775533, + "grad_norm": 2.087808847427368, + "learning_rate": 1e-06, + "loss": 1.0746, + "mean_token_accuracy": 0.6697378754615784, + "num_tokens": 170822886.0, + "step": 6687 + }, + { + "epoch": 0.7344607950801669, + "grad_norm": 2.316997528076172, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7148658037185669, + "num_tokens": 170848809.0, + "step": 6688 + }, + { + "epoch": 0.7345706127827806, + "grad_norm": 1.9836008548736572, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6957651376724243, + "num_tokens": 170877727.0, + "step": 6689 + }, + { + "epoch": 0.7346804304853942, + "grad_norm": 2.1746716499328613, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7090264558792114, + "num_tokens": 170904426.0, + "step": 6690 + }, + { + "epoch": 0.7347902481880079, + "grad_norm": 2.3242123126983643, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7208448052406311, + "num_tokens": 170926538.0, + "step": 6691 + }, + { + "epoch": 0.7349000658906216, + "grad_norm": 2.0993707180023193, + "learning_rate": 1e-06, + "loss": 1.0551, + "mean_token_accuracy": 0.6872323751449585, + "num_tokens": 170954600.0, + "step": 6692 + }, + { + "epoch": 0.7350098835932353, + "grad_norm": 2.0127367973327637, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.7041105628013611, + "num_tokens": 170983128.0, + "step": 6693 + }, + { + "epoch": 0.7351197012958489, + "grad_norm": 2.055986166000366, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7204331159591675, + "num_tokens": 171011744.0, + "step": 6694 + }, + { + "epoch": 0.7352295189984626, + "grad_norm": 2.4510068893432617, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7101829051971436, + "num_tokens": 171031533.0, + "step": 6695 + }, + { + "epoch": 0.7353393367010762, + "grad_norm": 2.303649425506592, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6880371570587158, + "num_tokens": 171057317.0, + "step": 6696 + }, + { + "epoch": 0.7354491544036899, + "grad_norm": 2.244800090789795, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.704010009765625, + "num_tokens": 171081723.0, + "step": 6697 + }, + { + "epoch": 0.7355589721063035, + "grad_norm": 2.2176668643951416, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6967133283615112, + "num_tokens": 171109219.0, + "step": 6698 + }, + { + "epoch": 0.7356687898089171, + "grad_norm": 2.239314079284668, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7157895565032959, + "num_tokens": 171134580.0, + "step": 6699 + }, + { + "epoch": 0.7357786075115309, + "grad_norm": 2.4118571281433105, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7237919569015503, + "num_tokens": 171155074.0, + "step": 6700 + }, + { + "epoch": 0.7358884252141445, + "grad_norm": 2.3155388832092285, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7099236249923706, + "num_tokens": 171177618.0, + "step": 6701 + }, + { + "epoch": 0.7359982429167582, + "grad_norm": 1.8899872303009033, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7198400497436523, + "num_tokens": 171209477.0, + "step": 6702 + }, + { + "epoch": 0.7361080606193718, + "grad_norm": 1.8756606578826904, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7128099203109741, + "num_tokens": 171241024.0, + "step": 6703 + }, + { + "epoch": 0.7362178783219855, + "grad_norm": 2.4818127155303955, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7314674854278564, + "num_tokens": 171261582.0, + "step": 6704 + }, + { + "epoch": 0.7363276960245991, + "grad_norm": 2.339703321456909, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7133957743644714, + "num_tokens": 171284379.0, + "step": 6705 + }, + { + "epoch": 0.7364375137272128, + "grad_norm": 2.6028003692626953, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7075939774513245, + "num_tokens": 171303952.0, + "step": 6706 + }, + { + "epoch": 0.7365473314298265, + "grad_norm": 2.229142665863037, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7285889387130737, + "num_tokens": 171327270.0, + "step": 6707 + }, + { + "epoch": 0.7366571491324402, + "grad_norm": 2.364051580429077, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7181224822998047, + "num_tokens": 171348413.0, + "step": 6708 + }, + { + "epoch": 0.7367669668350538, + "grad_norm": 2.1258366107940674, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6922000050544739, + "num_tokens": 171377592.0, + "step": 6709 + }, + { + "epoch": 0.7368767845376675, + "grad_norm": 2.477091073989868, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.6974092125892639, + "num_tokens": 171399543.0, + "step": 6710 + }, + { + "epoch": 0.7369866022402811, + "grad_norm": 2.4624080657958984, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7194459438323975, + "num_tokens": 171419682.0, + "step": 6711 + }, + { + "epoch": 0.7370964199428948, + "grad_norm": 2.0472681522369385, + "learning_rate": 1e-06, + "loss": 1.0391, + "mean_token_accuracy": 0.6903186440467834, + "num_tokens": 171449383.0, + "step": 6712 + }, + { + "epoch": 0.7372062376455084, + "grad_norm": 2.1397104263305664, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.7025889754295349, + "num_tokens": 171474883.0, + "step": 6713 + }, + { + "epoch": 0.7373160553481222, + "grad_norm": 2.093195676803589, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7178027629852295, + "num_tokens": 171500823.0, + "step": 6714 + }, + { + "epoch": 0.7374258730507358, + "grad_norm": 2.2116217613220215, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7032451033592224, + "num_tokens": 171530927.0, + "step": 6715 + }, + { + "epoch": 0.7375356907533495, + "grad_norm": 2.0944344997406006, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6973031759262085, + "num_tokens": 171558046.0, + "step": 6716 + }, + { + "epoch": 0.7376455084559631, + "grad_norm": 2.5219147205352783, + "learning_rate": 1e-06, + "loss": 1.0216, + "mean_token_accuracy": 0.6897251009941101, + "num_tokens": 171580123.0, + "step": 6717 + }, + { + "epoch": 0.7377553261585768, + "grad_norm": 2.3169033527374268, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7109969854354858, + "num_tokens": 171603335.0, + "step": 6718 + }, + { + "epoch": 0.7378651438611904, + "grad_norm": 2.337008237838745, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7081965208053589, + "num_tokens": 171626589.0, + "step": 6719 + }, + { + "epoch": 0.737974961563804, + "grad_norm": 2.329622745513916, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7262066602706909, + "num_tokens": 171649227.0, + "step": 6720 + }, + { + "epoch": 0.7380847792664178, + "grad_norm": 2.2916202545166016, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7102783918380737, + "num_tokens": 171672498.0, + "step": 6721 + }, + { + "epoch": 0.7381945969690314, + "grad_norm": 2.23626971244812, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7004439830780029, + "num_tokens": 171695820.0, + "step": 6722 + }, + { + "epoch": 0.7383044146716451, + "grad_norm": 2.280489444732666, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7107747197151184, + "num_tokens": 171720085.0, + "step": 6723 + }, + { + "epoch": 0.7384142323742587, + "grad_norm": 2.330509662628174, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7124476432800293, + "num_tokens": 171743425.0, + "step": 6724 + }, + { + "epoch": 0.7385240500768724, + "grad_norm": 2.427880048751831, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7396895885467529, + "num_tokens": 171761978.0, + "step": 6725 + }, + { + "epoch": 0.738633867779486, + "grad_norm": 2.1951756477355957, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.6984955072402954, + "num_tokens": 171787721.0, + "step": 6726 + }, + { + "epoch": 0.7387436854820997, + "grad_norm": 2.277095317840576, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7232967019081116, + "num_tokens": 171810401.0, + "step": 6727 + }, + { + "epoch": 0.7388535031847133, + "grad_norm": 1.8811063766479492, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7079253196716309, + "num_tokens": 171842507.0, + "step": 6728 + }, + { + "epoch": 0.7389633208873271, + "grad_norm": 2.5979292392730713, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7200839519500732, + "num_tokens": 171861877.0, + "step": 6729 + }, + { + "epoch": 0.7390731385899407, + "grad_norm": 2.1409313678741455, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7036144733428955, + "num_tokens": 171888542.0, + "step": 6730 + }, + { + "epoch": 0.7391829562925544, + "grad_norm": 2.127840280532837, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7125147581100464, + "num_tokens": 171914126.0, + "step": 6731 + }, + { + "epoch": 0.739292773995168, + "grad_norm": 2.4876761436462402, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7050305008888245, + "num_tokens": 171934149.0, + "step": 6732 + }, + { + "epoch": 0.7394025916977817, + "grad_norm": 2.3276209831237793, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7236109375953674, + "num_tokens": 171956454.0, + "step": 6733 + }, + { + "epoch": 0.7395124094003953, + "grad_norm": 2.3539884090423584, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.6936438679695129, + "num_tokens": 171980639.0, + "step": 6734 + }, + { + "epoch": 0.739622227103009, + "grad_norm": 2.2162458896636963, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.730905294418335, + "num_tokens": 172006700.0, + "step": 6735 + }, + { + "epoch": 0.7397320448056227, + "grad_norm": 2.266536235809326, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.693406879901886, + "num_tokens": 172035130.0, + "step": 6736 + }, + { + "epoch": 0.7398418625082364, + "grad_norm": 2.7458553314208984, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7079538106918335, + "num_tokens": 172052744.0, + "step": 6737 + }, + { + "epoch": 0.73995168021085, + "grad_norm": 2.328803300857544, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7090964913368225, + "num_tokens": 172075071.0, + "step": 6738 + }, + { + "epoch": 0.7400614979134637, + "grad_norm": 2.214672327041626, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.713370144367218, + "num_tokens": 172101170.0, + "step": 6739 + }, + { + "epoch": 0.7401713156160773, + "grad_norm": 2.200488328933716, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7215452790260315, + "num_tokens": 172126955.0, + "step": 6740 + }, + { + "epoch": 0.7402811333186909, + "grad_norm": 2.3480172157287598, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7352204322814941, + "num_tokens": 172148824.0, + "step": 6741 + }, + { + "epoch": 0.7403909510213046, + "grad_norm": 2.4710745811462402, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.7528929710388184, + "num_tokens": 172168837.0, + "step": 6742 + }, + { + "epoch": 0.7405007687239183, + "grad_norm": 1.9845482110977173, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7122571468353271, + "num_tokens": 172198439.0, + "step": 6743 + }, + { + "epoch": 0.740610586426532, + "grad_norm": 1.985209345817566, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.698294997215271, + "num_tokens": 172229412.0, + "step": 6744 + }, + { + "epoch": 0.7407204041291456, + "grad_norm": 2.5264501571655273, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7356138229370117, + "num_tokens": 172249390.0, + "step": 6745 + }, + { + "epoch": 0.7408302218317593, + "grad_norm": 2.2967731952667236, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7125900983810425, + "num_tokens": 172274423.0, + "step": 6746 + }, + { + "epoch": 0.7409400395343729, + "grad_norm": 2.2511844635009766, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7084593176841736, + "num_tokens": 172299919.0, + "step": 6747 + }, + { + "epoch": 0.7410498572369866, + "grad_norm": 2.0122456550598145, + "learning_rate": 1e-06, + "loss": 1.0464, + "mean_token_accuracy": 0.683003306388855, + "num_tokens": 172331126.0, + "step": 6748 + }, + { + "epoch": 0.7411596749396002, + "grad_norm": 2.044161081314087, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.6900110244750977, + "num_tokens": 172360439.0, + "step": 6749 + }, + { + "epoch": 0.741269492642214, + "grad_norm": 2.437001943588257, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.7057556509971619, + "num_tokens": 172382932.0, + "step": 6750 + }, + { + "epoch": 0.7413793103448276, + "grad_norm": 2.2735214233398438, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.6997889876365662, + "num_tokens": 172407140.0, + "step": 6751 + }, + { + "epoch": 0.7414891280474413, + "grad_norm": 2.3368475437164307, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7368878126144409, + "num_tokens": 172428406.0, + "step": 6752 + }, + { + "epoch": 0.7415989457500549, + "grad_norm": 2.410104513168335, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7062445878982544, + "num_tokens": 172449978.0, + "step": 6753 + }, + { + "epoch": 0.7417087634526686, + "grad_norm": 2.5042641162872314, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.688991367816925, + "num_tokens": 172473506.0, + "step": 6754 + }, + { + "epoch": 0.7418185811552822, + "grad_norm": 2.139932632446289, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.7023349404335022, + "num_tokens": 172499223.0, + "step": 6755 + }, + { + "epoch": 0.7419283988578959, + "grad_norm": 2.0978522300720215, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.715178370475769, + "num_tokens": 172525475.0, + "step": 6756 + }, + { + "epoch": 0.7420382165605095, + "grad_norm": 2.1298136711120605, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6980598568916321, + "num_tokens": 172553249.0, + "step": 6757 + }, + { + "epoch": 0.7421480342631233, + "grad_norm": 2.2240538597106934, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.6886618733406067, + "num_tokens": 172579377.0, + "step": 6758 + }, + { + "epoch": 0.7422578519657369, + "grad_norm": 2.1582043170928955, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6900988221168518, + "num_tokens": 172605269.0, + "step": 6759 + }, + { + "epoch": 0.7423676696683505, + "grad_norm": 2.1625096797943115, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7052538394927979, + "num_tokens": 172632416.0, + "step": 6760 + }, + { + "epoch": 0.7424774873709642, + "grad_norm": 2.4987847805023193, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.7021341919898987, + "num_tokens": 172653690.0, + "step": 6761 + }, + { + "epoch": 0.7425873050735778, + "grad_norm": 2.006700038909912, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6977676153182983, + "num_tokens": 172682389.0, + "step": 6762 + }, + { + "epoch": 0.7426971227761915, + "grad_norm": 2.060584545135498, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.702834963798523, + "num_tokens": 172708935.0, + "step": 6763 + }, + { + "epoch": 0.7428069404788051, + "grad_norm": 2.230093479156494, + "learning_rate": 1e-06, + "loss": 1.0209, + "mean_token_accuracy": 0.685443639755249, + "num_tokens": 172734735.0, + "step": 6764 + }, + { + "epoch": 0.7429167581814189, + "grad_norm": 2.124941349029541, + "learning_rate": 1e-06, + "loss": 1.0729, + "mean_token_accuracy": 0.6771057844161987, + "num_tokens": 172763035.0, + "step": 6765 + }, + { + "epoch": 0.7430265758840325, + "grad_norm": 2.145238161087036, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7047790288925171, + "num_tokens": 172789175.0, + "step": 6766 + }, + { + "epoch": 0.7431363935866462, + "grad_norm": 2.0702171325683594, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.692233681678772, + "num_tokens": 172817074.0, + "step": 6767 + }, + { + "epoch": 0.7432462112892598, + "grad_norm": 2.089567184448242, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.6960319876670837, + "num_tokens": 172844712.0, + "step": 6768 + }, + { + "epoch": 0.7433560289918735, + "grad_norm": 2.3001692295074463, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7236831188201904, + "num_tokens": 172867231.0, + "step": 6769 + }, + { + "epoch": 0.7434658466944871, + "grad_norm": 2.280555486679077, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6954119205474854, + "num_tokens": 172891896.0, + "step": 6770 + }, + { + "epoch": 0.7435756643971008, + "grad_norm": 2.231558084487915, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7251548171043396, + "num_tokens": 172915253.0, + "step": 6771 + }, + { + "epoch": 0.7436854820997145, + "grad_norm": 2.066873788833618, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7297861576080322, + "num_tokens": 172941297.0, + "step": 6772 + }, + { + "epoch": 0.7437952998023282, + "grad_norm": 2.233159065246582, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7245320081710815, + "num_tokens": 172964740.0, + "step": 6773 + }, + { + "epoch": 0.7439051175049418, + "grad_norm": 1.9955015182495117, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.7095742225646973, + "num_tokens": 172994009.0, + "step": 6774 + }, + { + "epoch": 0.7440149352075555, + "grad_norm": 2.174555540084839, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7093393206596375, + "num_tokens": 173018739.0, + "step": 6775 + }, + { + "epoch": 0.7441247529101691, + "grad_norm": 2.173448085784912, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.7001845240592957, + "num_tokens": 173044065.0, + "step": 6776 + }, + { + "epoch": 0.7442345706127828, + "grad_norm": 2.270159959793091, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7123454213142395, + "num_tokens": 173067324.0, + "step": 6777 + }, + { + "epoch": 0.7443443883153964, + "grad_norm": 2.524458646774292, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7090506553649902, + "num_tokens": 173087514.0, + "step": 6778 + }, + { + "epoch": 0.7444542060180102, + "grad_norm": 2.393731117248535, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7159019708633423, + "num_tokens": 173108126.0, + "step": 6779 + }, + { + "epoch": 0.7445640237206238, + "grad_norm": 2.214059591293335, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7338539361953735, + "num_tokens": 173132559.0, + "step": 6780 + }, + { + "epoch": 0.7446738414232374, + "grad_norm": 2.1716268062591553, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.6985359191894531, + "num_tokens": 173158736.0, + "step": 6781 + }, + { + "epoch": 0.7447836591258511, + "grad_norm": 2.315119743347168, + "learning_rate": 1e-06, + "loss": 1.0324, + "mean_token_accuracy": 0.6884890794754028, + "num_tokens": 173181030.0, + "step": 6782 + }, + { + "epoch": 0.7448934768284647, + "grad_norm": 2.5790200233459473, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7298970222473145, + "num_tokens": 173199540.0, + "step": 6783 + }, + { + "epoch": 0.7450032945310784, + "grad_norm": 2.2361068725585938, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.693823516368866, + "num_tokens": 173224909.0, + "step": 6784 + }, + { + "epoch": 0.745113112233692, + "grad_norm": 2.1372618675231934, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.6969991326332092, + "num_tokens": 173251833.0, + "step": 6785 + }, + { + "epoch": 0.7452229299363057, + "grad_norm": 2.1460177898406982, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6903735399246216, + "num_tokens": 173278956.0, + "step": 6786 + }, + { + "epoch": 0.7453327476389194, + "grad_norm": 2.5432446002960205, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7006607055664062, + "num_tokens": 173298626.0, + "step": 6787 + }, + { + "epoch": 0.7454425653415331, + "grad_norm": 1.9383574724197388, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.693716287612915, + "num_tokens": 173330140.0, + "step": 6788 + }, + { + "epoch": 0.7455523830441467, + "grad_norm": 2.223484516143799, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7027521133422852, + "num_tokens": 173354220.0, + "step": 6789 + }, + { + "epoch": 0.7456622007467604, + "grad_norm": 2.5219688415527344, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7211278080940247, + "num_tokens": 173373250.0, + "step": 6790 + }, + { + "epoch": 0.745772018449374, + "grad_norm": 2.55551815032959, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7272912263870239, + "num_tokens": 173392322.0, + "step": 6791 + }, + { + "epoch": 0.7458818361519877, + "grad_norm": 2.134260892868042, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7079506516456604, + "num_tokens": 173419937.0, + "step": 6792 + }, + { + "epoch": 0.7459916538546013, + "grad_norm": 1.8866596221923828, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6957660913467407, + "num_tokens": 173454515.0, + "step": 6793 + }, + { + "epoch": 0.7461014715572151, + "grad_norm": 2.1843349933624268, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6912831664085388, + "num_tokens": 173481942.0, + "step": 6794 + }, + { + "epoch": 0.7462112892598287, + "grad_norm": 2.4562458992004395, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.6936255693435669, + "num_tokens": 173505946.0, + "step": 6795 + }, + { + "epoch": 0.7463211069624424, + "grad_norm": 2.435795307159424, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7186152935028076, + "num_tokens": 173528671.0, + "step": 6796 + }, + { + "epoch": 0.746430924665056, + "grad_norm": 2.06180477142334, + "learning_rate": 1e-06, + "loss": 0.7715, + "mean_token_accuracy": 0.7497315406799316, + "num_tokens": 173553490.0, + "step": 6797 + }, + { + "epoch": 0.7465407423676697, + "grad_norm": 2.229405641555786, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7334839105606079, + "num_tokens": 173576835.0, + "step": 6798 + }, + { + "epoch": 0.7466505600702833, + "grad_norm": 2.057361602783203, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7198872566223145, + "num_tokens": 173604422.0, + "step": 6799 + }, + { + "epoch": 0.7467603777728969, + "grad_norm": 2.2111973762512207, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.734646201133728, + "num_tokens": 173628051.0, + "step": 6800 + }, + { + "epoch": 0.7468701954755107, + "grad_norm": 2.320708751678467, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.7027573585510254, + "num_tokens": 173650980.0, + "step": 6801 + }, + { + "epoch": 0.7469800131781243, + "grad_norm": 2.442660331726074, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7040784358978271, + "num_tokens": 173674488.0, + "step": 6802 + }, + { + "epoch": 0.747089830880738, + "grad_norm": 2.1515042781829834, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7272123098373413, + "num_tokens": 173698630.0, + "step": 6803 + }, + { + "epoch": 0.7471996485833516, + "grad_norm": 2.547614812850952, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7180173397064209, + "num_tokens": 173719115.0, + "step": 6804 + }, + { + "epoch": 0.7473094662859653, + "grad_norm": 2.2551119327545166, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.6973447203636169, + "num_tokens": 173743115.0, + "step": 6805 + }, + { + "epoch": 0.7474192839885789, + "grad_norm": 2.4875521659851074, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7118514180183411, + "num_tokens": 173765745.0, + "step": 6806 + }, + { + "epoch": 0.7475291016911926, + "grad_norm": 2.0906717777252197, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.700777530670166, + "num_tokens": 173794067.0, + "step": 6807 + }, + { + "epoch": 0.7476389193938063, + "grad_norm": 2.220059633255005, + "learning_rate": 1e-06, + "loss": 1.0418, + "mean_token_accuracy": 0.680635392665863, + "num_tokens": 173820217.0, + "step": 6808 + }, + { + "epoch": 0.74774873709642, + "grad_norm": 2.206928253173828, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7264350652694702, + "num_tokens": 173845012.0, + "step": 6809 + }, + { + "epoch": 0.7478585547990336, + "grad_norm": 1.9977823495864868, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7105115056037903, + "num_tokens": 173874534.0, + "step": 6810 + }, + { + "epoch": 0.7479683725016473, + "grad_norm": 2.524831771850586, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7283987402915955, + "num_tokens": 173893523.0, + "step": 6811 + }, + { + "epoch": 0.7480781902042609, + "grad_norm": 2.2976040840148926, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7231107950210571, + "num_tokens": 173915616.0, + "step": 6812 + }, + { + "epoch": 0.7481880079068746, + "grad_norm": 2.233572483062744, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7061572074890137, + "num_tokens": 173942787.0, + "step": 6813 + }, + { + "epoch": 0.7482978256094882, + "grad_norm": 2.1083390712738037, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7181532382965088, + "num_tokens": 173969048.0, + "step": 6814 + }, + { + "epoch": 0.7484076433121019, + "grad_norm": 2.0888402462005615, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7098750472068787, + "num_tokens": 173996969.0, + "step": 6815 + }, + { + "epoch": 0.7485174610147156, + "grad_norm": 2.424844980239868, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7028322219848633, + "num_tokens": 174018872.0, + "step": 6816 + }, + { + "epoch": 0.7486272787173293, + "grad_norm": 2.4043407440185547, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6971089839935303, + "num_tokens": 174040756.0, + "step": 6817 + }, + { + "epoch": 0.7487370964199429, + "grad_norm": 1.986781120300293, + "learning_rate": 1e-06, + "loss": 1.0348, + "mean_token_accuracy": 0.6941526532173157, + "num_tokens": 174069245.0, + "step": 6818 + }, + { + "epoch": 0.7488469141225566, + "grad_norm": 2.2305867671966553, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.721161961555481, + "num_tokens": 174091538.0, + "step": 6819 + }, + { + "epoch": 0.7489567318251702, + "grad_norm": 2.36541748046875, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7074511051177979, + "num_tokens": 174113571.0, + "step": 6820 + }, + { + "epoch": 0.7490665495277838, + "grad_norm": 2.494760036468506, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7039153575897217, + "num_tokens": 174134719.0, + "step": 6821 + }, + { + "epoch": 0.7491763672303975, + "grad_norm": 2.1760048866271973, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6904920339584351, + "num_tokens": 174164670.0, + "step": 6822 + }, + { + "epoch": 0.7492861849330112, + "grad_norm": 2.383167028427124, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6953811645507812, + "num_tokens": 174188306.0, + "step": 6823 + }, + { + "epoch": 0.7493960026356249, + "grad_norm": 2.0715668201446533, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7173354625701904, + "num_tokens": 174215532.0, + "step": 6824 + }, + { + "epoch": 0.7495058203382385, + "grad_norm": 2.3730669021606445, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7190142869949341, + "num_tokens": 174239140.0, + "step": 6825 + }, + { + "epoch": 0.7496156380408522, + "grad_norm": 2.190340757369995, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7056897282600403, + "num_tokens": 174266156.0, + "step": 6826 + }, + { + "epoch": 0.7497254557434658, + "grad_norm": 2.1832919120788574, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7101649045944214, + "num_tokens": 174292554.0, + "step": 6827 + }, + { + "epoch": 0.7498352734460795, + "grad_norm": 2.1550393104553223, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7017871141433716, + "num_tokens": 174318993.0, + "step": 6828 + }, + { + "epoch": 0.7499450911486931, + "grad_norm": 2.024358034133911, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7183469533920288, + "num_tokens": 174346588.0, + "step": 6829 + }, + { + "epoch": 0.7500549088513069, + "grad_norm": 2.0206971168518066, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7256730198860168, + "num_tokens": 174373523.0, + "step": 6830 + }, + { + "epoch": 0.7501647265539205, + "grad_norm": 2.2056875228881836, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.6867259740829468, + "num_tokens": 174399576.0, + "step": 6831 + }, + { + "epoch": 0.7502745442565342, + "grad_norm": 2.011312246322632, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7097535133361816, + "num_tokens": 174427480.0, + "step": 6832 + }, + { + "epoch": 0.7503843619591478, + "grad_norm": 2.303715705871582, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7092556953430176, + "num_tokens": 174450372.0, + "step": 6833 + }, + { + "epoch": 0.7504941796617615, + "grad_norm": 2.209641695022583, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7353688478469849, + "num_tokens": 174473977.0, + "step": 6834 + }, + { + "epoch": 0.7506039973643751, + "grad_norm": 2.3302128314971924, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.707423746585846, + "num_tokens": 174496054.0, + "step": 6835 + }, + { + "epoch": 0.7507138150669888, + "grad_norm": 2.158050537109375, + "learning_rate": 1e-06, + "loss": 1.0245, + "mean_token_accuracy": 0.6931807994842529, + "num_tokens": 174524968.0, + "step": 6836 + }, + { + "epoch": 0.7508236327696025, + "grad_norm": 2.318120241165161, + "learning_rate": 1e-06, + "loss": 1.0275, + "mean_token_accuracy": 0.6891181468963623, + "num_tokens": 174549870.0, + "step": 6837 + }, + { + "epoch": 0.7509334504722162, + "grad_norm": 2.0925374031066895, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.720117449760437, + "num_tokens": 174578181.0, + "step": 6838 + }, + { + "epoch": 0.7510432681748298, + "grad_norm": 1.9849294424057007, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.709782600402832, + "num_tokens": 174610632.0, + "step": 6839 + }, + { + "epoch": 0.7511530858774434, + "grad_norm": 2.122938394546509, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.710637629032135, + "num_tokens": 174635027.0, + "step": 6840 + }, + { + "epoch": 0.7512629035800571, + "grad_norm": 2.2692205905914307, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7140929102897644, + "num_tokens": 174657549.0, + "step": 6841 + }, + { + "epoch": 0.7513727212826707, + "grad_norm": 2.2781379222869873, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7312139272689819, + "num_tokens": 174680408.0, + "step": 6842 + }, + { + "epoch": 0.7514825389852844, + "grad_norm": 1.8539942502975464, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7060137987136841, + "num_tokens": 174714390.0, + "step": 6843 + }, + { + "epoch": 0.7515923566878981, + "grad_norm": 2.27022647857666, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.7010043859481812, + "num_tokens": 174738868.0, + "step": 6844 + }, + { + "epoch": 0.7517021743905118, + "grad_norm": 2.1542892456054688, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.7084803581237793, + "num_tokens": 174765343.0, + "step": 6845 + }, + { + "epoch": 0.7518119920931254, + "grad_norm": 1.7729767560958862, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.708740234375, + "num_tokens": 174801255.0, + "step": 6846 + }, + { + "epoch": 0.7519218097957391, + "grad_norm": 2.0503182411193848, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6955575942993164, + "num_tokens": 174830279.0, + "step": 6847 + }, + { + "epoch": 0.7520316274983527, + "grad_norm": 2.0757546424865723, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7022557258605957, + "num_tokens": 174858047.0, + "step": 6848 + }, + { + "epoch": 0.7521414452009664, + "grad_norm": 1.9105544090270996, + "learning_rate": 1e-06, + "loss": 1.0627, + "mean_token_accuracy": 0.6830999255180359, + "num_tokens": 174890055.0, + "step": 6849 + }, + { + "epoch": 0.75225126290358, + "grad_norm": 2.0172576904296875, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7020860910415649, + "num_tokens": 174918693.0, + "step": 6850 + }, + { + "epoch": 0.7523610806061937, + "grad_norm": 2.0298643112182617, + "learning_rate": 1e-06, + "loss": 1.0358, + "mean_token_accuracy": 0.6856244802474976, + "num_tokens": 174948066.0, + "step": 6851 + }, + { + "epoch": 0.7524708983088074, + "grad_norm": 2.2416880130767822, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6902139186859131, + "num_tokens": 174974224.0, + "step": 6852 + }, + { + "epoch": 0.7525807160114211, + "grad_norm": 2.355985641479492, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.7102289199829102, + "num_tokens": 174999093.0, + "step": 6853 + }, + { + "epoch": 0.7526905337140347, + "grad_norm": 2.1472761631011963, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.7029351592063904, + "num_tokens": 175024592.0, + "step": 6854 + }, + { + "epoch": 0.7528003514166484, + "grad_norm": 2.0757551193237305, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7242462635040283, + "num_tokens": 175051312.0, + "step": 6855 + }, + { + "epoch": 0.752910169119262, + "grad_norm": 2.126052141189575, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7094229459762573, + "num_tokens": 175080422.0, + "step": 6856 + }, + { + "epoch": 0.7530199868218757, + "grad_norm": 2.1090142726898193, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.6917780637741089, + "num_tokens": 175105911.0, + "step": 6857 + }, + { + "epoch": 0.7531298045244893, + "grad_norm": 2.1601130962371826, + "learning_rate": 1e-06, + "loss": 1.0526, + "mean_token_accuracy": 0.6781325936317444, + "num_tokens": 175133054.0, + "step": 6858 + }, + { + "epoch": 0.7532396222271031, + "grad_norm": 1.9172495603561401, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7074973583221436, + "num_tokens": 175163188.0, + "step": 6859 + }, + { + "epoch": 0.7533494399297167, + "grad_norm": 2.276238203048706, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7111949920654297, + "num_tokens": 175186062.0, + "step": 6860 + }, + { + "epoch": 0.7534592576323303, + "grad_norm": 2.31801700592041, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.6926876306533813, + "num_tokens": 175211302.0, + "step": 6861 + }, + { + "epoch": 0.753569075334944, + "grad_norm": 2.09592866897583, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7113875150680542, + "num_tokens": 175239225.0, + "step": 6862 + }, + { + "epoch": 0.7536788930375576, + "grad_norm": 1.9193865060806274, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7221692800521851, + "num_tokens": 175270568.0, + "step": 6863 + }, + { + "epoch": 0.7537887107401713, + "grad_norm": 2.2057535648345947, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6960971355438232, + "num_tokens": 175296411.0, + "step": 6864 + }, + { + "epoch": 0.7538985284427849, + "grad_norm": 2.07346248626709, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.7094898819923401, + "num_tokens": 175323285.0, + "step": 6865 + }, + { + "epoch": 0.7540083461453987, + "grad_norm": 2.1190898418426514, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7190033197402954, + "num_tokens": 175350570.0, + "step": 6866 + }, + { + "epoch": 0.7541181638480123, + "grad_norm": 2.5271291732788086, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7028262615203857, + "num_tokens": 175370877.0, + "step": 6867 + }, + { + "epoch": 0.754227981550626, + "grad_norm": 2.144775867462158, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7266721725463867, + "num_tokens": 175395464.0, + "step": 6868 + }, + { + "epoch": 0.7543377992532396, + "grad_norm": 2.0399019718170166, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7040877938270569, + "num_tokens": 175425265.0, + "step": 6869 + }, + { + "epoch": 0.7544476169558533, + "grad_norm": 1.971267819404602, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7055193781852722, + "num_tokens": 175454341.0, + "step": 6870 + }, + { + "epoch": 0.7545574346584669, + "grad_norm": 2.4413115978240967, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7150061130523682, + "num_tokens": 175474642.0, + "step": 6871 + }, + { + "epoch": 0.7546672523610806, + "grad_norm": 2.2992067337036133, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7115236520767212, + "num_tokens": 175500614.0, + "step": 6872 + }, + { + "epoch": 0.7547770700636943, + "grad_norm": 2.148226499557495, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7070058584213257, + "num_tokens": 175526086.0, + "step": 6873 + }, + { + "epoch": 0.754886887766308, + "grad_norm": 1.9906973838806152, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7149845361709595, + "num_tokens": 175556149.0, + "step": 6874 + }, + { + "epoch": 0.7549967054689216, + "grad_norm": 2.251337766647339, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.7042770385742188, + "num_tokens": 175579705.0, + "step": 6875 + }, + { + "epoch": 0.7551065231715353, + "grad_norm": 2.310974359512329, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.6987206935882568, + "num_tokens": 175605387.0, + "step": 6876 + }, + { + "epoch": 0.7552163408741489, + "grad_norm": 2.4333627223968506, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7012673616409302, + "num_tokens": 175626049.0, + "step": 6877 + }, + { + "epoch": 0.7553261585767626, + "grad_norm": 2.2965004444122314, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6838026642799377, + "num_tokens": 175650681.0, + "step": 6878 + }, + { + "epoch": 0.7554359762793762, + "grad_norm": 1.968061923980713, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7220098972320557, + "num_tokens": 175681775.0, + "step": 6879 + }, + { + "epoch": 0.7555457939819898, + "grad_norm": 2.720945358276367, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7165721654891968, + "num_tokens": 175698306.0, + "step": 6880 + }, + { + "epoch": 0.7556556116846036, + "grad_norm": 2.129420280456543, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7066003084182739, + "num_tokens": 175726542.0, + "step": 6881 + }, + { + "epoch": 0.7557654293872172, + "grad_norm": 2.0299549102783203, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7163861989974976, + "num_tokens": 175753851.0, + "step": 6882 + }, + { + "epoch": 0.7558752470898309, + "grad_norm": 2.2559914588928223, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.7040411233901978, + "num_tokens": 175780142.0, + "step": 6883 + }, + { + "epoch": 0.7559850647924445, + "grad_norm": 2.122675895690918, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7023786306381226, + "num_tokens": 175806868.0, + "step": 6884 + }, + { + "epoch": 0.7560948824950582, + "grad_norm": 2.448897123336792, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7270607948303223, + "num_tokens": 175827458.0, + "step": 6885 + }, + { + "epoch": 0.7562047001976718, + "grad_norm": 1.919321894645691, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6962435841560364, + "num_tokens": 175859946.0, + "step": 6886 + }, + { + "epoch": 0.7563145179002855, + "grad_norm": 2.209657669067383, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7208575010299683, + "num_tokens": 175883473.0, + "step": 6887 + }, + { + "epoch": 0.7564243356028992, + "grad_norm": 2.0147650241851807, + "learning_rate": 1e-06, + "loss": 1.0324, + "mean_token_accuracy": 0.698341965675354, + "num_tokens": 175913270.0, + "step": 6888 + }, + { + "epoch": 0.7565341533055129, + "grad_norm": 2.0417187213897705, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6923528909683228, + "num_tokens": 175941450.0, + "step": 6889 + }, + { + "epoch": 0.7566439710081265, + "grad_norm": 2.322563648223877, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7072786688804626, + "num_tokens": 175965882.0, + "step": 6890 + }, + { + "epoch": 0.7567537887107402, + "grad_norm": 1.9563080072402954, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7302948236465454, + "num_tokens": 175994429.0, + "step": 6891 + }, + { + "epoch": 0.7568636064133538, + "grad_norm": 2.4598634243011475, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7069780826568604, + "num_tokens": 176014252.0, + "step": 6892 + }, + { + "epoch": 0.7569734241159675, + "grad_norm": 2.307518243789673, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6971256732940674, + "num_tokens": 176038458.0, + "step": 6893 + }, + { + "epoch": 0.7570832418185811, + "grad_norm": 2.4982986450195312, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7201646566390991, + "num_tokens": 176057234.0, + "step": 6894 + }, + { + "epoch": 0.7571930595211949, + "grad_norm": 2.1292314529418945, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7224639654159546, + "num_tokens": 176081019.0, + "step": 6895 + }, + { + "epoch": 0.7573028772238085, + "grad_norm": 2.4029643535614014, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7213045358657837, + "num_tokens": 176102218.0, + "step": 6896 + }, + { + "epoch": 0.7574126949264222, + "grad_norm": 2.2034661769866943, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.6989949345588684, + "num_tokens": 176126386.0, + "step": 6897 + }, + { + "epoch": 0.7575225126290358, + "grad_norm": 2.3648393154144287, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7147049903869629, + "num_tokens": 176147440.0, + "step": 6898 + }, + { + "epoch": 0.7576323303316495, + "grad_norm": 1.9400399923324585, + "learning_rate": 1e-06, + "loss": 1.038, + "mean_token_accuracy": 0.686703622341156, + "num_tokens": 176179548.0, + "step": 6899 + }, + { + "epoch": 0.7577421480342631, + "grad_norm": 2.1183576583862305, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.697564959526062, + "num_tokens": 176206934.0, + "step": 6900 + }, + { + "epoch": 0.7578519657368767, + "grad_norm": 1.920608401298523, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7096637487411499, + "num_tokens": 176239938.0, + "step": 6901 + }, + { + "epoch": 0.7579617834394905, + "grad_norm": 2.4874320030212402, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.6984971165657043, + "num_tokens": 176258874.0, + "step": 6902 + }, + { + "epoch": 0.7580716011421041, + "grad_norm": 2.076097249984741, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7126488089561462, + "num_tokens": 176285134.0, + "step": 6903 + }, + { + "epoch": 0.7581814188447178, + "grad_norm": 2.0843470096588135, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.709058940410614, + "num_tokens": 176312252.0, + "step": 6904 + }, + { + "epoch": 0.7582912365473314, + "grad_norm": 1.995195746421814, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7176426649093628, + "num_tokens": 176340829.0, + "step": 6905 + }, + { + "epoch": 0.7584010542499451, + "grad_norm": 2.1242735385894775, + "learning_rate": 1e-06, + "loss": 1.0955, + "mean_token_accuracy": 0.6714431047439575, + "num_tokens": 176367599.0, + "step": 6906 + }, + { + "epoch": 0.7585108719525587, + "grad_norm": 2.4213268756866455, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7358822226524353, + "num_tokens": 176388835.0, + "step": 6907 + }, + { + "epoch": 0.7586206896551724, + "grad_norm": 2.1299238204956055, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6939072608947754, + "num_tokens": 176416551.0, + "step": 6908 + }, + { + "epoch": 0.758730507357786, + "grad_norm": 2.197664737701416, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7116482853889465, + "num_tokens": 176441163.0, + "step": 6909 + }, + { + "epoch": 0.7588403250603998, + "grad_norm": 2.37929368019104, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.726209282875061, + "num_tokens": 176461971.0, + "step": 6910 + }, + { + "epoch": 0.7589501427630134, + "grad_norm": 2.003002643585205, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7392255067825317, + "num_tokens": 176488548.0, + "step": 6911 + }, + { + "epoch": 0.7590599604656271, + "grad_norm": 1.9850353002548218, + "learning_rate": 1e-06, + "loss": 1.1041, + "mean_token_accuracy": 0.6680310964584351, + "num_tokens": 176522112.0, + "step": 6912 + }, + { + "epoch": 0.7591697781682407, + "grad_norm": 2.5890681743621826, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7103937864303589, + "num_tokens": 176542288.0, + "step": 6913 + }, + { + "epoch": 0.7592795958708544, + "grad_norm": 2.0963220596313477, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7084461450576782, + "num_tokens": 176569185.0, + "step": 6914 + }, + { + "epoch": 0.759389413573468, + "grad_norm": 2.186135768890381, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7059967517852783, + "num_tokens": 176594770.0, + "step": 6915 + }, + { + "epoch": 0.7594992312760817, + "grad_norm": 2.3154408931732178, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6967549324035645, + "num_tokens": 176618010.0, + "step": 6916 + }, + { + "epoch": 0.7596090489786954, + "grad_norm": 1.9595153331756592, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7011076211929321, + "num_tokens": 176649007.0, + "step": 6917 + }, + { + "epoch": 0.7597188666813091, + "grad_norm": 2.2048285007476807, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.6897942423820496, + "num_tokens": 176675254.0, + "step": 6918 + }, + { + "epoch": 0.7598286843839227, + "grad_norm": 2.1912899017333984, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7308485507965088, + "num_tokens": 176700460.0, + "step": 6919 + }, + { + "epoch": 0.7599385020865363, + "grad_norm": 2.0846426486968994, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7022964954376221, + "num_tokens": 176727876.0, + "step": 6920 + }, + { + "epoch": 0.76004831978915, + "grad_norm": 1.965780258178711, + "learning_rate": 1e-06, + "loss": 1.0986, + "mean_token_accuracy": 0.6721193194389343, + "num_tokens": 176761165.0, + "step": 6921 + }, + { + "epoch": 0.7601581374917636, + "grad_norm": 2.254305124282837, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7321294546127319, + "num_tokens": 176785918.0, + "step": 6922 + }, + { + "epoch": 0.7602679551943773, + "grad_norm": 2.2499783039093018, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7372969388961792, + "num_tokens": 176809084.0, + "step": 6923 + }, + { + "epoch": 0.760377772896991, + "grad_norm": 2.1990978717803955, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.6943762302398682, + "num_tokens": 176836409.0, + "step": 6924 + }, + { + "epoch": 0.7604875905996047, + "grad_norm": 2.6624858379364014, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7190444469451904, + "num_tokens": 176856177.0, + "step": 6925 + }, + { + "epoch": 0.7605974083022183, + "grad_norm": 2.446030855178833, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7241483330726624, + "num_tokens": 176876471.0, + "step": 6926 + }, + { + "epoch": 0.760707226004832, + "grad_norm": 2.1474874019622803, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7057443261146545, + "num_tokens": 176901917.0, + "step": 6927 + }, + { + "epoch": 0.7608170437074456, + "grad_norm": 1.8809669017791748, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6974261999130249, + "num_tokens": 176935894.0, + "step": 6928 + }, + { + "epoch": 0.7609268614100593, + "grad_norm": 2.622701644897461, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7240924835205078, + "num_tokens": 176953634.0, + "step": 6929 + }, + { + "epoch": 0.7610366791126729, + "grad_norm": 2.5418691635131836, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6913155913352966, + "num_tokens": 176975403.0, + "step": 6930 + }, + { + "epoch": 0.7611464968152867, + "grad_norm": 2.291487216949463, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7088791728019714, + "num_tokens": 176999572.0, + "step": 6931 + }, + { + "epoch": 0.7612563145179003, + "grad_norm": 2.183932304382324, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6947077512741089, + "num_tokens": 177025878.0, + "step": 6932 + }, + { + "epoch": 0.761366132220514, + "grad_norm": 2.131588935852051, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7101829648017883, + "num_tokens": 177050884.0, + "step": 6933 + }, + { + "epoch": 0.7614759499231276, + "grad_norm": 2.1662561893463135, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.722295343875885, + "num_tokens": 177074991.0, + "step": 6934 + }, + { + "epoch": 0.7615857676257413, + "grad_norm": 2.1808574199676514, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7080734968185425, + "num_tokens": 177100184.0, + "step": 6935 + }, + { + "epoch": 0.7616955853283549, + "grad_norm": 1.8100166320800781, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.6959133744239807, + "num_tokens": 177135997.0, + "step": 6936 + }, + { + "epoch": 0.7618054030309686, + "grad_norm": 2.523324966430664, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.714877188205719, + "num_tokens": 177156502.0, + "step": 6937 + }, + { + "epoch": 0.7619152207335822, + "grad_norm": 2.3157570362091064, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7320740222930908, + "num_tokens": 177178111.0, + "step": 6938 + }, + { + "epoch": 0.762025038436196, + "grad_norm": 2.050579309463501, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.727567195892334, + "num_tokens": 177206016.0, + "step": 6939 + }, + { + "epoch": 0.7621348561388096, + "grad_norm": 2.2055747509002686, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7063699960708618, + "num_tokens": 177231995.0, + "step": 6940 + }, + { + "epoch": 0.7622446738414232, + "grad_norm": 2.4464893341064453, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7265172600746155, + "num_tokens": 177253765.0, + "step": 6941 + }, + { + "epoch": 0.7623544915440369, + "grad_norm": 2.079359531402588, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7360556125640869, + "num_tokens": 177279760.0, + "step": 6942 + }, + { + "epoch": 0.7624643092466505, + "grad_norm": 2.1059467792510986, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7425675988197327, + "num_tokens": 177305321.0, + "step": 6943 + }, + { + "epoch": 0.7625741269492642, + "grad_norm": 2.289937734603882, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6930632591247559, + "num_tokens": 177329595.0, + "step": 6944 + }, + { + "epoch": 0.7626839446518778, + "grad_norm": 2.0972187519073486, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7122679352760315, + "num_tokens": 177355894.0, + "step": 6945 + }, + { + "epoch": 0.7627937623544916, + "grad_norm": 2.092090606689453, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7124810218811035, + "num_tokens": 177384168.0, + "step": 6946 + }, + { + "epoch": 0.7629035800571052, + "grad_norm": 2.515615701675415, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7147706747055054, + "num_tokens": 177403184.0, + "step": 6947 + }, + { + "epoch": 0.7630133977597189, + "grad_norm": 2.1471097469329834, + "learning_rate": 1e-06, + "loss": 1.0934, + "mean_token_accuracy": 0.6774094104766846, + "num_tokens": 177431771.0, + "step": 6948 + }, + { + "epoch": 0.7631232154623325, + "grad_norm": 2.216917037963867, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7251971960067749, + "num_tokens": 177455011.0, + "step": 6949 + }, + { + "epoch": 0.7632330331649462, + "grad_norm": 2.3139712810516357, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7123545408248901, + "num_tokens": 177477655.0, + "step": 6950 + }, + { + "epoch": 0.7633428508675598, + "grad_norm": 2.0177652835845947, + "learning_rate": 1e-06, + "loss": 1.0356, + "mean_token_accuracy": 0.6902837753295898, + "num_tokens": 177509010.0, + "step": 6951 + }, + { + "epoch": 0.7634526685701735, + "grad_norm": 2.095672130584717, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6951184272766113, + "num_tokens": 177537002.0, + "step": 6952 + }, + { + "epoch": 0.7635624862727872, + "grad_norm": 2.389768362045288, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7293250560760498, + "num_tokens": 177558956.0, + "step": 6953 + }, + { + "epoch": 0.7636723039754009, + "grad_norm": 2.213613986968994, + "learning_rate": 1e-06, + "loss": 1.0223, + "mean_token_accuracy": 0.6888138055801392, + "num_tokens": 177584670.0, + "step": 6954 + }, + { + "epoch": 0.7637821216780145, + "grad_norm": 2.037598133087158, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7366253733634949, + "num_tokens": 177609987.0, + "step": 6955 + }, + { + "epoch": 0.7638919393806282, + "grad_norm": 2.0177793502807617, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.6986189484596252, + "num_tokens": 177638066.0, + "step": 6956 + }, + { + "epoch": 0.7640017570832418, + "grad_norm": 2.370666980743408, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7029157280921936, + "num_tokens": 177659220.0, + "step": 6957 + }, + { + "epoch": 0.7641115747858555, + "grad_norm": 2.0813393592834473, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6925073862075806, + "num_tokens": 177686985.0, + "step": 6958 + }, + { + "epoch": 0.7642213924884691, + "grad_norm": 2.157644033432007, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7258002758026123, + "num_tokens": 177710979.0, + "step": 6959 + }, + { + "epoch": 0.7643312101910829, + "grad_norm": 2.3605661392211914, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7179285287857056, + "num_tokens": 177731563.0, + "step": 6960 + }, + { + "epoch": 0.7644410278936965, + "grad_norm": 2.7526559829711914, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7125818729400635, + "num_tokens": 177749587.0, + "step": 6961 + }, + { + "epoch": 0.7645508455963101, + "grad_norm": 2.1145339012145996, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6948084831237793, + "num_tokens": 177775915.0, + "step": 6962 + }, + { + "epoch": 0.7646606632989238, + "grad_norm": 1.8747130632400513, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6994272470474243, + "num_tokens": 177809690.0, + "step": 6963 + }, + { + "epoch": 0.7647704810015374, + "grad_norm": 2.1717584133148193, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.6994625926017761, + "num_tokens": 177835325.0, + "step": 6964 + }, + { + "epoch": 0.7648802987041511, + "grad_norm": 2.113158702850342, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.722205638885498, + "num_tokens": 177861857.0, + "step": 6965 + }, + { + "epoch": 0.7649901164067647, + "grad_norm": 2.125655174255371, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7057962417602539, + "num_tokens": 177891497.0, + "step": 6966 + }, + { + "epoch": 0.7650999341093784, + "grad_norm": 2.422610282897949, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7295687198638916, + "num_tokens": 177912800.0, + "step": 6967 + }, + { + "epoch": 0.7652097518119921, + "grad_norm": 2.197354555130005, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.6972866058349609, + "num_tokens": 177938768.0, + "step": 6968 + }, + { + "epoch": 0.7653195695146058, + "grad_norm": 2.0799431800842285, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7178555130958557, + "num_tokens": 177965391.0, + "step": 6969 + }, + { + "epoch": 0.7654293872172194, + "grad_norm": 1.9366968870162964, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6876079440116882, + "num_tokens": 177995416.0, + "step": 6970 + }, + { + "epoch": 0.7655392049198331, + "grad_norm": 2.257925033569336, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.6987147927284241, + "num_tokens": 178019075.0, + "step": 6971 + }, + { + "epoch": 0.7656490226224467, + "grad_norm": 2.3695425987243652, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.7137681841850281, + "num_tokens": 178040625.0, + "step": 6972 + }, + { + "epoch": 0.7657588403250604, + "grad_norm": 2.2612550258636475, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7283323407173157, + "num_tokens": 178062835.0, + "step": 6973 + }, + { + "epoch": 0.765868658027674, + "grad_norm": 2.298769950866699, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.702278733253479, + "num_tokens": 178085705.0, + "step": 6974 + }, + { + "epoch": 0.7659784757302878, + "grad_norm": 2.000380039215088, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.71064293384552, + "num_tokens": 178116779.0, + "step": 6975 + }, + { + "epoch": 0.7660882934329014, + "grad_norm": 2.7841849327087402, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.6953709721565247, + "num_tokens": 178141223.0, + "step": 6976 + }, + { + "epoch": 0.7661981111355151, + "grad_norm": 2.2445833683013916, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7035909295082092, + "num_tokens": 178164726.0, + "step": 6977 + }, + { + "epoch": 0.7663079288381287, + "grad_norm": 2.460578441619873, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7252414226531982, + "num_tokens": 178183388.0, + "step": 6978 + }, + { + "epoch": 0.7664177465407424, + "grad_norm": 2.431309461593628, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7049560546875, + "num_tokens": 178205127.0, + "step": 6979 + }, + { + "epoch": 0.766527564243356, + "grad_norm": 2.0543582439422607, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.6868945360183716, + "num_tokens": 178234013.0, + "step": 6980 + }, + { + "epoch": 0.7666373819459696, + "grad_norm": 2.1427454948425293, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.714502215385437, + "num_tokens": 178259803.0, + "step": 6981 + }, + { + "epoch": 0.7667471996485834, + "grad_norm": 2.105576276779175, + "learning_rate": 1e-06, + "loss": 1.0574, + "mean_token_accuracy": 0.6768226623535156, + "num_tokens": 178288124.0, + "step": 6982 + }, + { + "epoch": 0.766857017351197, + "grad_norm": 2.6247291564941406, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7149648666381836, + "num_tokens": 178306761.0, + "step": 6983 + }, + { + "epoch": 0.7669668350538107, + "grad_norm": 1.9957951307296753, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7062261700630188, + "num_tokens": 178336852.0, + "step": 6984 + }, + { + "epoch": 0.7670766527564243, + "grad_norm": 2.0869057178497314, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7311569452285767, + "num_tokens": 178362985.0, + "step": 6985 + }, + { + "epoch": 0.767186470459038, + "grad_norm": 2.349130630493164, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7150745391845703, + "num_tokens": 178384411.0, + "step": 6986 + }, + { + "epoch": 0.7672962881616516, + "grad_norm": 2.297151565551758, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.712493360042572, + "num_tokens": 178407349.0, + "step": 6987 + }, + { + "epoch": 0.7674061058642653, + "grad_norm": 1.9236571788787842, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.6946702003479004, + "num_tokens": 178440166.0, + "step": 6988 + }, + { + "epoch": 0.767515923566879, + "grad_norm": 2.208462953567505, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7047176957130432, + "num_tokens": 178464411.0, + "step": 6989 + }, + { + "epoch": 0.7676257412694927, + "grad_norm": 2.07232666015625, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.7022604942321777, + "num_tokens": 178495620.0, + "step": 6990 + }, + { + "epoch": 0.7677355589721063, + "grad_norm": 2.245002269744873, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7009921073913574, + "num_tokens": 178520006.0, + "step": 6991 + }, + { + "epoch": 0.76784537667472, + "grad_norm": 1.9691472053527832, + "learning_rate": 1e-06, + "loss": 1.0549, + "mean_token_accuracy": 0.6834028959274292, + "num_tokens": 178553861.0, + "step": 6992 + }, + { + "epoch": 0.7679551943773336, + "grad_norm": 2.0616514682769775, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7146685123443604, + "num_tokens": 178581850.0, + "step": 6993 + }, + { + "epoch": 0.7680650120799473, + "grad_norm": 2.2396152019500732, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7312990427017212, + "num_tokens": 178604332.0, + "step": 6994 + }, + { + "epoch": 0.7681748297825609, + "grad_norm": 2.109093427658081, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7048647999763489, + "num_tokens": 178631136.0, + "step": 6995 + }, + { + "epoch": 0.7682846474851747, + "grad_norm": 2.0312342643737793, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.6902032494544983, + "num_tokens": 178660901.0, + "step": 6996 + }, + { + "epoch": 0.7683944651877883, + "grad_norm": 2.2670626640319824, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.6967732310295105, + "num_tokens": 178684963.0, + "step": 6997 + }, + { + "epoch": 0.768504282890402, + "grad_norm": 2.217430353164673, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6885107755661011, + "num_tokens": 178710111.0, + "step": 6998 + }, + { + "epoch": 0.7686141005930156, + "grad_norm": 2.099116325378418, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7128146886825562, + "num_tokens": 178738217.0, + "step": 6999 + }, + { + "epoch": 0.7687239182956292, + "grad_norm": 2.6149580478668213, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7269282341003418, + "num_tokens": 178759565.0, + "step": 7000 + }, + { + "epoch": 0.7688337359982429, + "grad_norm": 2.2979073524475098, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.714934229850769, + "num_tokens": 178781019.0, + "step": 7001 + }, + { + "epoch": 0.7689435537008565, + "grad_norm": 2.1702778339385986, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7036519050598145, + "num_tokens": 178807268.0, + "step": 7002 + }, + { + "epoch": 0.7690533714034702, + "grad_norm": 2.0504837036132812, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7045814990997314, + "num_tokens": 178833949.0, + "step": 7003 + }, + { + "epoch": 0.7691631891060839, + "grad_norm": 2.0447516441345215, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7276268601417542, + "num_tokens": 178862409.0, + "step": 7004 + }, + { + "epoch": 0.7692730068086976, + "grad_norm": 2.3359968662261963, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7143563032150269, + "num_tokens": 178883685.0, + "step": 7005 + }, + { + "epoch": 0.7693828245113112, + "grad_norm": 2.4440367221832275, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7161357402801514, + "num_tokens": 178903945.0, + "step": 7006 + }, + { + "epoch": 0.7694926422139249, + "grad_norm": 2.0923304557800293, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7220059633255005, + "num_tokens": 178930825.0, + "step": 7007 + }, + { + "epoch": 0.7696024599165385, + "grad_norm": 2.419100761413574, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.702216625213623, + "num_tokens": 178956022.0, + "step": 7008 + }, + { + "epoch": 0.7697122776191522, + "grad_norm": 2.247955799102783, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7295582294464111, + "num_tokens": 178979474.0, + "step": 7009 + }, + { + "epoch": 0.7698220953217658, + "grad_norm": 1.9615099430084229, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.694129228591919, + "num_tokens": 179010213.0, + "step": 7010 + }, + { + "epoch": 0.7699319130243796, + "grad_norm": 2.1731088161468506, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7046948075294495, + "num_tokens": 179035190.0, + "step": 7011 + }, + { + "epoch": 0.7700417307269932, + "grad_norm": 2.236722230911255, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.7021081447601318, + "num_tokens": 179060588.0, + "step": 7012 + }, + { + "epoch": 0.7701515484296069, + "grad_norm": 2.467560052871704, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.729048490524292, + "num_tokens": 179080119.0, + "step": 7013 + }, + { + "epoch": 0.7702613661322205, + "grad_norm": 2.4990856647491455, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7222852110862732, + "num_tokens": 179099826.0, + "step": 7014 + }, + { + "epoch": 0.7703711838348342, + "grad_norm": 2.0467641353607178, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.694608211517334, + "num_tokens": 179127249.0, + "step": 7015 + }, + { + "epoch": 0.7704810015374478, + "grad_norm": 2.182943820953369, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7139278054237366, + "num_tokens": 179153307.0, + "step": 7016 + }, + { + "epoch": 0.7705908192400615, + "grad_norm": 2.345276117324829, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.709579586982727, + "num_tokens": 179175337.0, + "step": 7017 + }, + { + "epoch": 0.7707006369426752, + "grad_norm": 2.398178815841675, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7116144895553589, + "num_tokens": 179195300.0, + "step": 7018 + }, + { + "epoch": 0.7708104546452889, + "grad_norm": 2.082571268081665, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6914860010147095, + "num_tokens": 179223513.0, + "step": 7019 + }, + { + "epoch": 0.7709202723479025, + "grad_norm": 2.039173126220703, + "learning_rate": 1e-06, + "loss": 1.0406, + "mean_token_accuracy": 0.6843805313110352, + "num_tokens": 179252848.0, + "step": 7020 + }, + { + "epoch": 0.7710300900505161, + "grad_norm": 2.059263229370117, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7189310193061829, + "num_tokens": 179278028.0, + "step": 7021 + }, + { + "epoch": 0.7711399077531298, + "grad_norm": 2.165134906768799, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6922919750213623, + "num_tokens": 179306553.0, + "step": 7022 + }, + { + "epoch": 0.7712497254557434, + "grad_norm": 2.3747189044952393, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.7107089757919312, + "num_tokens": 179328121.0, + "step": 7023 + }, + { + "epoch": 0.7713595431583571, + "grad_norm": 2.264484167098999, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7204223275184631, + "num_tokens": 179352445.0, + "step": 7024 + }, + { + "epoch": 0.7714693608609708, + "grad_norm": 2.121004819869995, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.6968331336975098, + "num_tokens": 179380847.0, + "step": 7025 + }, + { + "epoch": 0.7715791785635845, + "grad_norm": 2.215702533721924, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7022197246551514, + "num_tokens": 179405903.0, + "step": 7026 + }, + { + "epoch": 0.7716889962661981, + "grad_norm": 2.3389317989349365, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7110697031021118, + "num_tokens": 179431795.0, + "step": 7027 + }, + { + "epoch": 0.7717988139688118, + "grad_norm": 2.340564489364624, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7109161019325256, + "num_tokens": 179454680.0, + "step": 7028 + }, + { + "epoch": 0.7719086316714254, + "grad_norm": 2.360863208770752, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7225155830383301, + "num_tokens": 179479292.0, + "step": 7029 + }, + { + "epoch": 0.7720184493740391, + "grad_norm": 2.5537502765655518, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.7101272344589233, + "num_tokens": 179499775.0, + "step": 7030 + }, + { + "epoch": 0.7721282670766527, + "grad_norm": 2.2248997688293457, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7272188067436218, + "num_tokens": 179523937.0, + "step": 7031 + }, + { + "epoch": 0.7722380847792664, + "grad_norm": 2.3409409523010254, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.689119279384613, + "num_tokens": 179548203.0, + "step": 7032 + }, + { + "epoch": 0.7723479024818801, + "grad_norm": 2.436628580093384, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7056973576545715, + "num_tokens": 179570583.0, + "step": 7033 + }, + { + "epoch": 0.7724577201844938, + "grad_norm": 2.167776346206665, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6914832592010498, + "num_tokens": 179597019.0, + "step": 7034 + }, + { + "epoch": 0.7725675378871074, + "grad_norm": 2.4143083095550537, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.728175699710846, + "num_tokens": 179617971.0, + "step": 7035 + }, + { + "epoch": 0.7726773555897211, + "grad_norm": 2.0161662101745605, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7233459949493408, + "num_tokens": 179645446.0, + "step": 7036 + }, + { + "epoch": 0.7727871732923347, + "grad_norm": 2.2428245544433594, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.6965459585189819, + "num_tokens": 179672387.0, + "step": 7037 + }, + { + "epoch": 0.7728969909949484, + "grad_norm": 2.5267257690429688, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7114232182502747, + "num_tokens": 179694061.0, + "step": 7038 + }, + { + "epoch": 0.773006808697562, + "grad_norm": 2.271066904067993, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7050528526306152, + "num_tokens": 179717588.0, + "step": 7039 + }, + { + "epoch": 0.7731166264001758, + "grad_norm": 2.3301239013671875, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6963347792625427, + "num_tokens": 179741570.0, + "step": 7040 + }, + { + "epoch": 0.7732264441027894, + "grad_norm": 2.2113301753997803, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7058677077293396, + "num_tokens": 179767442.0, + "step": 7041 + }, + { + "epoch": 0.773336261805403, + "grad_norm": 2.0135486125946045, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.71373450756073, + "num_tokens": 179795623.0, + "step": 7042 + }, + { + "epoch": 0.7734460795080167, + "grad_norm": 2.205493450164795, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7192891836166382, + "num_tokens": 179818644.0, + "step": 7043 + }, + { + "epoch": 0.7735558972106303, + "grad_norm": 2.169748067855835, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7187315225601196, + "num_tokens": 179845098.0, + "step": 7044 + }, + { + "epoch": 0.773665714913244, + "grad_norm": 2.0508060455322266, + "learning_rate": 1e-06, + "loss": 1.0313, + "mean_token_accuracy": 0.6868185997009277, + "num_tokens": 179875618.0, + "step": 7045 + }, + { + "epoch": 0.7737755326158576, + "grad_norm": 2.0295822620391846, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7030522227287292, + "num_tokens": 179902217.0, + "step": 7046 + }, + { + "epoch": 0.7738853503184714, + "grad_norm": 2.078767776489258, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7138200402259827, + "num_tokens": 179929782.0, + "step": 7047 + }, + { + "epoch": 0.773995168021085, + "grad_norm": 2.2994723320007324, + "learning_rate": 1e-06, + "loss": 1.0316, + "mean_token_accuracy": 0.6865338087081909, + "num_tokens": 179955503.0, + "step": 7048 + }, + { + "epoch": 0.7741049857236987, + "grad_norm": 2.2403624057769775, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7110949754714966, + "num_tokens": 179980243.0, + "step": 7049 + }, + { + "epoch": 0.7742148034263123, + "grad_norm": 2.3192453384399414, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7331327199935913, + "num_tokens": 180002271.0, + "step": 7050 + }, + { + "epoch": 0.774324621128926, + "grad_norm": 2.232804775238037, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7129300832748413, + "num_tokens": 180027012.0, + "step": 7051 + }, + { + "epoch": 0.7744344388315396, + "grad_norm": 2.2045671939849854, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7184170484542847, + "num_tokens": 180052047.0, + "step": 7052 + }, + { + "epoch": 0.7745442565341533, + "grad_norm": 2.6582887172698975, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7153204679489136, + "num_tokens": 180070666.0, + "step": 7053 + }, + { + "epoch": 0.774654074236767, + "grad_norm": 2.257324695587158, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7127958536148071, + "num_tokens": 180095707.0, + "step": 7054 + }, + { + "epoch": 0.7747638919393807, + "grad_norm": 2.284320831298828, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7222135066986084, + "num_tokens": 180119587.0, + "step": 7055 + }, + { + "epoch": 0.7748737096419943, + "grad_norm": 2.216923236846924, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7102534770965576, + "num_tokens": 180145128.0, + "step": 7056 + }, + { + "epoch": 0.774983527344608, + "grad_norm": 2.3687500953674316, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7127078771591187, + "num_tokens": 180167193.0, + "step": 7057 + }, + { + "epoch": 0.7750933450472216, + "grad_norm": 2.273690938949585, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.704074501991272, + "num_tokens": 180192638.0, + "step": 7058 + }, + { + "epoch": 0.7752031627498353, + "grad_norm": 2.008943796157837, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.6945813894271851, + "num_tokens": 180223175.0, + "step": 7059 + }, + { + "epoch": 0.7753129804524489, + "grad_norm": 2.5531022548675537, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7127768397331238, + "num_tokens": 180243286.0, + "step": 7060 + }, + { + "epoch": 0.7754227981550625, + "grad_norm": 2.7694668769836426, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7326223850250244, + "num_tokens": 180262242.0, + "step": 7061 + }, + { + "epoch": 0.7755326158576763, + "grad_norm": 2.3306100368499756, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7211437225341797, + "num_tokens": 180284424.0, + "step": 7062 + }, + { + "epoch": 0.77564243356029, + "grad_norm": 2.481707811355591, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7308191061019897, + "num_tokens": 180302983.0, + "step": 7063 + }, + { + "epoch": 0.7757522512629036, + "grad_norm": 2.011598587036133, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7088932991027832, + "num_tokens": 180330976.0, + "step": 7064 + }, + { + "epoch": 0.7758620689655172, + "grad_norm": 2.2187018394470215, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7051948308944702, + "num_tokens": 180355081.0, + "step": 7065 + }, + { + "epoch": 0.7759718866681309, + "grad_norm": 2.1627626419067383, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.7001637816429138, + "num_tokens": 180379368.0, + "step": 7066 + }, + { + "epoch": 0.7760817043707445, + "grad_norm": 2.32686710357666, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7342925071716309, + "num_tokens": 180401378.0, + "step": 7067 + }, + { + "epoch": 0.7761915220733582, + "grad_norm": 1.9869282245635986, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6997491121292114, + "num_tokens": 180431194.0, + "step": 7068 + }, + { + "epoch": 0.7763013397759719, + "grad_norm": 2.2032618522644043, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.708233654499054, + "num_tokens": 180455760.0, + "step": 7069 + }, + { + "epoch": 0.7764111574785856, + "grad_norm": 2.457599639892578, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7213746905326843, + "num_tokens": 180475045.0, + "step": 7070 + }, + { + "epoch": 0.7765209751811992, + "grad_norm": 2.1596415042877197, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7118632793426514, + "num_tokens": 180499867.0, + "step": 7071 + }, + { + "epoch": 0.7766307928838129, + "grad_norm": 2.6070687770843506, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.707368016242981, + "num_tokens": 180520480.0, + "step": 7072 + }, + { + "epoch": 0.7767406105864265, + "grad_norm": 2.4008078575134277, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6895997524261475, + "num_tokens": 180543847.0, + "step": 7073 + }, + { + "epoch": 0.7768504282890402, + "grad_norm": 2.1895225048065186, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7219681739807129, + "num_tokens": 180567995.0, + "step": 7074 + }, + { + "epoch": 0.7769602459916538, + "grad_norm": 2.2798397541046143, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7167567610740662, + "num_tokens": 180590530.0, + "step": 7075 + }, + { + "epoch": 0.7770700636942676, + "grad_norm": 2.0152029991149902, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7213982939720154, + "num_tokens": 180618497.0, + "step": 7076 + }, + { + "epoch": 0.7771798813968812, + "grad_norm": 2.153932809829712, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.6950111985206604, + "num_tokens": 180643130.0, + "step": 7077 + }, + { + "epoch": 0.7772896990994949, + "grad_norm": 1.9530659914016724, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6886630058288574, + "num_tokens": 180675682.0, + "step": 7078 + }, + { + "epoch": 0.7773995168021085, + "grad_norm": 2.2473273277282715, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7233693599700928, + "num_tokens": 180697766.0, + "step": 7079 + }, + { + "epoch": 0.7775093345047221, + "grad_norm": 2.0530574321746826, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.6922004222869873, + "num_tokens": 180726349.0, + "step": 7080 + }, + { + "epoch": 0.7776191522073358, + "grad_norm": 2.031172513961792, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7103232145309448, + "num_tokens": 180753718.0, + "step": 7081 + }, + { + "epoch": 0.7777289699099494, + "grad_norm": 2.442073345184326, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.7035413384437561, + "num_tokens": 180774813.0, + "step": 7082 + }, + { + "epoch": 0.7778387876125632, + "grad_norm": 1.873453140258789, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7083500623703003, + "num_tokens": 180806227.0, + "step": 7083 + }, + { + "epoch": 0.7779486053151768, + "grad_norm": 2.1372387409210205, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6912649869918823, + "num_tokens": 180833477.0, + "step": 7084 + }, + { + "epoch": 0.7780584230177905, + "grad_norm": 2.708218574523926, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7228183150291443, + "num_tokens": 180850614.0, + "step": 7085 + }, + { + "epoch": 0.7781682407204041, + "grad_norm": 2.257204055786133, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7114682793617249, + "num_tokens": 180874759.0, + "step": 7086 + }, + { + "epoch": 0.7782780584230178, + "grad_norm": 2.332085132598877, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7025572061538696, + "num_tokens": 180898986.0, + "step": 7087 + }, + { + "epoch": 0.7783878761256314, + "grad_norm": 2.0286567211151123, + "learning_rate": 1e-06, + "loss": 1.094, + "mean_token_accuracy": 0.6799099445343018, + "num_tokens": 180929954.0, + "step": 7088 + }, + { + "epoch": 0.7784976938282451, + "grad_norm": 2.4324324131011963, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7132226228713989, + "num_tokens": 180951124.0, + "step": 7089 + }, + { + "epoch": 0.7786075115308587, + "grad_norm": 2.545496940612793, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7429149150848389, + "num_tokens": 180968127.0, + "step": 7090 + }, + { + "epoch": 0.7787173292334725, + "grad_norm": 2.451190710067749, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7023835778236389, + "num_tokens": 180990619.0, + "step": 7091 + }, + { + "epoch": 0.7788271469360861, + "grad_norm": 2.1410508155822754, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.683655321598053, + "num_tokens": 181018284.0, + "step": 7092 + }, + { + "epoch": 0.7789369646386998, + "grad_norm": 1.9191964864730835, + "learning_rate": 1e-06, + "loss": 1.0487, + "mean_token_accuracy": 0.6829652786254883, + "num_tokens": 181052387.0, + "step": 7093 + }, + { + "epoch": 0.7790467823413134, + "grad_norm": 2.430655002593994, + "learning_rate": 1e-06, + "loss": 1.0285, + "mean_token_accuracy": 0.688422679901123, + "num_tokens": 181073449.0, + "step": 7094 + }, + { + "epoch": 0.7791566000439271, + "grad_norm": 2.3979578018188477, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7182998061180115, + "num_tokens": 181095179.0, + "step": 7095 + }, + { + "epoch": 0.7792664177465407, + "grad_norm": 2.375192165374756, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6943561434745789, + "num_tokens": 181120366.0, + "step": 7096 + }, + { + "epoch": 0.7793762354491544, + "grad_norm": 2.2824559211730957, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.706688404083252, + "num_tokens": 181142275.0, + "step": 7097 + }, + { + "epoch": 0.7794860531517681, + "grad_norm": 2.2396092414855957, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7229721546173096, + "num_tokens": 181165552.0, + "step": 7098 + }, + { + "epoch": 0.7795958708543818, + "grad_norm": 2.0232205390930176, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7074126601219177, + "num_tokens": 181194991.0, + "step": 7099 + }, + { + "epoch": 0.7797056885569954, + "grad_norm": 2.1316449642181396, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.711489200592041, + "num_tokens": 181221874.0, + "step": 7100 + }, + { + "epoch": 0.779815506259609, + "grad_norm": 2.190791130065918, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6923695802688599, + "num_tokens": 181248808.0, + "step": 7101 + }, + { + "epoch": 0.7799253239622227, + "grad_norm": 2.077221393585205, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7137595415115356, + "num_tokens": 181276772.0, + "step": 7102 + }, + { + "epoch": 0.7800351416648363, + "grad_norm": 2.297903299331665, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7123047113418579, + "num_tokens": 181299636.0, + "step": 7103 + }, + { + "epoch": 0.78014495936745, + "grad_norm": 2.142552614212036, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.7018067836761475, + "num_tokens": 181325780.0, + "step": 7104 + }, + { + "epoch": 0.7802547770700637, + "grad_norm": 2.056004047393799, + "learning_rate": 1e-06, + "loss": 1.078, + "mean_token_accuracy": 0.6756818890571594, + "num_tokens": 181356094.0, + "step": 7105 + }, + { + "epoch": 0.7803645947726774, + "grad_norm": 1.8969519138336182, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7057439684867859, + "num_tokens": 181387303.0, + "step": 7106 + }, + { + "epoch": 0.780474412475291, + "grad_norm": 2.3409764766693115, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.6943023800849915, + "num_tokens": 181410647.0, + "step": 7107 + }, + { + "epoch": 0.7805842301779047, + "grad_norm": 2.0007712841033936, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7136577367782593, + "num_tokens": 181436014.0, + "step": 7108 + }, + { + "epoch": 0.7806940478805183, + "grad_norm": 2.035576343536377, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7162430286407471, + "num_tokens": 181462429.0, + "step": 7109 + }, + { + "epoch": 0.780803865583132, + "grad_norm": 2.288541793823242, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7133346796035767, + "num_tokens": 181485175.0, + "step": 7110 + }, + { + "epoch": 0.7809136832857456, + "grad_norm": 2.1838858127593994, + "learning_rate": 1e-06, + "loss": 1.0264, + "mean_token_accuracy": 0.6912633776664734, + "num_tokens": 181510781.0, + "step": 7111 + }, + { + "epoch": 0.7810235009883594, + "grad_norm": 1.9552884101867676, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7012048959732056, + "num_tokens": 181540686.0, + "step": 7112 + }, + { + "epoch": 0.781133318690973, + "grad_norm": 2.2307488918304443, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7192341089248657, + "num_tokens": 181565834.0, + "step": 7113 + }, + { + "epoch": 0.7812431363935867, + "grad_norm": 2.2571702003479004, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7190337181091309, + "num_tokens": 181588867.0, + "step": 7114 + }, + { + "epoch": 0.7813529540962003, + "grad_norm": 2.2878377437591553, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7104289531707764, + "num_tokens": 181612235.0, + "step": 7115 + }, + { + "epoch": 0.781462771798814, + "grad_norm": 2.163511276245117, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7097325325012207, + "num_tokens": 181636893.0, + "step": 7116 + }, + { + "epoch": 0.7815725895014276, + "grad_norm": 2.1209912300109863, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6997065544128418, + "num_tokens": 181664026.0, + "step": 7117 + }, + { + "epoch": 0.7816824072040413, + "grad_norm": 2.4124419689178467, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7178036570549011, + "num_tokens": 181685267.0, + "step": 7118 + }, + { + "epoch": 0.7817922249066549, + "grad_norm": 2.0678369998931885, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7053354978561401, + "num_tokens": 181713814.0, + "step": 7119 + }, + { + "epoch": 0.7819020426092687, + "grad_norm": 2.1581993103027344, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.707114577293396, + "num_tokens": 181740248.0, + "step": 7120 + }, + { + "epoch": 0.7820118603118823, + "grad_norm": 2.221122980117798, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.7002407908439636, + "num_tokens": 181765453.0, + "step": 7121 + }, + { + "epoch": 0.782121678014496, + "grad_norm": 2.3626368045806885, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.6965690851211548, + "num_tokens": 181789009.0, + "step": 7122 + }, + { + "epoch": 0.7822314957171096, + "grad_norm": 2.0268943309783936, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.7005115151405334, + "num_tokens": 181819510.0, + "step": 7123 + }, + { + "epoch": 0.7823413134197232, + "grad_norm": 2.0783169269561768, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.7033158540725708, + "num_tokens": 181847729.0, + "step": 7124 + }, + { + "epoch": 0.7824511311223369, + "grad_norm": 2.037748098373413, + "learning_rate": 1e-06, + "loss": 1.037, + "mean_token_accuracy": 0.6875686645507812, + "num_tokens": 181877750.0, + "step": 7125 + }, + { + "epoch": 0.7825609488249505, + "grad_norm": 2.005380392074585, + "learning_rate": 1e-06, + "loss": 1.0703, + "mean_token_accuracy": 0.676649808883667, + "num_tokens": 181906963.0, + "step": 7126 + }, + { + "epoch": 0.7826707665275643, + "grad_norm": 2.3307700157165527, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7328004837036133, + "num_tokens": 181928051.0, + "step": 7127 + }, + { + "epoch": 0.7827805842301779, + "grad_norm": 2.1296772956848145, + "learning_rate": 1e-06, + "loss": 1.0821, + "mean_token_accuracy": 0.674081027507782, + "num_tokens": 181955237.0, + "step": 7128 + }, + { + "epoch": 0.7828904019327916, + "grad_norm": 2.1901116371154785, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.6961908340454102, + "num_tokens": 181980430.0, + "step": 7129 + }, + { + "epoch": 0.7830002196354052, + "grad_norm": 2.2153472900390625, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7346171140670776, + "num_tokens": 182002252.0, + "step": 7130 + }, + { + "epoch": 0.7831100373380189, + "grad_norm": 2.163336992263794, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6930376291275024, + "num_tokens": 182029595.0, + "step": 7131 + }, + { + "epoch": 0.7832198550406325, + "grad_norm": 2.2083377838134766, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7050067186355591, + "num_tokens": 182056007.0, + "step": 7132 + }, + { + "epoch": 0.7833296727432462, + "grad_norm": 2.0491390228271484, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.6990360021591187, + "num_tokens": 182084605.0, + "step": 7133 + }, + { + "epoch": 0.7834394904458599, + "grad_norm": 2.289121627807617, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7118238210678101, + "num_tokens": 182109077.0, + "step": 7134 + }, + { + "epoch": 0.7835493081484736, + "grad_norm": 2.3592894077301025, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7159892320632935, + "num_tokens": 182130212.0, + "step": 7135 + }, + { + "epoch": 0.7836591258510872, + "grad_norm": 2.3430747985839844, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6881648302078247, + "num_tokens": 182151586.0, + "step": 7136 + }, + { + "epoch": 0.7837689435537009, + "grad_norm": 2.028651237487793, + "learning_rate": 1e-06, + "loss": 1.0531, + "mean_token_accuracy": 0.6886764764785767, + "num_tokens": 182181743.0, + "step": 7137 + }, + { + "epoch": 0.7838787612563145, + "grad_norm": 2.596118927001953, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7071775197982788, + "num_tokens": 182204390.0, + "step": 7138 + }, + { + "epoch": 0.7839885789589282, + "grad_norm": 2.370863676071167, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7375755310058594, + "num_tokens": 182224738.0, + "step": 7139 + }, + { + "epoch": 0.7840983966615418, + "grad_norm": 2.1404359340667725, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7297815084457397, + "num_tokens": 182250263.0, + "step": 7140 + }, + { + "epoch": 0.7842082143641556, + "grad_norm": 1.9325311183929443, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.6903632879257202, + "num_tokens": 182282758.0, + "step": 7141 + }, + { + "epoch": 0.7843180320667692, + "grad_norm": 2.088719367980957, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7095389366149902, + "num_tokens": 182309065.0, + "step": 7142 + }, + { + "epoch": 0.7844278497693828, + "grad_norm": 2.171027660369873, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6951072216033936, + "num_tokens": 182336713.0, + "step": 7143 + }, + { + "epoch": 0.7845376674719965, + "grad_norm": 2.174161195755005, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7120493650436401, + "num_tokens": 182361193.0, + "step": 7144 + }, + { + "epoch": 0.7846474851746101, + "grad_norm": 2.0315935611724854, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7056511640548706, + "num_tokens": 182388521.0, + "step": 7145 + }, + { + "epoch": 0.7847573028772238, + "grad_norm": 2.361555814743042, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.7037012577056885, + "num_tokens": 182411217.0, + "step": 7146 + }, + { + "epoch": 0.7848671205798374, + "grad_norm": 2.4062271118164062, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7120094895362854, + "num_tokens": 182435893.0, + "step": 7147 + }, + { + "epoch": 0.7849769382824512, + "grad_norm": 2.2772839069366455, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.7097494006156921, + "num_tokens": 182459955.0, + "step": 7148 + }, + { + "epoch": 0.7850867559850648, + "grad_norm": 1.953120470046997, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.7069307565689087, + "num_tokens": 182489351.0, + "step": 7149 + }, + { + "epoch": 0.7851965736876785, + "grad_norm": 2.113924980163574, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.7024470567703247, + "num_tokens": 182516887.0, + "step": 7150 + }, + { + "epoch": 0.7853063913902921, + "grad_norm": 2.234781503677368, + "learning_rate": 1e-06, + "loss": 1.0321, + "mean_token_accuracy": 0.684074878692627, + "num_tokens": 182542284.0, + "step": 7151 + }, + { + "epoch": 0.7854162090929058, + "grad_norm": 2.3437082767486572, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7201570272445679, + "num_tokens": 182564398.0, + "step": 7152 + }, + { + "epoch": 0.7855260267955194, + "grad_norm": 2.079800605773926, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7047950029373169, + "num_tokens": 182592683.0, + "step": 7153 + }, + { + "epoch": 0.7856358444981331, + "grad_norm": 2.160001277923584, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.7044250965118408, + "num_tokens": 182620525.0, + "step": 7154 + }, + { + "epoch": 0.7857456622007467, + "grad_norm": 2.439424991607666, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6894219517707825, + "num_tokens": 182642324.0, + "step": 7155 + }, + { + "epoch": 0.7858554799033605, + "grad_norm": 2.216789484024048, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7143489122390747, + "num_tokens": 182667217.0, + "step": 7156 + }, + { + "epoch": 0.7859652976059741, + "grad_norm": 2.1757657527923584, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.7016133666038513, + "num_tokens": 182693513.0, + "step": 7157 + }, + { + "epoch": 0.7860751153085878, + "grad_norm": 2.126431465148926, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7429530620574951, + "num_tokens": 182716369.0, + "step": 7158 + }, + { + "epoch": 0.7861849330112014, + "grad_norm": 2.6177542209625244, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7017863988876343, + "num_tokens": 182736305.0, + "step": 7159 + }, + { + "epoch": 0.786294750713815, + "grad_norm": 2.3525145053863525, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7233051657676697, + "num_tokens": 182759465.0, + "step": 7160 + }, + { + "epoch": 0.7864045684164287, + "grad_norm": 2.154750108718872, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.6954004764556885, + "num_tokens": 182784979.0, + "step": 7161 + }, + { + "epoch": 0.7865143861190423, + "grad_norm": 2.1255738735198975, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7362067699432373, + "num_tokens": 182808945.0, + "step": 7162 + }, + { + "epoch": 0.7866242038216561, + "grad_norm": 2.3308911323547363, + "learning_rate": 1e-06, + "loss": 1.0383, + "mean_token_accuracy": 0.6830621957778931, + "num_tokens": 182834370.0, + "step": 7163 + }, + { + "epoch": 0.7867340215242697, + "grad_norm": 2.080056667327881, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6981604099273682, + "num_tokens": 182864264.0, + "step": 7164 + }, + { + "epoch": 0.7868438392268834, + "grad_norm": 2.05937123298645, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.710398256778717, + "num_tokens": 182891940.0, + "step": 7165 + }, + { + "epoch": 0.786953656929497, + "grad_norm": 2.0707480907440186, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7247618436813354, + "num_tokens": 182916877.0, + "step": 7166 + }, + { + "epoch": 0.7870634746321107, + "grad_norm": 2.474604368209839, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7062331438064575, + "num_tokens": 182939967.0, + "step": 7167 + }, + { + "epoch": 0.7871732923347243, + "grad_norm": 2.364227294921875, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7292352914810181, + "num_tokens": 182961401.0, + "step": 7168 + }, + { + "epoch": 0.787283110037338, + "grad_norm": 1.9224491119384766, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.7040033340454102, + "num_tokens": 182992412.0, + "step": 7169 + }, + { + "epoch": 0.7873929277399517, + "grad_norm": 2.1239430904388428, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7067717909812927, + "num_tokens": 183019334.0, + "step": 7170 + }, + { + "epoch": 0.7875027454425654, + "grad_norm": 2.2451555728912354, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7134581208229065, + "num_tokens": 183042516.0, + "step": 7171 + }, + { + "epoch": 0.787612563145179, + "grad_norm": 2.3822317123413086, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7116340398788452, + "num_tokens": 183063570.0, + "step": 7172 + }, + { + "epoch": 0.7877223808477927, + "grad_norm": 2.054006576538086, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.7030719518661499, + "num_tokens": 183091459.0, + "step": 7173 + }, + { + "epoch": 0.7878321985504063, + "grad_norm": 2.2526397705078125, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6986984610557556, + "num_tokens": 183116674.0, + "step": 7174 + }, + { + "epoch": 0.78794201625302, + "grad_norm": 2.539123058319092, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.699720025062561, + "num_tokens": 183137938.0, + "step": 7175 + }, + { + "epoch": 0.7880518339556336, + "grad_norm": 2.4024882316589355, + "learning_rate": 1e-06, + "loss": 1.0867, + "mean_token_accuracy": 0.670411229133606, + "num_tokens": 183162027.0, + "step": 7176 + }, + { + "epoch": 0.7881616516582474, + "grad_norm": 2.251096248626709, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.7009624242782593, + "num_tokens": 183186009.0, + "step": 7177 + }, + { + "epoch": 0.788271469360861, + "grad_norm": 2.1587119102478027, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7122749090194702, + "num_tokens": 183211713.0, + "step": 7178 + }, + { + "epoch": 0.7883812870634747, + "grad_norm": 2.2358248233795166, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6966720819473267, + "num_tokens": 183238881.0, + "step": 7179 + }, + { + "epoch": 0.7884911047660883, + "grad_norm": 2.383859872817993, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7099535465240479, + "num_tokens": 183261763.0, + "step": 7180 + }, + { + "epoch": 0.788600922468702, + "grad_norm": 2.067004919052124, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7289543151855469, + "num_tokens": 183286409.0, + "step": 7181 + }, + { + "epoch": 0.7887107401713156, + "grad_norm": 2.2061078548431396, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7133890986442566, + "num_tokens": 183310758.0, + "step": 7182 + }, + { + "epoch": 0.7888205578739292, + "grad_norm": 2.322885274887085, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7039520740509033, + "num_tokens": 183333342.0, + "step": 7183 + }, + { + "epoch": 0.7889303755765429, + "grad_norm": 2.450568199157715, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.71197509765625, + "num_tokens": 183355459.0, + "step": 7184 + }, + { + "epoch": 0.7890401932791566, + "grad_norm": 2.336350917816162, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.6961318850517273, + "num_tokens": 183379958.0, + "step": 7185 + }, + { + "epoch": 0.7891500109817703, + "grad_norm": 1.9038209915161133, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7015190124511719, + "num_tokens": 183411207.0, + "step": 7186 + }, + { + "epoch": 0.7892598286843839, + "grad_norm": 2.0617074966430664, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7109364867210388, + "num_tokens": 183438445.0, + "step": 7187 + }, + { + "epoch": 0.7893696463869976, + "grad_norm": 2.4753060340881348, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7130680680274963, + "num_tokens": 183459034.0, + "step": 7188 + }, + { + "epoch": 0.7894794640896112, + "grad_norm": 2.270857095718384, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7175258994102478, + "num_tokens": 183481865.0, + "step": 7189 + }, + { + "epoch": 0.7895892817922249, + "grad_norm": 2.228454828262329, + "learning_rate": 1e-06, + "loss": 1.043, + "mean_token_accuracy": 0.6930093765258789, + "num_tokens": 183507087.0, + "step": 7190 + }, + { + "epoch": 0.7896990994948385, + "grad_norm": 2.3816161155700684, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7066711783409119, + "num_tokens": 183529526.0, + "step": 7191 + }, + { + "epoch": 0.7898089171974523, + "grad_norm": 1.9772427082061768, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7083812952041626, + "num_tokens": 183560299.0, + "step": 7192 + }, + { + "epoch": 0.7899187349000659, + "grad_norm": 1.9872803688049316, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7049753665924072, + "num_tokens": 183589009.0, + "step": 7193 + }, + { + "epoch": 0.7900285526026796, + "grad_norm": 2.237276554107666, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.731116533279419, + "num_tokens": 183612112.0, + "step": 7194 + }, + { + "epoch": 0.7901383703052932, + "grad_norm": 2.0103330612182617, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.7020460367202759, + "num_tokens": 183643729.0, + "step": 7195 + }, + { + "epoch": 0.7902481880079069, + "grad_norm": 2.392831802368164, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7257003784179688, + "num_tokens": 183665203.0, + "step": 7196 + }, + { + "epoch": 0.7903580057105205, + "grad_norm": 1.9421828985214233, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7126933336257935, + "num_tokens": 183696142.0, + "step": 7197 + }, + { + "epoch": 0.7904678234131342, + "grad_norm": 2.285092830657959, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.716942548751831, + "num_tokens": 183718566.0, + "step": 7198 + }, + { + "epoch": 0.7905776411157479, + "grad_norm": 2.20259428024292, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7168830633163452, + "num_tokens": 183743087.0, + "step": 7199 + }, + { + "epoch": 0.7906874588183616, + "grad_norm": 2.266038417816162, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.725409984588623, + "num_tokens": 183766681.0, + "step": 7200 + }, + { + "epoch": 0.7907972765209752, + "grad_norm": 2.3413784503936768, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7285176515579224, + "num_tokens": 183789717.0, + "step": 7201 + }, + { + "epoch": 0.7909070942235888, + "grad_norm": 2.074821710586548, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7160754203796387, + "num_tokens": 183815535.0, + "step": 7202 + }, + { + "epoch": 0.7910169119262025, + "grad_norm": 2.0499513149261475, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7219347357749939, + "num_tokens": 183840935.0, + "step": 7203 + }, + { + "epoch": 0.7911267296288161, + "grad_norm": 2.3111183643341064, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7105370759963989, + "num_tokens": 183864691.0, + "step": 7204 + }, + { + "epoch": 0.7912365473314298, + "grad_norm": 2.293948173522949, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.691766619682312, + "num_tokens": 183888605.0, + "step": 7205 + }, + { + "epoch": 0.7913463650340435, + "grad_norm": 2.0627331733703613, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7215858697891235, + "num_tokens": 183913767.0, + "step": 7206 + }, + { + "epoch": 0.7914561827366572, + "grad_norm": 2.2083704471588135, + "learning_rate": 1e-06, + "loss": 1.0379, + "mean_token_accuracy": 0.686936616897583, + "num_tokens": 183940045.0, + "step": 7207 + }, + { + "epoch": 0.7915660004392708, + "grad_norm": 2.244260311126709, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7131170034408569, + "num_tokens": 183965057.0, + "step": 7208 + }, + { + "epoch": 0.7916758181418845, + "grad_norm": 2.2686831951141357, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6916084289550781, + "num_tokens": 183990469.0, + "step": 7209 + }, + { + "epoch": 0.7917856358444981, + "grad_norm": 2.3504817485809326, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7255257964134216, + "num_tokens": 184011786.0, + "step": 7210 + }, + { + "epoch": 0.7918954535471118, + "grad_norm": 2.223209857940674, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7222741842269897, + "num_tokens": 184036549.0, + "step": 7211 + }, + { + "epoch": 0.7920052712497254, + "grad_norm": 2.240727186203003, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7189924120903015, + "num_tokens": 184059870.0, + "step": 7212 + }, + { + "epoch": 0.7921150889523391, + "grad_norm": 2.0035629272460938, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7185316681861877, + "num_tokens": 184087008.0, + "step": 7213 + }, + { + "epoch": 0.7922249066549528, + "grad_norm": 2.0835044384002686, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7059503793716431, + "num_tokens": 184113290.0, + "step": 7214 + }, + { + "epoch": 0.7923347243575665, + "grad_norm": 2.4043827056884766, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7150548100471497, + "num_tokens": 184134044.0, + "step": 7215 + }, + { + "epoch": 0.7924445420601801, + "grad_norm": 2.39357852935791, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.6893582344055176, + "num_tokens": 184155676.0, + "step": 7216 + }, + { + "epoch": 0.7925543597627938, + "grad_norm": 1.9337414503097534, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7173739671707153, + "num_tokens": 184184111.0, + "step": 7217 + }, + { + "epoch": 0.7926641774654074, + "grad_norm": 2.21168851852417, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.696513831615448, + "num_tokens": 184208344.0, + "step": 7218 + }, + { + "epoch": 0.792773995168021, + "grad_norm": 1.8474419116973877, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6936380863189697, + "num_tokens": 184243157.0, + "step": 7219 + }, + { + "epoch": 0.7928838128706347, + "grad_norm": 2.063429832458496, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6941457986831665, + "num_tokens": 184271017.0, + "step": 7220 + }, + { + "epoch": 0.7929936305732485, + "grad_norm": 1.9873936176300049, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7111389636993408, + "num_tokens": 184300614.0, + "step": 7221 + }, + { + "epoch": 0.7931034482758621, + "grad_norm": 2.037346601486206, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.7038928270339966, + "num_tokens": 184331277.0, + "step": 7222 + }, + { + "epoch": 0.7932132659784757, + "grad_norm": 2.5967907905578613, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7164531946182251, + "num_tokens": 184351672.0, + "step": 7223 + }, + { + "epoch": 0.7933230836810894, + "grad_norm": 2.0742807388305664, + "learning_rate": 1e-06, + "loss": 1.0252, + "mean_token_accuracy": 0.6852843761444092, + "num_tokens": 184382861.0, + "step": 7224 + }, + { + "epoch": 0.793432901383703, + "grad_norm": 2.2911577224731445, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7171739339828491, + "num_tokens": 184405136.0, + "step": 7225 + }, + { + "epoch": 0.7935427190863167, + "grad_norm": 2.523496627807617, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7139455080032349, + "num_tokens": 184424804.0, + "step": 7226 + }, + { + "epoch": 0.7936525367889303, + "grad_norm": 2.268599510192871, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7017241716384888, + "num_tokens": 184449896.0, + "step": 7227 + }, + { + "epoch": 0.7937623544915441, + "grad_norm": 2.2799317836761475, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7051657438278198, + "num_tokens": 184475737.0, + "step": 7228 + }, + { + "epoch": 0.7938721721941577, + "grad_norm": 2.205775260925293, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7106374502182007, + "num_tokens": 184499540.0, + "step": 7229 + }, + { + "epoch": 0.7939819898967714, + "grad_norm": 2.1588990688323975, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6915040612220764, + "num_tokens": 184525405.0, + "step": 7230 + }, + { + "epoch": 0.794091807599385, + "grad_norm": 2.5696911811828613, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7306907176971436, + "num_tokens": 184545054.0, + "step": 7231 + }, + { + "epoch": 0.7942016253019987, + "grad_norm": 2.035731077194214, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.7060633301734924, + "num_tokens": 184574272.0, + "step": 7232 + }, + { + "epoch": 0.7943114430046123, + "grad_norm": 2.254883050918579, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7266777753829956, + "num_tokens": 184598434.0, + "step": 7233 + }, + { + "epoch": 0.794421260707226, + "grad_norm": 2.0894672870635986, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.6990057229995728, + "num_tokens": 184624186.0, + "step": 7234 + }, + { + "epoch": 0.7945310784098397, + "grad_norm": 2.133409261703491, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.7003380656242371, + "num_tokens": 184650785.0, + "step": 7235 + }, + { + "epoch": 0.7946408961124534, + "grad_norm": 2.172666549682617, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7232711315155029, + "num_tokens": 184679129.0, + "step": 7236 + }, + { + "epoch": 0.794750713815067, + "grad_norm": 2.451651096343994, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7162883281707764, + "num_tokens": 184702338.0, + "step": 7237 + }, + { + "epoch": 0.7948605315176807, + "grad_norm": 1.9574843645095825, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7058128714561462, + "num_tokens": 184733082.0, + "step": 7238 + }, + { + "epoch": 0.7949703492202943, + "grad_norm": 2.006356954574585, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7183370590209961, + "num_tokens": 184762739.0, + "step": 7239 + }, + { + "epoch": 0.795080166922908, + "grad_norm": 2.132228374481201, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7091895341873169, + "num_tokens": 184790221.0, + "step": 7240 + }, + { + "epoch": 0.7951899846255216, + "grad_norm": 2.6033921241760254, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7351797819137573, + "num_tokens": 184807796.0, + "step": 7241 + }, + { + "epoch": 0.7952998023281352, + "grad_norm": 2.2675695419311523, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7228764295578003, + "num_tokens": 184829662.0, + "step": 7242 + }, + { + "epoch": 0.795409620030749, + "grad_norm": 2.2789692878723145, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7008633613586426, + "num_tokens": 184853160.0, + "step": 7243 + }, + { + "epoch": 0.7955194377333626, + "grad_norm": 2.390779495239258, + "learning_rate": 1e-06, + "loss": 1.058, + "mean_token_accuracy": 0.6742525696754456, + "num_tokens": 184875997.0, + "step": 7244 + }, + { + "epoch": 0.7956292554359763, + "grad_norm": 2.2303225994110107, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7154887318611145, + "num_tokens": 184901595.0, + "step": 7245 + }, + { + "epoch": 0.7957390731385899, + "grad_norm": 2.149366617202759, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.711891233921051, + "num_tokens": 184926413.0, + "step": 7246 + }, + { + "epoch": 0.7958488908412036, + "grad_norm": 2.0276851654052734, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.6985757350921631, + "num_tokens": 184955166.0, + "step": 7247 + }, + { + "epoch": 0.7959587085438172, + "grad_norm": 2.1271278858184814, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7270950675010681, + "num_tokens": 184979498.0, + "step": 7248 + }, + { + "epoch": 0.7960685262464309, + "grad_norm": 2.276149272918701, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7137935757637024, + "num_tokens": 185002369.0, + "step": 7249 + }, + { + "epoch": 0.7961783439490446, + "grad_norm": 2.198883533477783, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7058306932449341, + "num_tokens": 185026220.0, + "step": 7250 + }, + { + "epoch": 0.7962881616516583, + "grad_norm": 2.0032546520233154, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.696355938911438, + "num_tokens": 185056636.0, + "step": 7251 + }, + { + "epoch": 0.7963979793542719, + "grad_norm": 2.128932237625122, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.6989469528198242, + "num_tokens": 185083815.0, + "step": 7252 + }, + { + "epoch": 0.7965077970568856, + "grad_norm": 1.9874900579452515, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.7022383213043213, + "num_tokens": 185113778.0, + "step": 7253 + }, + { + "epoch": 0.7966176147594992, + "grad_norm": 2.2127392292022705, + "learning_rate": 1e-06, + "loss": 1.0288, + "mean_token_accuracy": 0.685181736946106, + "num_tokens": 185138777.0, + "step": 7254 + }, + { + "epoch": 0.7967274324621129, + "grad_norm": 2.5624232292175293, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.707579493522644, + "num_tokens": 185158019.0, + "step": 7255 + }, + { + "epoch": 0.7968372501647265, + "grad_norm": 2.076653242111206, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7033734321594238, + "num_tokens": 185185688.0, + "step": 7256 + }, + { + "epoch": 0.7969470678673403, + "grad_norm": 2.1680996417999268, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7405523061752319, + "num_tokens": 185209284.0, + "step": 7257 + }, + { + "epoch": 0.7970568855699539, + "grad_norm": 1.8000389337539673, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6923518776893616, + "num_tokens": 185244872.0, + "step": 7258 + }, + { + "epoch": 0.7971667032725676, + "grad_norm": 2.075451612472534, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6952595710754395, + "num_tokens": 185272668.0, + "step": 7259 + }, + { + "epoch": 0.7972765209751812, + "grad_norm": 2.1684422492980957, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7066158652305603, + "num_tokens": 185298472.0, + "step": 7260 + }, + { + "epoch": 0.7973863386777948, + "grad_norm": 2.4783008098602295, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7294320464134216, + "num_tokens": 185323719.0, + "step": 7261 + }, + { + "epoch": 0.7974961563804085, + "grad_norm": 2.1312193870544434, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.6977769136428833, + "num_tokens": 185349842.0, + "step": 7262 + }, + { + "epoch": 0.7976059740830221, + "grad_norm": 2.1714518070220947, + "learning_rate": 1e-06, + "loss": 1.0738, + "mean_token_accuracy": 0.6774497032165527, + "num_tokens": 185377082.0, + "step": 7263 + }, + { + "epoch": 0.7977157917856359, + "grad_norm": 1.9566868543624878, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7097915410995483, + "num_tokens": 185407485.0, + "step": 7264 + }, + { + "epoch": 0.7978256094882495, + "grad_norm": 1.9358232021331787, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7162024974822998, + "num_tokens": 185436825.0, + "step": 7265 + }, + { + "epoch": 0.7979354271908632, + "grad_norm": 2.35235595703125, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7319116592407227, + "num_tokens": 185458217.0, + "step": 7266 + }, + { + "epoch": 0.7980452448934768, + "grad_norm": 2.108506202697754, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.6988821625709534, + "num_tokens": 185484954.0, + "step": 7267 + }, + { + "epoch": 0.7981550625960905, + "grad_norm": 2.2310397624969482, + "learning_rate": 1e-06, + "loss": 1.1054, + "mean_token_accuracy": 0.6639620065689087, + "num_tokens": 185511901.0, + "step": 7268 + }, + { + "epoch": 0.7982648802987041, + "grad_norm": 2.6380560398101807, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7172359228134155, + "num_tokens": 185530617.0, + "step": 7269 + }, + { + "epoch": 0.7983746980013178, + "grad_norm": 2.110978603363037, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.6974995136260986, + "num_tokens": 185556538.0, + "step": 7270 + }, + { + "epoch": 0.7984845157039314, + "grad_norm": 2.510694980621338, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7284798622131348, + "num_tokens": 185575227.0, + "step": 7271 + }, + { + "epoch": 0.7985943334065452, + "grad_norm": 2.5526702404022217, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7370659112930298, + "num_tokens": 185594543.0, + "step": 7272 + }, + { + "epoch": 0.7987041511091588, + "grad_norm": 2.1839702129364014, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7138394117355347, + "num_tokens": 185620642.0, + "step": 7273 + }, + { + "epoch": 0.7988139688117725, + "grad_norm": 2.3286287784576416, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7279490828514099, + "num_tokens": 185641984.0, + "step": 7274 + }, + { + "epoch": 0.7989237865143861, + "grad_norm": 2.4087183475494385, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7259559631347656, + "num_tokens": 185662466.0, + "step": 7275 + }, + { + "epoch": 0.7990336042169998, + "grad_norm": 2.047581195831299, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.6850720643997192, + "num_tokens": 185691935.0, + "step": 7276 + }, + { + "epoch": 0.7991434219196134, + "grad_norm": 2.0387659072875977, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7112511396408081, + "num_tokens": 185718250.0, + "step": 7277 + }, + { + "epoch": 0.799253239622227, + "grad_norm": 2.2581942081451416, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7188608646392822, + "num_tokens": 185741697.0, + "step": 7278 + }, + { + "epoch": 0.7993630573248408, + "grad_norm": 2.196834087371826, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6937233805656433, + "num_tokens": 185768312.0, + "step": 7279 + }, + { + "epoch": 0.7994728750274545, + "grad_norm": 2.260984420776367, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.700472891330719, + "num_tokens": 185790994.0, + "step": 7280 + }, + { + "epoch": 0.7995826927300681, + "grad_norm": 2.238760471343994, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.723724365234375, + "num_tokens": 185813023.0, + "step": 7281 + }, + { + "epoch": 0.7996925104326817, + "grad_norm": 2.0681705474853516, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7316989898681641, + "num_tokens": 185839727.0, + "step": 7282 + }, + { + "epoch": 0.7998023281352954, + "grad_norm": 2.0791471004486084, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7191708087921143, + "num_tokens": 185867387.0, + "step": 7283 + }, + { + "epoch": 0.799912145837909, + "grad_norm": 2.141786575317383, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7065908908843994, + "num_tokens": 185894039.0, + "step": 7284 + }, + { + "epoch": 0.8000219635405227, + "grad_norm": 1.9955819845199585, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6970623135566711, + "num_tokens": 185923458.0, + "step": 7285 + }, + { + "epoch": 0.8001317812431364, + "grad_norm": 2.2704577445983887, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.7020416855812073, + "num_tokens": 185947615.0, + "step": 7286 + }, + { + "epoch": 0.8002415989457501, + "grad_norm": 2.181957960128784, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.7010635733604431, + "num_tokens": 185972640.0, + "step": 7287 + }, + { + "epoch": 0.8003514166483637, + "grad_norm": 2.6292543411254883, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7103359699249268, + "num_tokens": 185990702.0, + "step": 7288 + }, + { + "epoch": 0.8004612343509774, + "grad_norm": 2.286485195159912, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7366239428520203, + "num_tokens": 186014397.0, + "step": 7289 + }, + { + "epoch": 0.800571052053591, + "grad_norm": 2.207148313522339, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.6999268531799316, + "num_tokens": 186038913.0, + "step": 7290 + }, + { + "epoch": 0.8006808697562047, + "grad_norm": 2.128499746322632, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7171970009803772, + "num_tokens": 186066227.0, + "step": 7291 + }, + { + "epoch": 0.8007906874588183, + "grad_norm": 1.9239940643310547, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6920256018638611, + "num_tokens": 186098482.0, + "step": 7292 + }, + { + "epoch": 0.8009005051614321, + "grad_norm": 2.2437710762023926, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7163746356964111, + "num_tokens": 186121432.0, + "step": 7293 + }, + { + "epoch": 0.8010103228640457, + "grad_norm": 1.9724315404891968, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7051554918289185, + "num_tokens": 186154068.0, + "step": 7294 + }, + { + "epoch": 0.8011201405666594, + "grad_norm": 1.9044562578201294, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.6973395943641663, + "num_tokens": 186187389.0, + "step": 7295 + }, + { + "epoch": 0.801229958269273, + "grad_norm": 2.024597406387329, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7373193502426147, + "num_tokens": 186214483.0, + "step": 7296 + }, + { + "epoch": 0.8013397759718867, + "grad_norm": 2.1079370975494385, + "learning_rate": 1e-06, + "loss": 1.0362, + "mean_token_accuracy": 0.68475341796875, + "num_tokens": 186243361.0, + "step": 7297 + }, + { + "epoch": 0.8014495936745003, + "grad_norm": 2.3689637184143066, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7245032787322998, + "num_tokens": 186264843.0, + "step": 7298 + }, + { + "epoch": 0.801559411377114, + "grad_norm": 2.093996286392212, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7058334946632385, + "num_tokens": 186290816.0, + "step": 7299 + }, + { + "epoch": 0.8016692290797276, + "grad_norm": 1.924447774887085, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.6868312954902649, + "num_tokens": 186322888.0, + "step": 7300 + }, + { + "epoch": 0.8017790467823414, + "grad_norm": 2.1010916233062744, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.704353928565979, + "num_tokens": 186350123.0, + "step": 7301 + }, + { + "epoch": 0.801888864484955, + "grad_norm": 2.1710071563720703, + "learning_rate": 1e-06, + "loss": 1.0478, + "mean_token_accuracy": 0.6809449791908264, + "num_tokens": 186377327.0, + "step": 7302 + }, + { + "epoch": 0.8019986821875686, + "grad_norm": 2.346839666366577, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.705757737159729, + "num_tokens": 186401140.0, + "step": 7303 + }, + { + "epoch": 0.8021084998901823, + "grad_norm": 2.304530620574951, + "learning_rate": 1e-06, + "loss": 1.0588, + "mean_token_accuracy": 0.6812657117843628, + "num_tokens": 186424718.0, + "step": 7304 + }, + { + "epoch": 0.8022183175927959, + "grad_norm": 2.5403599739074707, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7314590811729431, + "num_tokens": 186443304.0, + "step": 7305 + }, + { + "epoch": 0.8023281352954096, + "grad_norm": 2.252257823944092, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.687751293182373, + "num_tokens": 186467356.0, + "step": 7306 + }, + { + "epoch": 0.8024379529980232, + "grad_norm": 2.313889265060425, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7332091331481934, + "num_tokens": 186487470.0, + "step": 7307 + }, + { + "epoch": 0.802547770700637, + "grad_norm": 2.1927742958068848, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.710678219795227, + "num_tokens": 186511739.0, + "step": 7308 + }, + { + "epoch": 0.8026575884032506, + "grad_norm": 2.5667359828948975, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7114048600196838, + "num_tokens": 186531538.0, + "step": 7309 + }, + { + "epoch": 0.8027674061058643, + "grad_norm": 2.37925386428833, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7224773168563843, + "num_tokens": 186551646.0, + "step": 7310 + }, + { + "epoch": 0.8028772238084779, + "grad_norm": 2.2900009155273438, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7171457409858704, + "num_tokens": 186574915.0, + "step": 7311 + }, + { + "epoch": 0.8029870415110916, + "grad_norm": 2.0497286319732666, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7163106203079224, + "num_tokens": 186602578.0, + "step": 7312 + }, + { + "epoch": 0.8030968592137052, + "grad_norm": 2.2086944580078125, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.708938479423523, + "num_tokens": 186626095.0, + "step": 7313 + }, + { + "epoch": 0.8032066769163189, + "grad_norm": 2.3707056045532227, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7132978439331055, + "num_tokens": 186647371.0, + "step": 7314 + }, + { + "epoch": 0.8033164946189326, + "grad_norm": 2.069467067718506, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7342562079429626, + "num_tokens": 186673048.0, + "step": 7315 + }, + { + "epoch": 0.8034263123215463, + "grad_norm": 2.388747215270996, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7160378098487854, + "num_tokens": 186692791.0, + "step": 7316 + }, + { + "epoch": 0.8035361300241599, + "grad_norm": 2.157677412033081, + "learning_rate": 1e-06, + "loss": 1.0246, + "mean_token_accuracy": 0.6878156065940857, + "num_tokens": 186718208.0, + "step": 7317 + }, + { + "epoch": 0.8036459477267736, + "grad_norm": 2.2001781463623047, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7207537889480591, + "num_tokens": 186742071.0, + "step": 7318 + }, + { + "epoch": 0.8037557654293872, + "grad_norm": 2.0757977962493896, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.716324508190155, + "num_tokens": 186768667.0, + "step": 7319 + }, + { + "epoch": 0.8038655831320008, + "grad_norm": 2.430814743041992, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7021543979644775, + "num_tokens": 186789898.0, + "step": 7320 + }, + { + "epoch": 0.8039754008346145, + "grad_norm": 2.355259418487549, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7328903079032898, + "num_tokens": 186810896.0, + "step": 7321 + }, + { + "epoch": 0.8040852185372283, + "grad_norm": 2.0083603858947754, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.7041723728179932, + "num_tokens": 186840557.0, + "step": 7322 + }, + { + "epoch": 0.8041950362398419, + "grad_norm": 2.192204475402832, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7309861779212952, + "num_tokens": 186864733.0, + "step": 7323 + }, + { + "epoch": 0.8043048539424555, + "grad_norm": 2.200103521347046, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.6984320878982544, + "num_tokens": 186890402.0, + "step": 7324 + }, + { + "epoch": 0.8044146716450692, + "grad_norm": 1.853798508644104, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6942518353462219, + "num_tokens": 186925158.0, + "step": 7325 + }, + { + "epoch": 0.8045244893476828, + "grad_norm": 2.041564702987671, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.723699688911438, + "num_tokens": 186953218.0, + "step": 7326 + }, + { + "epoch": 0.8046343070502965, + "grad_norm": 1.9825313091278076, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7066439390182495, + "num_tokens": 186980843.0, + "step": 7327 + }, + { + "epoch": 0.8047441247529101, + "grad_norm": 2.073417901992798, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.7046177387237549, + "num_tokens": 187007286.0, + "step": 7328 + }, + { + "epoch": 0.8048539424555239, + "grad_norm": 2.13236665725708, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.6998347043991089, + "num_tokens": 187033966.0, + "step": 7329 + }, + { + "epoch": 0.8049637601581375, + "grad_norm": 2.4950029850006104, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7218978404998779, + "num_tokens": 187054046.0, + "step": 7330 + }, + { + "epoch": 0.8050735778607512, + "grad_norm": 2.145303964614868, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7073692083358765, + "num_tokens": 187081264.0, + "step": 7331 + }, + { + "epoch": 0.8051833955633648, + "grad_norm": 2.22529673576355, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7116984724998474, + "num_tokens": 187107073.0, + "step": 7332 + }, + { + "epoch": 0.8052932132659785, + "grad_norm": 1.9134671688079834, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7059910893440247, + "num_tokens": 187138168.0, + "step": 7333 + }, + { + "epoch": 0.8054030309685921, + "grad_norm": 2.124255895614624, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7027643918991089, + "num_tokens": 187166525.0, + "step": 7334 + }, + { + "epoch": 0.8055128486712058, + "grad_norm": 1.984107494354248, + "learning_rate": 1e-06, + "loss": 1.0341, + "mean_token_accuracy": 0.6828818321228027, + "num_tokens": 187197680.0, + "step": 7335 + }, + { + "epoch": 0.8056226663738194, + "grad_norm": 1.9813834428787231, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.701101541519165, + "num_tokens": 187225563.0, + "step": 7336 + }, + { + "epoch": 0.8057324840764332, + "grad_norm": 2.512697696685791, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.737838089466095, + "num_tokens": 187243987.0, + "step": 7337 + }, + { + "epoch": 0.8058423017790468, + "grad_norm": 1.9049508571624756, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.700524091720581, + "num_tokens": 187274497.0, + "step": 7338 + }, + { + "epoch": 0.8059521194816605, + "grad_norm": 2.180351734161377, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7188053131103516, + "num_tokens": 187298795.0, + "step": 7339 + }, + { + "epoch": 0.8060619371842741, + "grad_norm": 2.228196859359741, + "learning_rate": 1e-06, + "loss": 1.0373, + "mean_token_accuracy": 0.7019616365432739, + "num_tokens": 187322761.0, + "step": 7340 + }, + { + "epoch": 0.8061717548868877, + "grad_norm": 2.244542121887207, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7070186734199524, + "num_tokens": 187347963.0, + "step": 7341 + }, + { + "epoch": 0.8062815725895014, + "grad_norm": 2.1892905235290527, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7139493227005005, + "num_tokens": 187373124.0, + "step": 7342 + }, + { + "epoch": 0.806391390292115, + "grad_norm": 2.200054883956909, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7095344662666321, + "num_tokens": 187397167.0, + "step": 7343 + }, + { + "epoch": 0.8065012079947288, + "grad_norm": 2.3080527782440186, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7176731824874878, + "num_tokens": 187419917.0, + "step": 7344 + }, + { + "epoch": 0.8066110256973424, + "grad_norm": 2.2634451389312744, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.6856218576431274, + "num_tokens": 187446453.0, + "step": 7345 + }, + { + "epoch": 0.8067208433999561, + "grad_norm": 2.141671895980835, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.720266580581665, + "num_tokens": 187472741.0, + "step": 7346 + }, + { + "epoch": 0.8068306611025697, + "grad_norm": 2.062605857849121, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.694269061088562, + "num_tokens": 187501240.0, + "step": 7347 + }, + { + "epoch": 0.8069404788051834, + "grad_norm": 1.8708664178848267, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6882972717285156, + "num_tokens": 187535072.0, + "step": 7348 + }, + { + "epoch": 0.807050296507797, + "grad_norm": 2.286957025527954, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.7037643194198608, + "num_tokens": 187558581.0, + "step": 7349 + }, + { + "epoch": 0.8071601142104107, + "grad_norm": 1.9700325727462769, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.7047770619392395, + "num_tokens": 187586817.0, + "step": 7350 + }, + { + "epoch": 0.8072699319130244, + "grad_norm": 2.2601070404052734, + "learning_rate": 1e-06, + "loss": 1.0342, + "mean_token_accuracy": 0.6870489120483398, + "num_tokens": 187612173.0, + "step": 7351 + }, + { + "epoch": 0.8073797496156381, + "grad_norm": 2.1698157787323, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.6918627023696899, + "num_tokens": 187639320.0, + "step": 7352 + }, + { + "epoch": 0.8074895673182517, + "grad_norm": 2.0905942916870117, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7181714773178101, + "num_tokens": 187665670.0, + "step": 7353 + }, + { + "epoch": 0.8075993850208654, + "grad_norm": 2.2133631706237793, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.6829435229301453, + "num_tokens": 187690867.0, + "step": 7354 + }, + { + "epoch": 0.807709202723479, + "grad_norm": 2.2402124404907227, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7184559106826782, + "num_tokens": 187713975.0, + "step": 7355 + }, + { + "epoch": 0.8078190204260927, + "grad_norm": 2.0862996578216553, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6918281316757202, + "num_tokens": 187742118.0, + "step": 7356 + }, + { + "epoch": 0.8079288381287063, + "grad_norm": 2.0065197944641113, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7283629179000854, + "num_tokens": 187768456.0, + "step": 7357 + }, + { + "epoch": 0.8080386558313201, + "grad_norm": 2.2164525985717773, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7060401439666748, + "num_tokens": 187791530.0, + "step": 7358 + }, + { + "epoch": 0.8081484735339337, + "grad_norm": 2.431793212890625, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7141975164413452, + "num_tokens": 187813145.0, + "step": 7359 + }, + { + "epoch": 0.8082582912365474, + "grad_norm": 2.2062323093414307, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.728274941444397, + "num_tokens": 187835984.0, + "step": 7360 + }, + { + "epoch": 0.808368108939161, + "grad_norm": 1.9633300304412842, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7090891599655151, + "num_tokens": 187866956.0, + "step": 7361 + }, + { + "epoch": 0.8084779266417746, + "grad_norm": 2.2844111919403076, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7136259078979492, + "num_tokens": 187889329.0, + "step": 7362 + }, + { + "epoch": 0.8085877443443883, + "grad_norm": 2.199636459350586, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7074654698371887, + "num_tokens": 187914379.0, + "step": 7363 + }, + { + "epoch": 0.8086975620470019, + "grad_norm": 2.3038623332977295, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7062438726425171, + "num_tokens": 187938143.0, + "step": 7364 + }, + { + "epoch": 0.8088073797496156, + "grad_norm": 2.224569082260132, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.721144437789917, + "num_tokens": 187961933.0, + "step": 7365 + }, + { + "epoch": 0.8089171974522293, + "grad_norm": 1.9794625043869019, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.6901513338088989, + "num_tokens": 187993231.0, + "step": 7366 + }, + { + "epoch": 0.809027015154843, + "grad_norm": 2.3329062461853027, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7261621952056885, + "num_tokens": 188015454.0, + "step": 7367 + }, + { + "epoch": 0.8091368328574566, + "grad_norm": 2.4277780055999756, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7034883499145508, + "num_tokens": 188037741.0, + "step": 7368 + }, + { + "epoch": 0.8092466505600703, + "grad_norm": 2.017425775527954, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.7004907131195068, + "num_tokens": 188067753.0, + "step": 7369 + }, + { + "epoch": 0.8093564682626839, + "grad_norm": 2.3332009315490723, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7111684083938599, + "num_tokens": 188092893.0, + "step": 7370 + }, + { + "epoch": 0.8094662859652976, + "grad_norm": 2.1247639656066895, + "learning_rate": 1e-06, + "loss": 1.0341, + "mean_token_accuracy": 0.6953655481338501, + "num_tokens": 188121314.0, + "step": 7371 + }, + { + "epoch": 0.8095761036679112, + "grad_norm": 2.20025634765625, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7241337299346924, + "num_tokens": 188146643.0, + "step": 7372 + }, + { + "epoch": 0.809685921370525, + "grad_norm": 2.1489198207855225, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6941973567008972, + "num_tokens": 188173702.0, + "step": 7373 + }, + { + "epoch": 0.8097957390731386, + "grad_norm": 2.370344400405884, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7087504863739014, + "num_tokens": 188195886.0, + "step": 7374 + }, + { + "epoch": 0.8099055567757523, + "grad_norm": 2.042571783065796, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6953199505805969, + "num_tokens": 188224566.0, + "step": 7375 + }, + { + "epoch": 0.8100153744783659, + "grad_norm": 2.189366579055786, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7127127647399902, + "num_tokens": 188249923.0, + "step": 7376 + }, + { + "epoch": 0.8101251921809796, + "grad_norm": 2.0683460235595703, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.7080473899841309, + "num_tokens": 188276532.0, + "step": 7377 + }, + { + "epoch": 0.8102350098835932, + "grad_norm": 2.236466407775879, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7092339396476746, + "num_tokens": 188300218.0, + "step": 7378 + }, + { + "epoch": 0.8103448275862069, + "grad_norm": 2.244471549987793, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.6925783157348633, + "num_tokens": 188324902.0, + "step": 7379 + }, + { + "epoch": 0.8104546452888206, + "grad_norm": 2.2421875, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6968984603881836, + "num_tokens": 188348502.0, + "step": 7380 + }, + { + "epoch": 0.8105644629914343, + "grad_norm": 2.1073217391967773, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6942970752716064, + "num_tokens": 188374591.0, + "step": 7381 + }, + { + "epoch": 0.8106742806940479, + "grad_norm": 2.237791061401367, + "learning_rate": 1e-06, + "loss": 0.9913, + "mean_token_accuracy": 0.6962205171585083, + "num_tokens": 188398906.0, + "step": 7382 + }, + { + "epoch": 0.8107840983966615, + "grad_norm": 2.472296953201294, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7175568342208862, + "num_tokens": 188419039.0, + "step": 7383 + }, + { + "epoch": 0.8108939160992752, + "grad_norm": 2.1014485359191895, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7088865041732788, + "num_tokens": 188445449.0, + "step": 7384 + }, + { + "epoch": 0.8110037338018888, + "grad_norm": 2.2431490421295166, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7462055087089539, + "num_tokens": 188469622.0, + "step": 7385 + }, + { + "epoch": 0.8111135515045025, + "grad_norm": 2.2369680404663086, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7154257893562317, + "num_tokens": 188493872.0, + "step": 7386 + }, + { + "epoch": 0.8112233692071162, + "grad_norm": 2.1999404430389404, + "learning_rate": 1e-06, + "loss": 1.0563, + "mean_token_accuracy": 0.6893317699432373, + "num_tokens": 188520260.0, + "step": 7387 + }, + { + "epoch": 0.8113331869097299, + "grad_norm": 2.0870211124420166, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7120043635368347, + "num_tokens": 188546661.0, + "step": 7388 + }, + { + "epoch": 0.8114430046123435, + "grad_norm": 2.0321857929229736, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7058869004249573, + "num_tokens": 188577010.0, + "step": 7389 + }, + { + "epoch": 0.8115528223149572, + "grad_norm": 2.1546261310577393, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.6977362036705017, + "num_tokens": 188603509.0, + "step": 7390 + }, + { + "epoch": 0.8116626400175708, + "grad_norm": 2.2796406745910645, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7202972769737244, + "num_tokens": 188625890.0, + "step": 7391 + }, + { + "epoch": 0.8117724577201845, + "grad_norm": 2.0768327713012695, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.7057024240493774, + "num_tokens": 188653585.0, + "step": 7392 + }, + { + "epoch": 0.8118822754227981, + "grad_norm": 2.1853742599487305, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7131173014640808, + "num_tokens": 188677529.0, + "step": 7393 + }, + { + "epoch": 0.8119920931254118, + "grad_norm": 2.5545876026153564, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7187594771385193, + "num_tokens": 188696650.0, + "step": 7394 + }, + { + "epoch": 0.8121019108280255, + "grad_norm": 2.0525996685028076, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.7082149982452393, + "num_tokens": 188723951.0, + "step": 7395 + }, + { + "epoch": 0.8122117285306392, + "grad_norm": 2.2122557163238525, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7105017304420471, + "num_tokens": 188746477.0, + "step": 7396 + }, + { + "epoch": 0.8123215462332528, + "grad_norm": 2.248328924179077, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.715472400188446, + "num_tokens": 188769993.0, + "step": 7397 + }, + { + "epoch": 0.8124313639358665, + "grad_norm": 2.0777623653411865, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7039374709129333, + "num_tokens": 188797192.0, + "step": 7398 + }, + { + "epoch": 0.8125411816384801, + "grad_norm": 2.4361493587493896, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7301880121231079, + "num_tokens": 188817720.0, + "step": 7399 + }, + { + "epoch": 0.8126509993410937, + "grad_norm": 2.4826505184173584, + "learning_rate": 1e-06, + "loss": 1.0325, + "mean_token_accuracy": 0.6882985234260559, + "num_tokens": 188838936.0, + "step": 7400 + }, + { + "epoch": 0.8127608170437074, + "grad_norm": 2.4471237659454346, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7263636589050293, + "num_tokens": 188858959.0, + "step": 7401 + }, + { + "epoch": 0.8128706347463212, + "grad_norm": 2.2496228218078613, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7175986170768738, + "num_tokens": 188883943.0, + "step": 7402 + }, + { + "epoch": 0.8129804524489348, + "grad_norm": 2.1206235885620117, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7133074998855591, + "num_tokens": 188910853.0, + "step": 7403 + }, + { + "epoch": 0.8130902701515484, + "grad_norm": 2.2491822242736816, + "learning_rate": 1e-06, + "loss": 1.0924, + "mean_token_accuracy": 0.6860020756721497, + "num_tokens": 188935778.0, + "step": 7404 + }, + { + "epoch": 0.8132000878541621, + "grad_norm": 1.9838073253631592, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.710840106010437, + "num_tokens": 188966174.0, + "step": 7405 + }, + { + "epoch": 0.8133099055567757, + "grad_norm": 2.054547071456909, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7094157338142395, + "num_tokens": 188993313.0, + "step": 7406 + }, + { + "epoch": 0.8134197232593894, + "grad_norm": 1.9949116706848145, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.703760027885437, + "num_tokens": 189025340.0, + "step": 7407 + }, + { + "epoch": 0.813529540962003, + "grad_norm": 2.2892701625823975, + "learning_rate": 1e-06, + "loss": 1.0522, + "mean_token_accuracy": 0.6822316646575928, + "num_tokens": 189050498.0, + "step": 7408 + }, + { + "epoch": 0.8136393586646168, + "grad_norm": 2.195199966430664, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7051712274551392, + "num_tokens": 189075897.0, + "step": 7409 + }, + { + "epoch": 0.8137491763672304, + "grad_norm": 2.1405723094940186, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7224831581115723, + "num_tokens": 189101830.0, + "step": 7410 + }, + { + "epoch": 0.8138589940698441, + "grad_norm": 2.4707486629486084, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.711661159992218, + "num_tokens": 189122246.0, + "step": 7411 + }, + { + "epoch": 0.8139688117724577, + "grad_norm": 2.285346746444702, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.737214207649231, + "num_tokens": 189147129.0, + "step": 7412 + }, + { + "epoch": 0.8140786294750714, + "grad_norm": 2.2800378799438477, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7025130987167358, + "num_tokens": 189172671.0, + "step": 7413 + }, + { + "epoch": 0.814188447177685, + "grad_norm": 2.1535167694091797, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7262201309204102, + "num_tokens": 189198121.0, + "step": 7414 + }, + { + "epoch": 0.8142982648802987, + "grad_norm": 2.3406405448913574, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.704764723777771, + "num_tokens": 189221391.0, + "step": 7415 + }, + { + "epoch": 0.8144080825829124, + "grad_norm": 2.2323834896087646, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7141478061676025, + "num_tokens": 189244549.0, + "step": 7416 + }, + { + "epoch": 0.8145179002855261, + "grad_norm": 2.4974405765533447, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6987742185592651, + "num_tokens": 189265767.0, + "step": 7417 + }, + { + "epoch": 0.8146277179881397, + "grad_norm": 2.617095708847046, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.7071675658226013, + "num_tokens": 189286142.0, + "step": 7418 + }, + { + "epoch": 0.8147375356907534, + "grad_norm": 2.1618425846099854, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7408807873725891, + "num_tokens": 189309597.0, + "step": 7419 + }, + { + "epoch": 0.814847353393367, + "grad_norm": 1.9629216194152832, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6911975145339966, + "num_tokens": 189341349.0, + "step": 7420 + }, + { + "epoch": 0.8149571710959806, + "grad_norm": 2.383216619491577, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7367837429046631, + "num_tokens": 189363113.0, + "step": 7421 + }, + { + "epoch": 0.8150669887985943, + "grad_norm": 2.1174328327178955, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6910686492919922, + "num_tokens": 189390910.0, + "step": 7422 + }, + { + "epoch": 0.8151768065012079, + "grad_norm": 2.04463267326355, + "learning_rate": 1e-06, + "loss": 1.0594, + "mean_token_accuracy": 0.6781497597694397, + "num_tokens": 189420827.0, + "step": 7423 + }, + { + "epoch": 0.8152866242038217, + "grad_norm": 2.544513463973999, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7246167659759521, + "num_tokens": 189439239.0, + "step": 7424 + }, + { + "epoch": 0.8153964419064353, + "grad_norm": 1.9327322244644165, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6882364749908447, + "num_tokens": 189470533.0, + "step": 7425 + }, + { + "epoch": 0.815506259609049, + "grad_norm": 2.5599184036254883, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7309679985046387, + "num_tokens": 189489620.0, + "step": 7426 + }, + { + "epoch": 0.8156160773116626, + "grad_norm": 2.2431235313415527, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7226489782333374, + "num_tokens": 189514939.0, + "step": 7427 + }, + { + "epoch": 0.8157258950142763, + "grad_norm": 2.0030503273010254, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.69691002368927, + "num_tokens": 189547145.0, + "step": 7428 + }, + { + "epoch": 0.8158357127168899, + "grad_norm": 2.557729721069336, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.718929648399353, + "num_tokens": 189566354.0, + "step": 7429 + }, + { + "epoch": 0.8159455304195036, + "grad_norm": 2.200547933578491, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6932287216186523, + "num_tokens": 189591296.0, + "step": 7430 + }, + { + "epoch": 0.8160553481221173, + "grad_norm": 2.1174192428588867, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7116733193397522, + "num_tokens": 189618090.0, + "step": 7431 + }, + { + "epoch": 0.816165165824731, + "grad_norm": 2.1517493724823, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7094910740852356, + "num_tokens": 189643416.0, + "step": 7432 + }, + { + "epoch": 0.8162749835273446, + "grad_norm": 2.3539681434631348, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7139588594436646, + "num_tokens": 189665688.0, + "step": 7433 + }, + { + "epoch": 0.8163848012299583, + "grad_norm": 2.098958730697632, + "learning_rate": 1e-06, + "loss": 1.0538, + "mean_token_accuracy": 0.6847726702690125, + "num_tokens": 189693320.0, + "step": 7434 + }, + { + "epoch": 0.8164946189325719, + "grad_norm": 2.0560452938079834, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.71113520860672, + "num_tokens": 189723688.0, + "step": 7435 + }, + { + "epoch": 0.8166044366351856, + "grad_norm": 1.939566969871521, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.6964242458343506, + "num_tokens": 189756072.0, + "step": 7436 + }, + { + "epoch": 0.8167142543377992, + "grad_norm": 2.1138007640838623, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7071422338485718, + "num_tokens": 189780327.0, + "step": 7437 + }, + { + "epoch": 0.816824072040413, + "grad_norm": 1.9911872148513794, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.692313551902771, + "num_tokens": 189810273.0, + "step": 7438 + }, + { + "epoch": 0.8169338897430266, + "grad_norm": 2.3848888874053955, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7232258319854736, + "num_tokens": 189831003.0, + "step": 7439 + }, + { + "epoch": 0.8170437074456403, + "grad_norm": 2.2086915969848633, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7100221514701843, + "num_tokens": 189856901.0, + "step": 7440 + }, + { + "epoch": 0.8171535251482539, + "grad_norm": 1.963051438331604, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6966555118560791, + "num_tokens": 189888499.0, + "step": 7441 + }, + { + "epoch": 0.8172633428508675, + "grad_norm": 2.117189407348633, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7273339629173279, + "num_tokens": 189914406.0, + "step": 7442 + }, + { + "epoch": 0.8173731605534812, + "grad_norm": 2.3676891326904297, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7078864574432373, + "num_tokens": 189935943.0, + "step": 7443 + }, + { + "epoch": 0.8174829782560948, + "grad_norm": 2.5191025733947754, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7125126123428345, + "num_tokens": 189956889.0, + "step": 7444 + }, + { + "epoch": 0.8175927959587086, + "grad_norm": 2.0361061096191406, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.6991775631904602, + "num_tokens": 189986039.0, + "step": 7445 + }, + { + "epoch": 0.8177026136613222, + "grad_norm": 2.2814650535583496, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.708582878112793, + "num_tokens": 190010156.0, + "step": 7446 + }, + { + "epoch": 0.8178124313639359, + "grad_norm": 2.1269686222076416, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7388498783111572, + "num_tokens": 190036818.0, + "step": 7447 + }, + { + "epoch": 0.8179222490665495, + "grad_norm": 2.227205991744995, + "learning_rate": 1e-06, + "loss": 1.032, + "mean_token_accuracy": 0.6810823082923889, + "num_tokens": 190061867.0, + "step": 7448 + }, + { + "epoch": 0.8180320667691632, + "grad_norm": 2.1131985187530518, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7070004940032959, + "num_tokens": 190091255.0, + "step": 7449 + }, + { + "epoch": 0.8181418844717768, + "grad_norm": 2.405909299850464, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6911977529525757, + "num_tokens": 190115114.0, + "step": 7450 + }, + { + "epoch": 0.8182517021743905, + "grad_norm": 2.183574676513672, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7263846397399902, + "num_tokens": 190139711.0, + "step": 7451 + }, + { + "epoch": 0.8183615198770041, + "grad_norm": 1.9126032590866089, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.696138322353363, + "num_tokens": 190171101.0, + "step": 7452 + }, + { + "epoch": 0.8184713375796179, + "grad_norm": 2.252133369445801, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6972000598907471, + "num_tokens": 190197744.0, + "step": 7453 + }, + { + "epoch": 0.8185811552822315, + "grad_norm": 2.2140884399414062, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7354509234428406, + "num_tokens": 190221331.0, + "step": 7454 + }, + { + "epoch": 0.8186909729848452, + "grad_norm": 2.3481791019439697, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7007215023040771, + "num_tokens": 190243378.0, + "step": 7455 + }, + { + "epoch": 0.8188007906874588, + "grad_norm": 2.0235116481781006, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7065235376358032, + "num_tokens": 190271064.0, + "step": 7456 + }, + { + "epoch": 0.8189106083900725, + "grad_norm": 2.1089773178100586, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.688625693321228, + "num_tokens": 190299055.0, + "step": 7457 + }, + { + "epoch": 0.8190204260926861, + "grad_norm": 2.225461483001709, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.7088918089866638, + "num_tokens": 190325786.0, + "step": 7458 + }, + { + "epoch": 0.8191302437952998, + "grad_norm": 2.022948980331421, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.713496208190918, + "num_tokens": 190353783.0, + "step": 7459 + }, + { + "epoch": 0.8192400614979135, + "grad_norm": 2.261737108230591, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7212466597557068, + "num_tokens": 190375119.0, + "step": 7460 + }, + { + "epoch": 0.8193498792005272, + "grad_norm": 2.1847665309906006, + "learning_rate": 1e-06, + "loss": 1.0471, + "mean_token_accuracy": 0.6894838809967041, + "num_tokens": 190399966.0, + "step": 7461 + }, + { + "epoch": 0.8194596969031408, + "grad_norm": 2.25827956199646, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6895627975463867, + "num_tokens": 190425166.0, + "step": 7462 + }, + { + "epoch": 0.8195695146057544, + "grad_norm": 1.984616994857788, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7002578973770142, + "num_tokens": 190454213.0, + "step": 7463 + }, + { + "epoch": 0.8196793323083681, + "grad_norm": 2.2215583324432373, + "learning_rate": 1e-06, + "loss": 1.0704, + "mean_token_accuracy": 0.6743204593658447, + "num_tokens": 190480354.0, + "step": 7464 + }, + { + "epoch": 0.8197891500109817, + "grad_norm": 2.1873252391815186, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7125942707061768, + "num_tokens": 190503141.0, + "step": 7465 + }, + { + "epoch": 0.8198989677135954, + "grad_norm": 2.2266030311584473, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7071948647499084, + "num_tokens": 190526385.0, + "step": 7466 + }, + { + "epoch": 0.8200087854162091, + "grad_norm": 2.377840518951416, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7104909420013428, + "num_tokens": 190548702.0, + "step": 7467 + }, + { + "epoch": 0.8201186031188228, + "grad_norm": 2.2175612449645996, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7169466018676758, + "num_tokens": 190572841.0, + "step": 7468 + }, + { + "epoch": 0.8202284208214364, + "grad_norm": 2.063261032104492, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6958720684051514, + "num_tokens": 190602023.0, + "step": 7469 + }, + { + "epoch": 0.8203382385240501, + "grad_norm": 2.0775129795074463, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.696149468421936, + "num_tokens": 190630510.0, + "step": 7470 + }, + { + "epoch": 0.8204480562266637, + "grad_norm": 2.359896183013916, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7125962972640991, + "num_tokens": 190653610.0, + "step": 7471 + }, + { + "epoch": 0.8205578739292774, + "grad_norm": 2.1355984210968018, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7229138016700745, + "num_tokens": 190679359.0, + "step": 7472 + }, + { + "epoch": 0.820667691631891, + "grad_norm": 2.2703592777252197, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7193560004234314, + "num_tokens": 190702499.0, + "step": 7473 + }, + { + "epoch": 0.8207775093345048, + "grad_norm": 2.3432257175445557, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7073987126350403, + "num_tokens": 190724715.0, + "step": 7474 + }, + { + "epoch": 0.8208873270371184, + "grad_norm": 2.470555305480957, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7337523102760315, + "num_tokens": 190745462.0, + "step": 7475 + }, + { + "epoch": 0.8209971447397321, + "grad_norm": 2.1516757011413574, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.729257345199585, + "num_tokens": 190770446.0, + "step": 7476 + }, + { + "epoch": 0.8211069624423457, + "grad_norm": 2.2383837699890137, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.7028712034225464, + "num_tokens": 190795303.0, + "step": 7477 + }, + { + "epoch": 0.8212167801449594, + "grad_norm": 2.334179639816284, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7003272771835327, + "num_tokens": 190818359.0, + "step": 7478 + }, + { + "epoch": 0.821326597847573, + "grad_norm": 2.0899124145507812, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7027076482772827, + "num_tokens": 190846343.0, + "step": 7479 + }, + { + "epoch": 0.8214364155501866, + "grad_norm": 2.063791036605835, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.6834884881973267, + "num_tokens": 190876062.0, + "step": 7480 + }, + { + "epoch": 0.8215462332528004, + "grad_norm": 1.9755771160125732, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6991056203842163, + "num_tokens": 190905424.0, + "step": 7481 + }, + { + "epoch": 0.821656050955414, + "grad_norm": 2.0752344131469727, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7103667855262756, + "num_tokens": 190932834.0, + "step": 7482 + }, + { + "epoch": 0.8217658686580277, + "grad_norm": 2.1213552951812744, + "learning_rate": 1e-06, + "loss": 1.0527, + "mean_token_accuracy": 0.6803128123283386, + "num_tokens": 190962444.0, + "step": 7483 + }, + { + "epoch": 0.8218756863606413, + "grad_norm": 2.1435298919677734, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7024081945419312, + "num_tokens": 190988259.0, + "step": 7484 + }, + { + "epoch": 0.821985504063255, + "grad_norm": 2.1446497440338135, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7069183588027954, + "num_tokens": 191015158.0, + "step": 7485 + }, + { + "epoch": 0.8220953217658686, + "grad_norm": 2.1083719730377197, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6948114633560181, + "num_tokens": 191040774.0, + "step": 7486 + }, + { + "epoch": 0.8222051394684823, + "grad_norm": 2.5107967853546143, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.718246340751648, + "num_tokens": 191061232.0, + "step": 7487 + }, + { + "epoch": 0.8223149571710959, + "grad_norm": 2.313263416290283, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7351272106170654, + "num_tokens": 191084431.0, + "step": 7488 + }, + { + "epoch": 0.8224247748737097, + "grad_norm": 2.2963626384735107, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.7014486193656921, + "num_tokens": 191109465.0, + "step": 7489 + }, + { + "epoch": 0.8225345925763233, + "grad_norm": 1.9211794137954712, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6916230916976929, + "num_tokens": 191140759.0, + "step": 7490 + }, + { + "epoch": 0.822644410278937, + "grad_norm": 2.038818359375, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7250219583511353, + "num_tokens": 191169781.0, + "step": 7491 + }, + { + "epoch": 0.8227542279815506, + "grad_norm": 2.24872088432312, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.7010902762413025, + "num_tokens": 191193303.0, + "step": 7492 + }, + { + "epoch": 0.8228640456841643, + "grad_norm": 2.0929415225982666, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6964071393013, + "num_tokens": 191222703.0, + "step": 7493 + }, + { + "epoch": 0.8229738633867779, + "grad_norm": 2.288675546646118, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7052587866783142, + "num_tokens": 191245926.0, + "step": 7494 + }, + { + "epoch": 0.8230836810893916, + "grad_norm": 2.2749738693237305, + "learning_rate": 1e-06, + "loss": 1.0509, + "mean_token_accuracy": 0.6878125667572021, + "num_tokens": 191271376.0, + "step": 7495 + }, + { + "epoch": 0.8231934987920053, + "grad_norm": 2.1393465995788574, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.700893759727478, + "num_tokens": 191296866.0, + "step": 7496 + }, + { + "epoch": 0.823303316494619, + "grad_norm": 2.2866106033325195, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7079607248306274, + "num_tokens": 191318725.0, + "step": 7497 + }, + { + "epoch": 0.8234131341972326, + "grad_norm": 2.1949408054351807, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7136608362197876, + "num_tokens": 191344506.0, + "step": 7498 + }, + { + "epoch": 0.8235229518998463, + "grad_norm": 2.260620594024658, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7248846888542175, + "num_tokens": 191365579.0, + "step": 7499 + }, + { + "epoch": 0.8236327696024599, + "grad_norm": 2.112010955810547, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7101883888244629, + "num_tokens": 191392400.0, + "step": 7500 + }, + { + "epoch": 0.8237425873050735, + "grad_norm": 2.2534220218658447, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7228737473487854, + "num_tokens": 191414548.0, + "step": 7501 + }, + { + "epoch": 0.8238524050076872, + "grad_norm": 2.3793489933013916, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7308182120323181, + "num_tokens": 191434936.0, + "step": 7502 + }, + { + "epoch": 0.823962222710301, + "grad_norm": 2.1809909343719482, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7045807838439941, + "num_tokens": 191459850.0, + "step": 7503 + }, + { + "epoch": 0.8240720404129146, + "grad_norm": 2.4667301177978516, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7433935403823853, + "num_tokens": 191478884.0, + "step": 7504 + }, + { + "epoch": 0.8241818581155282, + "grad_norm": 2.250866651535034, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7044157981872559, + "num_tokens": 191501677.0, + "step": 7505 + }, + { + "epoch": 0.8242916758181419, + "grad_norm": 2.131502389907837, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7270892262458801, + "num_tokens": 191525668.0, + "step": 7506 + }, + { + "epoch": 0.8244014935207555, + "grad_norm": 2.452913522720337, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7182847857475281, + "num_tokens": 191546144.0, + "step": 7507 + }, + { + "epoch": 0.8245113112233692, + "grad_norm": 2.0641016960144043, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7317043542861938, + "num_tokens": 191573175.0, + "step": 7508 + }, + { + "epoch": 0.8246211289259828, + "grad_norm": 2.2880687713623047, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7076454162597656, + "num_tokens": 191598479.0, + "step": 7509 + }, + { + "epoch": 0.8247309466285966, + "grad_norm": 2.3001046180725098, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7074267864227295, + "num_tokens": 191621791.0, + "step": 7510 + }, + { + "epoch": 0.8248407643312102, + "grad_norm": 2.128875255584717, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.6937142610549927, + "num_tokens": 191649449.0, + "step": 7511 + }, + { + "epoch": 0.8249505820338239, + "grad_norm": 2.2390198707580566, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6936733722686768, + "num_tokens": 191674044.0, + "step": 7512 + }, + { + "epoch": 0.8250603997364375, + "grad_norm": 2.408426284790039, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7332568168640137, + "num_tokens": 191694413.0, + "step": 7513 + }, + { + "epoch": 0.8251702174390512, + "grad_norm": 2.1090753078460693, + "learning_rate": 1e-06, + "loss": 1.0722, + "mean_token_accuracy": 0.6740654706954956, + "num_tokens": 191722502.0, + "step": 7514 + }, + { + "epoch": 0.8252800351416648, + "grad_norm": 2.47255802154541, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7126117944717407, + "num_tokens": 191743041.0, + "step": 7515 + }, + { + "epoch": 0.8253898528442785, + "grad_norm": 2.175344467163086, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7108231782913208, + "num_tokens": 191771855.0, + "step": 7516 + }, + { + "epoch": 0.8254996705468921, + "grad_norm": 2.0428261756896973, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7068803310394287, + "num_tokens": 191799586.0, + "step": 7517 + }, + { + "epoch": 0.8256094882495059, + "grad_norm": 2.1799135208129883, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6960750818252563, + "num_tokens": 191824534.0, + "step": 7518 + }, + { + "epoch": 0.8257193059521195, + "grad_norm": 2.124101161956787, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6925442814826965, + "num_tokens": 191851459.0, + "step": 7519 + }, + { + "epoch": 0.8258291236547332, + "grad_norm": 2.2253360748291016, + "learning_rate": 1e-06, + "loss": 1.036, + "mean_token_accuracy": 0.6938751935958862, + "num_tokens": 191875395.0, + "step": 7520 + }, + { + "epoch": 0.8259389413573468, + "grad_norm": 2.131044387817383, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7056816816329956, + "num_tokens": 191902101.0, + "step": 7521 + }, + { + "epoch": 0.8260487590599604, + "grad_norm": 2.3227462768554688, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6936980485916138, + "num_tokens": 191926151.0, + "step": 7522 + }, + { + "epoch": 0.8261585767625741, + "grad_norm": 2.2253713607788086, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7112299203872681, + "num_tokens": 191949803.0, + "step": 7523 + }, + { + "epoch": 0.8262683944651877, + "grad_norm": 2.2287633419036865, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7293063402175903, + "num_tokens": 191973109.0, + "step": 7524 + }, + { + "epoch": 0.8263782121678015, + "grad_norm": 2.364798069000244, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7291173934936523, + "num_tokens": 191994462.0, + "step": 7525 + }, + { + "epoch": 0.8264880298704151, + "grad_norm": 2.192941665649414, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7096669673919678, + "num_tokens": 192020209.0, + "step": 7526 + }, + { + "epoch": 0.8265978475730288, + "grad_norm": 2.1508545875549316, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.705105185508728, + "num_tokens": 192047414.0, + "step": 7527 + }, + { + "epoch": 0.8267076652756424, + "grad_norm": 2.326226234436035, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7314039468765259, + "num_tokens": 192072441.0, + "step": 7528 + }, + { + "epoch": 0.8268174829782561, + "grad_norm": 1.8572442531585693, + "learning_rate": 1e-06, + "loss": 1.0585, + "mean_token_accuracy": 0.6803194880485535, + "num_tokens": 192108276.0, + "step": 7529 + }, + { + "epoch": 0.8269273006808697, + "grad_norm": 2.2432398796081543, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.6998749375343323, + "num_tokens": 192131852.0, + "step": 7530 + }, + { + "epoch": 0.8270371183834834, + "grad_norm": 1.9488264322280884, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7059693336486816, + "num_tokens": 192163262.0, + "step": 7531 + }, + { + "epoch": 0.8271469360860971, + "grad_norm": 2.046329975128174, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.716403067111969, + "num_tokens": 192189817.0, + "step": 7532 + }, + { + "epoch": 0.8272567537887108, + "grad_norm": 2.170135498046875, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6945439577102661, + "num_tokens": 192216312.0, + "step": 7533 + }, + { + "epoch": 0.8273665714913244, + "grad_norm": 2.164942979812622, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.6978937387466431, + "num_tokens": 192241658.0, + "step": 7534 + }, + { + "epoch": 0.8274763891939381, + "grad_norm": 2.6008386611938477, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.722313404083252, + "num_tokens": 192259929.0, + "step": 7535 + }, + { + "epoch": 0.8275862068965517, + "grad_norm": 2.120492696762085, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7161099314689636, + "num_tokens": 192285130.0, + "step": 7536 + }, + { + "epoch": 0.8276960245991654, + "grad_norm": 2.6692230701446533, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7300969362258911, + "num_tokens": 192302567.0, + "step": 7537 + }, + { + "epoch": 0.827805842301779, + "grad_norm": 2.0149590969085693, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.716081976890564, + "num_tokens": 192333411.0, + "step": 7538 + }, + { + "epoch": 0.8279156600043928, + "grad_norm": 2.0118045806884766, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7213590145111084, + "num_tokens": 192360943.0, + "step": 7539 + }, + { + "epoch": 0.8280254777070064, + "grad_norm": 2.0863683223724365, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7025167942047119, + "num_tokens": 192387355.0, + "step": 7540 + }, + { + "epoch": 0.82813529540962, + "grad_norm": 2.5397133827209473, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7223836183547974, + "num_tokens": 192406611.0, + "step": 7541 + }, + { + "epoch": 0.8282451131122337, + "grad_norm": 2.079252243041992, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7214611172676086, + "num_tokens": 192431593.0, + "step": 7542 + }, + { + "epoch": 0.8283549308148473, + "grad_norm": 1.9110667705535889, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7176586389541626, + "num_tokens": 192463032.0, + "step": 7543 + }, + { + "epoch": 0.828464748517461, + "grad_norm": 2.2568817138671875, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7055231332778931, + "num_tokens": 192487753.0, + "step": 7544 + }, + { + "epoch": 0.8285745662200746, + "grad_norm": 2.2150495052337646, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7106829881668091, + "num_tokens": 192511686.0, + "step": 7545 + }, + { + "epoch": 0.8286843839226883, + "grad_norm": 2.2063345909118652, + "learning_rate": 1e-06, + "loss": 1.0364, + "mean_token_accuracy": 0.6879073977470398, + "num_tokens": 192537887.0, + "step": 7546 + }, + { + "epoch": 0.828794201625302, + "grad_norm": 1.9559617042541504, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7204830646514893, + "num_tokens": 192567390.0, + "step": 7547 + }, + { + "epoch": 0.8289040193279157, + "grad_norm": 2.232445478439331, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.6985987424850464, + "num_tokens": 192592520.0, + "step": 7548 + }, + { + "epoch": 0.8290138370305293, + "grad_norm": 1.9794535636901855, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.698636531829834, + "num_tokens": 192622800.0, + "step": 7549 + }, + { + "epoch": 0.829123654733143, + "grad_norm": 2.337568998336792, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.7008423209190369, + "num_tokens": 192645014.0, + "step": 7550 + }, + { + "epoch": 0.8292334724357566, + "grad_norm": 2.2808985710144043, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7148561477661133, + "num_tokens": 192669145.0, + "step": 7551 + }, + { + "epoch": 0.8293432901383703, + "grad_norm": 2.27017879486084, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7067351341247559, + "num_tokens": 192693365.0, + "step": 7552 + }, + { + "epoch": 0.8294531078409839, + "grad_norm": 2.0125036239624023, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7369077801704407, + "num_tokens": 192721503.0, + "step": 7553 + }, + { + "epoch": 0.8295629255435977, + "grad_norm": 1.9835753440856934, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7219454050064087, + "num_tokens": 192752281.0, + "step": 7554 + }, + { + "epoch": 0.8296727432462113, + "grad_norm": 2.0104732513427734, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.700647234916687, + "num_tokens": 192782734.0, + "step": 7555 + }, + { + "epoch": 0.829782560948825, + "grad_norm": 2.319854497909546, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.7498406171798706, + "num_tokens": 192804581.0, + "step": 7556 + }, + { + "epoch": 0.8298923786514386, + "grad_norm": 1.9475520849227905, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.6977829933166504, + "num_tokens": 192837674.0, + "step": 7557 + }, + { + "epoch": 0.8300021963540523, + "grad_norm": 2.5121989250183105, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7286519408226013, + "num_tokens": 192857306.0, + "step": 7558 + }, + { + "epoch": 0.8301120140566659, + "grad_norm": 2.1153650283813477, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7000183463096619, + "num_tokens": 192885070.0, + "step": 7559 + }, + { + "epoch": 0.8302218317592795, + "grad_norm": 2.0596227645874023, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7105851173400879, + "num_tokens": 192914973.0, + "step": 7560 + }, + { + "epoch": 0.8303316494618933, + "grad_norm": 2.364426851272583, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7145920395851135, + "num_tokens": 192938079.0, + "step": 7561 + }, + { + "epoch": 0.830441467164507, + "grad_norm": 2.3768796920776367, + "learning_rate": 1e-06, + "loss": 1.0713, + "mean_token_accuracy": 0.6785808205604553, + "num_tokens": 192961945.0, + "step": 7562 + }, + { + "epoch": 0.8305512848671206, + "grad_norm": 2.1013340950012207, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6892094612121582, + "num_tokens": 192990370.0, + "step": 7563 + }, + { + "epoch": 0.8306611025697342, + "grad_norm": 2.231320858001709, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7072936296463013, + "num_tokens": 193013219.0, + "step": 7564 + }, + { + "epoch": 0.8307709202723479, + "grad_norm": 2.1464881896972656, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7193433046340942, + "num_tokens": 193038783.0, + "step": 7565 + }, + { + "epoch": 0.8308807379749615, + "grad_norm": 2.2521700859069824, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7085000276565552, + "num_tokens": 193062554.0, + "step": 7566 + }, + { + "epoch": 0.8309905556775752, + "grad_norm": 1.9209022521972656, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6897907853126526, + "num_tokens": 193095462.0, + "step": 7567 + }, + { + "epoch": 0.8311003733801889, + "grad_norm": 2.2635796070098877, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7454872131347656, + "num_tokens": 193118394.0, + "step": 7568 + }, + { + "epoch": 0.8312101910828026, + "grad_norm": 2.1420626640319824, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6901720762252808, + "num_tokens": 193145406.0, + "step": 7569 + }, + { + "epoch": 0.8313200087854162, + "grad_norm": 2.5467541217803955, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.719159722328186, + "num_tokens": 193163070.0, + "step": 7570 + }, + { + "epoch": 0.8314298264880299, + "grad_norm": 2.0564708709716797, + "learning_rate": 1e-06, + "loss": 1.0462, + "mean_token_accuracy": 0.6895797848701477, + "num_tokens": 193192584.0, + "step": 7571 + }, + { + "epoch": 0.8315396441906435, + "grad_norm": 2.1581883430480957, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.6924156546592712, + "num_tokens": 193217597.0, + "step": 7572 + }, + { + "epoch": 0.8316494618932572, + "grad_norm": 2.340496778488159, + "learning_rate": 1e-06, + "loss": 1.0564, + "mean_token_accuracy": 0.6924721002578735, + "num_tokens": 193242160.0, + "step": 7573 + }, + { + "epoch": 0.8317592795958708, + "grad_norm": 2.231539011001587, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7173409461975098, + "num_tokens": 193266999.0, + "step": 7574 + }, + { + "epoch": 0.8318690972984845, + "grad_norm": 2.162184238433838, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6839048862457275, + "num_tokens": 193294349.0, + "step": 7575 + }, + { + "epoch": 0.8319789150010982, + "grad_norm": 2.220813751220703, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7164740562438965, + "num_tokens": 193320894.0, + "step": 7576 + }, + { + "epoch": 0.8320887327037119, + "grad_norm": 2.2193970680236816, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.697679877281189, + "num_tokens": 193345004.0, + "step": 7577 + }, + { + "epoch": 0.8321985504063255, + "grad_norm": 2.1684134006500244, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.7035731673240662, + "num_tokens": 193371360.0, + "step": 7578 + }, + { + "epoch": 0.8323083681089392, + "grad_norm": 2.145559310913086, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7135555744171143, + "num_tokens": 193397752.0, + "step": 7579 + }, + { + "epoch": 0.8324181858115528, + "grad_norm": 2.0929787158966064, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7163420915603638, + "num_tokens": 193423007.0, + "step": 7580 + }, + { + "epoch": 0.8325280035141664, + "grad_norm": 2.3108978271484375, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7091792821884155, + "num_tokens": 193444627.0, + "step": 7581 + }, + { + "epoch": 0.8326378212167801, + "grad_norm": 2.2064692974090576, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.7017777562141418, + "num_tokens": 193470166.0, + "step": 7582 + }, + { + "epoch": 0.8327476389193939, + "grad_norm": 2.333516836166382, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7270717024803162, + "num_tokens": 193492047.0, + "step": 7583 + }, + { + "epoch": 0.8328574566220075, + "grad_norm": 1.9898923635482788, + "learning_rate": 1e-06, + "loss": 1.1047, + "mean_token_accuracy": 0.6678321957588196, + "num_tokens": 193523528.0, + "step": 7584 + }, + { + "epoch": 0.8329672743246211, + "grad_norm": 2.521944999694824, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7291977405548096, + "num_tokens": 193543545.0, + "step": 7585 + }, + { + "epoch": 0.8330770920272348, + "grad_norm": 2.207893133163452, + "learning_rate": 1e-06, + "loss": 1.0238, + "mean_token_accuracy": 0.687302827835083, + "num_tokens": 193567429.0, + "step": 7586 + }, + { + "epoch": 0.8331869097298484, + "grad_norm": 2.154134750366211, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6901478171348572, + "num_tokens": 193593676.0, + "step": 7587 + }, + { + "epoch": 0.8332967274324621, + "grad_norm": 2.011101245880127, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.694644570350647, + "num_tokens": 193622155.0, + "step": 7588 + }, + { + "epoch": 0.8334065451350757, + "grad_norm": 2.0483992099761963, + "learning_rate": 1e-06, + "loss": 1.0118, + "mean_token_accuracy": 0.6978950500488281, + "num_tokens": 193650493.0, + "step": 7589 + }, + { + "epoch": 0.8335163628376895, + "grad_norm": 2.037527322769165, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.704915463924408, + "num_tokens": 193678931.0, + "step": 7590 + }, + { + "epoch": 0.8336261805403031, + "grad_norm": 2.193984270095825, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6969902515411377, + "num_tokens": 193704744.0, + "step": 7591 + }, + { + "epoch": 0.8337359982429168, + "grad_norm": 2.052111864089966, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.7027405500411987, + "num_tokens": 193732910.0, + "step": 7592 + }, + { + "epoch": 0.8338458159455304, + "grad_norm": 2.2372334003448486, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7085339426994324, + "num_tokens": 193758183.0, + "step": 7593 + }, + { + "epoch": 0.8339556336481441, + "grad_norm": 2.195584535598755, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.722439706325531, + "num_tokens": 193781195.0, + "step": 7594 + }, + { + "epoch": 0.8340654513507577, + "grad_norm": 2.195657253265381, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.7092512249946594, + "num_tokens": 193805233.0, + "step": 7595 + }, + { + "epoch": 0.8341752690533714, + "grad_norm": 1.9827104806900024, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7088097929954529, + "num_tokens": 193833887.0, + "step": 7596 + }, + { + "epoch": 0.8342850867559851, + "grad_norm": 1.8647229671478271, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.7085666060447693, + "num_tokens": 193865786.0, + "step": 7597 + }, + { + "epoch": 0.8343949044585988, + "grad_norm": 2.0508615970611572, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6964795589447021, + "num_tokens": 193894334.0, + "step": 7598 + }, + { + "epoch": 0.8345047221612124, + "grad_norm": 2.5084784030914307, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7318320274353027, + "num_tokens": 193914299.0, + "step": 7599 + }, + { + "epoch": 0.834614539863826, + "grad_norm": 2.126171827316284, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7060282230377197, + "num_tokens": 193941367.0, + "step": 7600 + }, + { + "epoch": 0.8347243575664397, + "grad_norm": 2.2301230430603027, + "learning_rate": 1e-06, + "loss": 1.0531, + "mean_token_accuracy": 0.6889939904212952, + "num_tokens": 193966593.0, + "step": 7601 + }, + { + "epoch": 0.8348341752690533, + "grad_norm": 2.2776377201080322, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7222573757171631, + "num_tokens": 193989699.0, + "step": 7602 + }, + { + "epoch": 0.834943992971667, + "grad_norm": 2.2493104934692383, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7253715991973877, + "num_tokens": 194013667.0, + "step": 7603 + }, + { + "epoch": 0.8350538106742806, + "grad_norm": 2.023739814758301, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.6978287100791931, + "num_tokens": 194042651.0, + "step": 7604 + }, + { + "epoch": 0.8351636283768944, + "grad_norm": 2.212017774581909, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7096124887466431, + "num_tokens": 194066682.0, + "step": 7605 + }, + { + "epoch": 0.835273446079508, + "grad_norm": 2.1332035064697266, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.704644501209259, + "num_tokens": 194091820.0, + "step": 7606 + }, + { + "epoch": 0.8353832637821217, + "grad_norm": 2.162219762802124, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6878327131271362, + "num_tokens": 194120147.0, + "step": 7607 + }, + { + "epoch": 0.8354930814847353, + "grad_norm": 2.262798547744751, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.7026035189628601, + "num_tokens": 194145282.0, + "step": 7608 + }, + { + "epoch": 0.835602899187349, + "grad_norm": 2.198955774307251, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.6931567192077637, + "num_tokens": 194170746.0, + "step": 7609 + }, + { + "epoch": 0.8357127168899626, + "grad_norm": 2.1571643352508545, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7090296149253845, + "num_tokens": 194197455.0, + "step": 7610 + }, + { + "epoch": 0.8358225345925763, + "grad_norm": 2.009037733078003, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7138680219650269, + "num_tokens": 194224846.0, + "step": 7611 + }, + { + "epoch": 0.83593235229519, + "grad_norm": 2.4432191848754883, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.6998686194419861, + "num_tokens": 194248326.0, + "step": 7612 + }, + { + "epoch": 0.8360421699978037, + "grad_norm": 2.2445921897888184, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.69134521484375, + "num_tokens": 194273611.0, + "step": 7613 + }, + { + "epoch": 0.8361519877004173, + "grad_norm": 2.511845588684082, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7109410166740417, + "num_tokens": 194294498.0, + "step": 7614 + }, + { + "epoch": 0.836261805403031, + "grad_norm": 2.105412244796753, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7072595953941345, + "num_tokens": 194322071.0, + "step": 7615 + }, + { + "epoch": 0.8363716231056446, + "grad_norm": 2.018162727355957, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7172095775604248, + "num_tokens": 194351626.0, + "step": 7616 + }, + { + "epoch": 0.8364814408082583, + "grad_norm": 2.279628276824951, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7076247930526733, + "num_tokens": 194376506.0, + "step": 7617 + }, + { + "epoch": 0.8365912585108719, + "grad_norm": 2.264801263809204, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7485862970352173, + "num_tokens": 194398708.0, + "step": 7618 + }, + { + "epoch": 0.8367010762134857, + "grad_norm": 2.273320436477661, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7234443426132202, + "num_tokens": 194421651.0, + "step": 7619 + }, + { + "epoch": 0.8368108939160993, + "grad_norm": 2.142775297164917, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7163739204406738, + "num_tokens": 194447823.0, + "step": 7620 + }, + { + "epoch": 0.836920711618713, + "grad_norm": 2.4460790157318115, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.7013362646102905, + "num_tokens": 194470641.0, + "step": 7621 + }, + { + "epoch": 0.8370305293213266, + "grad_norm": 2.154771089553833, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7153252363204956, + "num_tokens": 194496199.0, + "step": 7622 + }, + { + "epoch": 0.8371403470239402, + "grad_norm": 2.3539891242980957, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.6980685591697693, + "num_tokens": 194519335.0, + "step": 7623 + }, + { + "epoch": 0.8372501647265539, + "grad_norm": 2.5463547706604004, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7224010229110718, + "num_tokens": 194539916.0, + "step": 7624 + }, + { + "epoch": 0.8373599824291675, + "grad_norm": 2.3862390518188477, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7231903672218323, + "num_tokens": 194562408.0, + "step": 7625 + }, + { + "epoch": 0.8374698001317813, + "grad_norm": 2.260589361190796, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7314156293869019, + "num_tokens": 194585968.0, + "step": 7626 + }, + { + "epoch": 0.8375796178343949, + "grad_norm": 2.4089162349700928, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7340266108512878, + "num_tokens": 194605608.0, + "step": 7627 + }, + { + "epoch": 0.8376894355370086, + "grad_norm": 2.2258331775665283, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6957579851150513, + "num_tokens": 194630255.0, + "step": 7628 + }, + { + "epoch": 0.8377992532396222, + "grad_norm": 2.0081753730773926, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7396405935287476, + "num_tokens": 194659070.0, + "step": 7629 + }, + { + "epoch": 0.8379090709422359, + "grad_norm": 2.3384315967559814, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.717837929725647, + "num_tokens": 194682088.0, + "step": 7630 + }, + { + "epoch": 0.8380188886448495, + "grad_norm": 2.203674077987671, + "learning_rate": 1e-06, + "loss": 1.0829, + "mean_token_accuracy": 0.6714000701904297, + "num_tokens": 194708725.0, + "step": 7631 + }, + { + "epoch": 0.8381287063474632, + "grad_norm": 2.2209222316741943, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7256736755371094, + "num_tokens": 194732566.0, + "step": 7632 + }, + { + "epoch": 0.8382385240500769, + "grad_norm": 2.08332896232605, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.718895673751831, + "num_tokens": 194760027.0, + "step": 7633 + }, + { + "epoch": 0.8383483417526906, + "grad_norm": 2.346229314804077, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7083756923675537, + "num_tokens": 194783674.0, + "step": 7634 + }, + { + "epoch": 0.8384581594553042, + "grad_norm": 2.2228810787200928, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7079463005065918, + "num_tokens": 194809806.0, + "step": 7635 + }, + { + "epoch": 0.8385679771579179, + "grad_norm": 1.977087378501892, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7080036997795105, + "num_tokens": 194840777.0, + "step": 7636 + }, + { + "epoch": 0.8386777948605315, + "grad_norm": 2.004124402999878, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.7062352299690247, + "num_tokens": 194870590.0, + "step": 7637 + }, + { + "epoch": 0.8387876125631452, + "grad_norm": 2.0101850032806396, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7158827781677246, + "num_tokens": 194900086.0, + "step": 7638 + }, + { + "epoch": 0.8388974302657588, + "grad_norm": 2.1595306396484375, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7200977802276611, + "num_tokens": 194925299.0, + "step": 7639 + }, + { + "epoch": 0.8390072479683724, + "grad_norm": 2.2758395671844482, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6996581554412842, + "num_tokens": 194948640.0, + "step": 7640 + }, + { + "epoch": 0.8391170656709862, + "grad_norm": 2.1699764728546143, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7355480194091797, + "num_tokens": 194974572.0, + "step": 7641 + }, + { + "epoch": 0.8392268833735999, + "grad_norm": 2.1623501777648926, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7158405184745789, + "num_tokens": 195002491.0, + "step": 7642 + }, + { + "epoch": 0.8393367010762135, + "grad_norm": 2.230269432067871, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7061468362808228, + "num_tokens": 195026099.0, + "step": 7643 + }, + { + "epoch": 0.8394465187788271, + "grad_norm": 2.0857927799224854, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6998906135559082, + "num_tokens": 195052775.0, + "step": 7644 + }, + { + "epoch": 0.8395563364814408, + "grad_norm": 2.0590929985046387, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7111873626708984, + "num_tokens": 195081544.0, + "step": 7645 + }, + { + "epoch": 0.8396661541840544, + "grad_norm": 2.301982879638672, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.70900559425354, + "num_tokens": 195104432.0, + "step": 7646 + }, + { + "epoch": 0.8397759718866681, + "grad_norm": 2.0537002086639404, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.7008002400398254, + "num_tokens": 195133405.0, + "step": 7647 + }, + { + "epoch": 0.8398857895892818, + "grad_norm": 2.209139823913574, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7072018384933472, + "num_tokens": 195159942.0, + "step": 7648 + }, + { + "epoch": 0.8399956072918955, + "grad_norm": 2.001359701156616, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.6959282159805298, + "num_tokens": 195189972.0, + "step": 7649 + }, + { + "epoch": 0.8401054249945091, + "grad_norm": 2.3623735904693604, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7117508053779602, + "num_tokens": 195212608.0, + "step": 7650 + }, + { + "epoch": 0.8402152426971228, + "grad_norm": 2.2409920692443848, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7158645987510681, + "num_tokens": 195238689.0, + "step": 7651 + }, + { + "epoch": 0.8403250603997364, + "grad_norm": 2.1793835163116455, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.6918338537216187, + "num_tokens": 195265913.0, + "step": 7652 + }, + { + "epoch": 0.8404348781023501, + "grad_norm": 2.260103464126587, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.6996452212333679, + "num_tokens": 195288463.0, + "step": 7653 + }, + { + "epoch": 0.8405446958049637, + "grad_norm": 2.0168821811676025, + "learning_rate": 1e-06, + "loss": 1.0114, + "mean_token_accuracy": 0.6932013630867004, + "num_tokens": 195316087.0, + "step": 7654 + }, + { + "epoch": 0.8406545135075775, + "grad_norm": 2.2700486183166504, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7486380338668823, + "num_tokens": 195338520.0, + "step": 7655 + }, + { + "epoch": 0.8407643312101911, + "grad_norm": 2.549757480621338, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.701222836971283, + "num_tokens": 195359636.0, + "step": 7656 + }, + { + "epoch": 0.8408741489128048, + "grad_norm": 2.180652141571045, + "learning_rate": 1e-06, + "loss": 1.0425, + "mean_token_accuracy": 0.687350869178772, + "num_tokens": 195385121.0, + "step": 7657 + }, + { + "epoch": 0.8409839666154184, + "grad_norm": 2.2051477432250977, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7227023839950562, + "num_tokens": 195407964.0, + "step": 7658 + }, + { + "epoch": 0.8410937843180321, + "grad_norm": 2.2297849655151367, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6915684938430786, + "num_tokens": 195435295.0, + "step": 7659 + }, + { + "epoch": 0.8412036020206457, + "grad_norm": 2.138166904449463, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6922356486320496, + "num_tokens": 195462724.0, + "step": 7660 + }, + { + "epoch": 0.8413134197232593, + "grad_norm": 2.154977321624756, + "learning_rate": 1e-06, + "loss": 1.0566, + "mean_token_accuracy": 0.6826776266098022, + "num_tokens": 195488992.0, + "step": 7661 + }, + { + "epoch": 0.8414232374258731, + "grad_norm": 2.5729639530181885, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7172945141792297, + "num_tokens": 195507895.0, + "step": 7662 + }, + { + "epoch": 0.8415330551284868, + "grad_norm": 2.1406123638153076, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7192680835723877, + "num_tokens": 195536089.0, + "step": 7663 + }, + { + "epoch": 0.8416428728311004, + "grad_norm": 2.288381576538086, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7120400071144104, + "num_tokens": 195558160.0, + "step": 7664 + }, + { + "epoch": 0.841752690533714, + "grad_norm": 2.1151647567749023, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7094985246658325, + "num_tokens": 195585234.0, + "step": 7665 + }, + { + "epoch": 0.8418625082363277, + "grad_norm": 2.1655421257019043, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6978490948677063, + "num_tokens": 195610997.0, + "step": 7666 + }, + { + "epoch": 0.8419723259389413, + "grad_norm": 2.2147817611694336, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7191767692565918, + "num_tokens": 195635313.0, + "step": 7667 + }, + { + "epoch": 0.842082143641555, + "grad_norm": 2.4168004989624023, + "learning_rate": 1e-06, + "loss": 1.0321, + "mean_token_accuracy": 0.6929093599319458, + "num_tokens": 195658673.0, + "step": 7668 + }, + { + "epoch": 0.8421919613441686, + "grad_norm": 1.9759514331817627, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.703961968421936, + "num_tokens": 195687708.0, + "step": 7669 + }, + { + "epoch": 0.8423017790467824, + "grad_norm": 2.3634033203125, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7083730697631836, + "num_tokens": 195711068.0, + "step": 7670 + }, + { + "epoch": 0.842411596749396, + "grad_norm": 2.132065773010254, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7194986343383789, + "num_tokens": 195737520.0, + "step": 7671 + }, + { + "epoch": 0.8425214144520097, + "grad_norm": 2.3340749740600586, + "learning_rate": 1e-06, + "loss": 1.0228, + "mean_token_accuracy": 0.6972640156745911, + "num_tokens": 195762210.0, + "step": 7672 + }, + { + "epoch": 0.8426312321546233, + "grad_norm": 2.140958070755005, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.6965070366859436, + "num_tokens": 195790118.0, + "step": 7673 + }, + { + "epoch": 0.842741049857237, + "grad_norm": 2.4363439083099365, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7216453552246094, + "num_tokens": 195813329.0, + "step": 7674 + }, + { + "epoch": 0.8428508675598506, + "grad_norm": 2.143001079559326, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7387682199478149, + "num_tokens": 195838020.0, + "step": 7675 + }, + { + "epoch": 0.8429606852624643, + "grad_norm": 2.2593276500701904, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.693930983543396, + "num_tokens": 195863636.0, + "step": 7676 + }, + { + "epoch": 0.843070502965078, + "grad_norm": 2.4580869674682617, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6914441585540771, + "num_tokens": 195885035.0, + "step": 7677 + }, + { + "epoch": 0.8431803206676917, + "grad_norm": 2.1606531143188477, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7173652648925781, + "num_tokens": 195909285.0, + "step": 7678 + }, + { + "epoch": 0.8432901383703053, + "grad_norm": 2.085603952407837, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7110321521759033, + "num_tokens": 195935055.0, + "step": 7679 + }, + { + "epoch": 0.843399956072919, + "grad_norm": 2.0756542682647705, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7243154644966125, + "num_tokens": 195960318.0, + "step": 7680 + }, + { + "epoch": 0.8435097737755326, + "grad_norm": 2.386528730392456, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7071417570114136, + "num_tokens": 195980490.0, + "step": 7681 + }, + { + "epoch": 0.8436195914781462, + "grad_norm": 2.5282158851623535, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.7537533044815063, + "num_tokens": 195997734.0, + "step": 7682 + }, + { + "epoch": 0.8437294091807599, + "grad_norm": 2.261084794998169, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7257078886032104, + "num_tokens": 196020055.0, + "step": 7683 + }, + { + "epoch": 0.8438392268833736, + "grad_norm": 1.964889645576477, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7080082893371582, + "num_tokens": 196050519.0, + "step": 7684 + }, + { + "epoch": 0.8439490445859873, + "grad_norm": 2.326960325241089, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.6959593296051025, + "num_tokens": 196074700.0, + "step": 7685 + }, + { + "epoch": 0.8440588622886009, + "grad_norm": 2.4871749877929688, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7055327892303467, + "num_tokens": 196095157.0, + "step": 7686 + }, + { + "epoch": 0.8441686799912146, + "grad_norm": 2.102327823638916, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6946314573287964, + "num_tokens": 196122092.0, + "step": 7687 + }, + { + "epoch": 0.8442784976938282, + "grad_norm": 2.290066957473755, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7091021537780762, + "num_tokens": 196145925.0, + "step": 7688 + }, + { + "epoch": 0.8443883153964419, + "grad_norm": 2.2019970417022705, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.732251763343811, + "num_tokens": 196171443.0, + "step": 7689 + }, + { + "epoch": 0.8444981330990555, + "grad_norm": 2.4085617065429688, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7109822034835815, + "num_tokens": 196191737.0, + "step": 7690 + }, + { + "epoch": 0.8446079508016693, + "grad_norm": 2.3172554969787598, + "learning_rate": 1e-06, + "loss": 1.0313, + "mean_token_accuracy": 0.6815239787101746, + "num_tokens": 196215796.0, + "step": 7691 + }, + { + "epoch": 0.8447177685042829, + "grad_norm": 1.889582872390747, + "learning_rate": 1e-06, + "loss": 1.0195, + "mean_token_accuracy": 0.6934642791748047, + "num_tokens": 196249030.0, + "step": 7692 + }, + { + "epoch": 0.8448275862068966, + "grad_norm": 2.013241767883301, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.698269248008728, + "num_tokens": 196279065.0, + "step": 7693 + }, + { + "epoch": 0.8449374039095102, + "grad_norm": 2.058182716369629, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7225723266601562, + "num_tokens": 196306862.0, + "step": 7694 + }, + { + "epoch": 0.8450472216121239, + "grad_norm": 2.6095495223999023, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7278031706809998, + "num_tokens": 196324283.0, + "step": 7695 + }, + { + "epoch": 0.8451570393147375, + "grad_norm": 2.100468397140503, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7319318056106567, + "num_tokens": 196349856.0, + "step": 7696 + }, + { + "epoch": 0.8452668570173512, + "grad_norm": 2.37274169921875, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7173600196838379, + "num_tokens": 196372483.0, + "step": 7697 + }, + { + "epoch": 0.8453766747199648, + "grad_norm": 2.1821248531341553, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7146431803703308, + "num_tokens": 196398557.0, + "step": 7698 + }, + { + "epoch": 0.8454864924225786, + "grad_norm": 2.054140090942383, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.685883641242981, + "num_tokens": 196429341.0, + "step": 7699 + }, + { + "epoch": 0.8455963101251922, + "grad_norm": 2.099980115890503, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7074770331382751, + "num_tokens": 196455462.0, + "step": 7700 + }, + { + "epoch": 0.8457061278278059, + "grad_norm": 2.0278189182281494, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.6993758082389832, + "num_tokens": 196484440.0, + "step": 7701 + }, + { + "epoch": 0.8458159455304195, + "grad_norm": 2.3193061351776123, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7238738536834717, + "num_tokens": 196508253.0, + "step": 7702 + }, + { + "epoch": 0.8459257632330331, + "grad_norm": 2.1196837425231934, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7220597267150879, + "num_tokens": 196534019.0, + "step": 7703 + }, + { + "epoch": 0.8460355809356468, + "grad_norm": 2.322291135787964, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.7001559138298035, + "num_tokens": 196557556.0, + "step": 7704 + }, + { + "epoch": 0.8461453986382604, + "grad_norm": 1.8411284685134888, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7065064311027527, + "num_tokens": 196591183.0, + "step": 7705 + }, + { + "epoch": 0.8462552163408742, + "grad_norm": 1.759995698928833, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6925148963928223, + "num_tokens": 196628782.0, + "step": 7706 + }, + { + "epoch": 0.8463650340434878, + "grad_norm": 2.1747777462005615, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7058627605438232, + "num_tokens": 196651948.0, + "step": 7707 + }, + { + "epoch": 0.8464748517461015, + "grad_norm": 1.961585283279419, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7083548903465271, + "num_tokens": 196680893.0, + "step": 7708 + }, + { + "epoch": 0.8465846694487151, + "grad_norm": 2.2646074295043945, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7174696326255798, + "num_tokens": 196703572.0, + "step": 7709 + }, + { + "epoch": 0.8466944871513288, + "grad_norm": 2.177083969116211, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7087782621383667, + "num_tokens": 196729520.0, + "step": 7710 + }, + { + "epoch": 0.8468043048539424, + "grad_norm": 2.18727970123291, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7153905630111694, + "num_tokens": 196754183.0, + "step": 7711 + }, + { + "epoch": 0.8469141225565561, + "grad_norm": 1.8484678268432617, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.698137640953064, + "num_tokens": 196789357.0, + "step": 7712 + }, + { + "epoch": 0.8470239402591698, + "grad_norm": 2.2199196815490723, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6997398734092712, + "num_tokens": 196814601.0, + "step": 7713 + }, + { + "epoch": 0.8471337579617835, + "grad_norm": 2.148388624191284, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.717781662940979, + "num_tokens": 196839519.0, + "step": 7714 + }, + { + "epoch": 0.8472435756643971, + "grad_norm": 2.2670185565948486, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6930751800537109, + "num_tokens": 196865052.0, + "step": 7715 + }, + { + "epoch": 0.8473533933670108, + "grad_norm": 1.9933592081069946, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.7066267728805542, + "num_tokens": 196894700.0, + "step": 7716 + }, + { + "epoch": 0.8474632110696244, + "grad_norm": 2.330106735229492, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6972532272338867, + "num_tokens": 196919941.0, + "step": 7717 + }, + { + "epoch": 0.8475730287722381, + "grad_norm": 2.1216609477996826, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.6877725720405579, + "num_tokens": 196946645.0, + "step": 7718 + }, + { + "epoch": 0.8476828464748517, + "grad_norm": 2.0217130184173584, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7232834696769714, + "num_tokens": 196974862.0, + "step": 7719 + }, + { + "epoch": 0.8477926641774655, + "grad_norm": 2.0801517963409424, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7222626209259033, + "num_tokens": 197000678.0, + "step": 7720 + }, + { + "epoch": 0.8479024818800791, + "grad_norm": 2.006345510482788, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.6897492408752441, + "num_tokens": 197031369.0, + "step": 7721 + }, + { + "epoch": 0.8480122995826928, + "grad_norm": 2.2876410484313965, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.722856879234314, + "num_tokens": 197052704.0, + "step": 7722 + }, + { + "epoch": 0.8481221172853064, + "grad_norm": 2.2067301273345947, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6928690671920776, + "num_tokens": 197079675.0, + "step": 7723 + }, + { + "epoch": 0.84823193498792, + "grad_norm": 2.0773942470550537, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.7022162079811096, + "num_tokens": 197105357.0, + "step": 7724 + }, + { + "epoch": 0.8483417526905337, + "grad_norm": 2.1575095653533936, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7023322582244873, + "num_tokens": 197131746.0, + "step": 7725 + }, + { + "epoch": 0.8484515703931473, + "grad_norm": 2.3555877208709717, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.701612651348114, + "num_tokens": 197154817.0, + "step": 7726 + }, + { + "epoch": 0.848561388095761, + "grad_norm": 2.1470932960510254, + "learning_rate": 1e-06, + "loss": 1.049, + "mean_token_accuracy": 0.6805106401443481, + "num_tokens": 197183057.0, + "step": 7727 + }, + { + "epoch": 0.8486712057983747, + "grad_norm": 2.049445867538452, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7033370137214661, + "num_tokens": 197212748.0, + "step": 7728 + }, + { + "epoch": 0.8487810235009884, + "grad_norm": 2.1638436317443848, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.6998869776725769, + "num_tokens": 197239373.0, + "step": 7729 + }, + { + "epoch": 0.848890841203602, + "grad_norm": 2.029217004776001, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7116686701774597, + "num_tokens": 197268041.0, + "step": 7730 + }, + { + "epoch": 0.8490006589062157, + "grad_norm": 2.042269468307495, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6880989074707031, + "num_tokens": 197297284.0, + "step": 7731 + }, + { + "epoch": 0.8491104766088293, + "grad_norm": 2.242222547531128, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7206713557243347, + "num_tokens": 197319971.0, + "step": 7732 + }, + { + "epoch": 0.849220294311443, + "grad_norm": 2.5066847801208496, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7061545848846436, + "num_tokens": 197340201.0, + "step": 7733 + }, + { + "epoch": 0.8493301120140566, + "grad_norm": 2.6228771209716797, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7181993722915649, + "num_tokens": 197357680.0, + "step": 7734 + }, + { + "epoch": 0.8494399297166704, + "grad_norm": 2.146472930908203, + "learning_rate": 1e-06, + "loss": 1.0769, + "mean_token_accuracy": 0.6798784732818604, + "num_tokens": 197385168.0, + "step": 7735 + }, + { + "epoch": 0.849549747419284, + "grad_norm": 2.102145195007324, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.6971025466918945, + "num_tokens": 197412123.0, + "step": 7736 + }, + { + "epoch": 0.8496595651218977, + "grad_norm": 2.1777830123901367, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.6940509080886841, + "num_tokens": 197437894.0, + "step": 7737 + }, + { + "epoch": 0.8497693828245113, + "grad_norm": 2.4469642639160156, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7323173880577087, + "num_tokens": 197457322.0, + "step": 7738 + }, + { + "epoch": 0.849879200527125, + "grad_norm": 2.253288745880127, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7381353974342346, + "num_tokens": 197479501.0, + "step": 7739 + }, + { + "epoch": 0.8499890182297386, + "grad_norm": 1.9997048377990723, + "learning_rate": 1e-06, + "loss": 1.0365, + "mean_token_accuracy": 0.695391833782196, + "num_tokens": 197508957.0, + "step": 7740 + }, + { + "epoch": 0.8500988359323522, + "grad_norm": 2.3158299922943115, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7053394317626953, + "num_tokens": 197532123.0, + "step": 7741 + }, + { + "epoch": 0.850208653634966, + "grad_norm": 2.2366437911987305, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7302922010421753, + "num_tokens": 197553647.0, + "step": 7742 + }, + { + "epoch": 0.8503184713375797, + "grad_norm": 1.994059681892395, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.718592643737793, + "num_tokens": 197582758.0, + "step": 7743 + }, + { + "epoch": 0.8504282890401933, + "grad_norm": 2.006655216217041, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7264897227287292, + "num_tokens": 197613750.0, + "step": 7744 + }, + { + "epoch": 0.8505381067428069, + "grad_norm": 2.316716432571411, + "learning_rate": 1e-06, + "loss": 1.0603, + "mean_token_accuracy": 0.6861318945884705, + "num_tokens": 197637704.0, + "step": 7745 + }, + { + "epoch": 0.8506479244454206, + "grad_norm": 1.900192379951477, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6905441284179688, + "num_tokens": 197669052.0, + "step": 7746 + }, + { + "epoch": 0.8507577421480342, + "grad_norm": 2.3903558254241943, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7060615420341492, + "num_tokens": 197689929.0, + "step": 7747 + }, + { + "epoch": 0.8508675598506479, + "grad_norm": 2.2321736812591553, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7285012006759644, + "num_tokens": 197712233.0, + "step": 7748 + }, + { + "epoch": 0.8509773775532616, + "grad_norm": 2.555513620376587, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.712217390537262, + "num_tokens": 197731160.0, + "step": 7749 + }, + { + "epoch": 0.8510871952558753, + "grad_norm": 1.9669291973114014, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7121248841285706, + "num_tokens": 197760867.0, + "step": 7750 + }, + { + "epoch": 0.8511970129584889, + "grad_norm": 2.1902430057525635, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6942557096481323, + "num_tokens": 197788470.0, + "step": 7751 + }, + { + "epoch": 0.8513068306611026, + "grad_norm": 2.2657129764556885, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7002460360527039, + "num_tokens": 197812516.0, + "step": 7752 + }, + { + "epoch": 0.8514166483637162, + "grad_norm": 1.9173532724380493, + "learning_rate": 1e-06, + "loss": 1.0481, + "mean_token_accuracy": 0.6810286045074463, + "num_tokens": 197845880.0, + "step": 7753 + }, + { + "epoch": 0.8515264660663299, + "grad_norm": 2.210402250289917, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.704112708568573, + "num_tokens": 197871725.0, + "step": 7754 + }, + { + "epoch": 0.8516362837689435, + "grad_norm": 2.5972819328308105, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7280409336090088, + "num_tokens": 197890371.0, + "step": 7755 + }, + { + "epoch": 0.8517461014715572, + "grad_norm": 2.3750171661376953, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7224037647247314, + "num_tokens": 197911131.0, + "step": 7756 + }, + { + "epoch": 0.8518559191741709, + "grad_norm": 2.4037716388702393, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7244308590888977, + "num_tokens": 197936228.0, + "step": 7757 + }, + { + "epoch": 0.8519657368767846, + "grad_norm": 2.145460605621338, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7033767104148865, + "num_tokens": 197964427.0, + "step": 7758 + }, + { + "epoch": 0.8520755545793982, + "grad_norm": 2.5039873123168945, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.6972580552101135, + "num_tokens": 197986642.0, + "step": 7759 + }, + { + "epoch": 0.8521853722820119, + "grad_norm": 2.1692936420440674, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7032244801521301, + "num_tokens": 198016383.0, + "step": 7760 + }, + { + "epoch": 0.8522951899846255, + "grad_norm": 2.1685664653778076, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6905769109725952, + "num_tokens": 198042773.0, + "step": 7761 + }, + { + "epoch": 0.8524050076872391, + "grad_norm": 2.2722361087799072, + "learning_rate": 1e-06, + "loss": 1.0493, + "mean_token_accuracy": 0.676295280456543, + "num_tokens": 198068228.0, + "step": 7762 + }, + { + "epoch": 0.8525148253898528, + "grad_norm": 2.1352698802948, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7067873477935791, + "num_tokens": 198092786.0, + "step": 7763 + }, + { + "epoch": 0.8526246430924665, + "grad_norm": 2.2008392810821533, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.700869083404541, + "num_tokens": 198117369.0, + "step": 7764 + }, + { + "epoch": 0.8527344607950802, + "grad_norm": 2.1734886169433594, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7047801613807678, + "num_tokens": 198142199.0, + "step": 7765 + }, + { + "epoch": 0.8528442784976938, + "grad_norm": 2.247004270553589, + "learning_rate": 1e-06, + "loss": 1.0476, + "mean_token_accuracy": 0.6847249269485474, + "num_tokens": 198168125.0, + "step": 7766 + }, + { + "epoch": 0.8529540962003075, + "grad_norm": 2.507373332977295, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6953307390213013, + "num_tokens": 198189433.0, + "step": 7767 + }, + { + "epoch": 0.8530639139029211, + "grad_norm": 2.2240936756134033, + "learning_rate": 1e-06, + "loss": 1.0537, + "mean_token_accuracy": 0.675855815410614, + "num_tokens": 198217787.0, + "step": 7768 + }, + { + "epoch": 0.8531737316055348, + "grad_norm": 2.601346969604492, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7173928022384644, + "num_tokens": 198239449.0, + "step": 7769 + }, + { + "epoch": 0.8532835493081484, + "grad_norm": 2.209785223007202, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7203918695449829, + "num_tokens": 198263601.0, + "step": 7770 + }, + { + "epoch": 0.8533933670107622, + "grad_norm": 2.2599287033081055, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7199777364730835, + "num_tokens": 198286648.0, + "step": 7771 + }, + { + "epoch": 0.8535031847133758, + "grad_norm": 2.0508909225463867, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7099241614341736, + "num_tokens": 198316676.0, + "step": 7772 + }, + { + "epoch": 0.8536130024159895, + "grad_norm": 2.071620225906372, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7337076663970947, + "num_tokens": 198343710.0, + "step": 7773 + }, + { + "epoch": 0.8537228201186031, + "grad_norm": 2.305168867111206, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.7119224667549133, + "num_tokens": 198367825.0, + "step": 7774 + }, + { + "epoch": 0.8538326378212168, + "grad_norm": 2.0263993740081787, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7227067947387695, + "num_tokens": 198396994.0, + "step": 7775 + }, + { + "epoch": 0.8539424555238304, + "grad_norm": 2.2824866771698, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.698035717010498, + "num_tokens": 198420533.0, + "step": 7776 + }, + { + "epoch": 0.8540522732264441, + "grad_norm": 2.1806812286376953, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.714181661605835, + "num_tokens": 198445972.0, + "step": 7777 + }, + { + "epoch": 0.8541620909290578, + "grad_norm": 2.221743583679199, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7464126348495483, + "num_tokens": 198467598.0, + "step": 7778 + }, + { + "epoch": 0.8542719086316715, + "grad_norm": 2.241445541381836, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.6933715343475342, + "num_tokens": 198493084.0, + "step": 7779 + }, + { + "epoch": 0.8543817263342851, + "grad_norm": 1.9385403394699097, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.70289146900177, + "num_tokens": 198524918.0, + "step": 7780 + }, + { + "epoch": 0.8544915440368988, + "grad_norm": 2.1113364696502686, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6989200711250305, + "num_tokens": 198552792.0, + "step": 7781 + }, + { + "epoch": 0.8546013617395124, + "grad_norm": 1.988990306854248, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.6984130144119263, + "num_tokens": 198580002.0, + "step": 7782 + }, + { + "epoch": 0.854711179442126, + "grad_norm": 2.0047173500061035, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7134103178977966, + "num_tokens": 198607274.0, + "step": 7783 + }, + { + "epoch": 0.8548209971447397, + "grad_norm": 2.0555338859558105, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6909979581832886, + "num_tokens": 198635169.0, + "step": 7784 + }, + { + "epoch": 0.8549308148473534, + "grad_norm": 2.332637071609497, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7212420105934143, + "num_tokens": 198655849.0, + "step": 7785 + }, + { + "epoch": 0.8550406325499671, + "grad_norm": 2.5036346912384033, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.701208770275116, + "num_tokens": 198677254.0, + "step": 7786 + }, + { + "epoch": 0.8551504502525807, + "grad_norm": 2.1037960052490234, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.7054516673088074, + "num_tokens": 198703409.0, + "step": 7787 + }, + { + "epoch": 0.8552602679551944, + "grad_norm": 2.001744508743286, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7065180540084839, + "num_tokens": 198733632.0, + "step": 7788 + }, + { + "epoch": 0.855370085657808, + "grad_norm": 2.110236167907715, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7121044397354126, + "num_tokens": 198760006.0, + "step": 7789 + }, + { + "epoch": 0.8554799033604217, + "grad_norm": 2.0974435806274414, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7098353505134583, + "num_tokens": 198785594.0, + "step": 7790 + }, + { + "epoch": 0.8555897210630353, + "grad_norm": 2.1122329235076904, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7173429131507874, + "num_tokens": 198809283.0, + "step": 7791 + }, + { + "epoch": 0.855699538765649, + "grad_norm": 2.3562684059143066, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.721149206161499, + "num_tokens": 198831310.0, + "step": 7792 + }, + { + "epoch": 0.8558093564682627, + "grad_norm": 2.094521999359131, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7060933113098145, + "num_tokens": 198857971.0, + "step": 7793 + }, + { + "epoch": 0.8559191741708764, + "grad_norm": 2.1663615703582764, + "learning_rate": 1e-06, + "loss": 1.0339, + "mean_token_accuracy": 0.6910233497619629, + "num_tokens": 198881645.0, + "step": 7794 + }, + { + "epoch": 0.85602899187349, + "grad_norm": 1.887758493423462, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.7091621160507202, + "num_tokens": 198914187.0, + "step": 7795 + }, + { + "epoch": 0.8561388095761037, + "grad_norm": 2.3929483890533447, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6982915997505188, + "num_tokens": 198937522.0, + "step": 7796 + }, + { + "epoch": 0.8562486272787173, + "grad_norm": 2.6442840099334717, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.7010865211486816, + "num_tokens": 198956409.0, + "step": 7797 + }, + { + "epoch": 0.856358444981331, + "grad_norm": 2.4129533767700195, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7196696996688843, + "num_tokens": 198977075.0, + "step": 7798 + }, + { + "epoch": 0.8564682626839446, + "grad_norm": 2.610872745513916, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7212576270103455, + "num_tokens": 198995411.0, + "step": 7799 + }, + { + "epoch": 0.8565780803865584, + "grad_norm": 2.5504205226898193, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7303968667984009, + "num_tokens": 199012848.0, + "step": 7800 + }, + { + "epoch": 0.856687898089172, + "grad_norm": 2.3614888191223145, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7239820957183838, + "num_tokens": 199035230.0, + "step": 7801 + }, + { + "epoch": 0.8567977157917857, + "grad_norm": 2.0748302936553955, + "learning_rate": 1e-06, + "loss": 1.0536, + "mean_token_accuracy": 0.6804789304733276, + "num_tokens": 199063676.0, + "step": 7802 + }, + { + "epoch": 0.8569075334943993, + "grad_norm": 2.36429500579834, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7095216512680054, + "num_tokens": 199085977.0, + "step": 7803 + }, + { + "epoch": 0.8570173511970129, + "grad_norm": 2.162161111831665, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7153680324554443, + "num_tokens": 199112643.0, + "step": 7804 + }, + { + "epoch": 0.8571271688996266, + "grad_norm": 2.2392210960388184, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.691423237323761, + "num_tokens": 199136978.0, + "step": 7805 + }, + { + "epoch": 0.8572369866022402, + "grad_norm": 1.9972246885299683, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.704479455947876, + "num_tokens": 199165601.0, + "step": 7806 + }, + { + "epoch": 0.857346804304854, + "grad_norm": 2.1081008911132812, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7003510594367981, + "num_tokens": 199191517.0, + "step": 7807 + }, + { + "epoch": 0.8574566220074676, + "grad_norm": 2.1482696533203125, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.702281653881073, + "num_tokens": 199218954.0, + "step": 7808 + }, + { + "epoch": 0.8575664397100813, + "grad_norm": 1.9746503829956055, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6954178810119629, + "num_tokens": 199249487.0, + "step": 7809 + }, + { + "epoch": 0.8576762574126949, + "grad_norm": 2.09122896194458, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7241602540016174, + "num_tokens": 199275942.0, + "step": 7810 + }, + { + "epoch": 0.8577860751153086, + "grad_norm": 2.366631269454956, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7147207260131836, + "num_tokens": 199297521.0, + "step": 7811 + }, + { + "epoch": 0.8578958928179222, + "grad_norm": 2.2593753337860107, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7121192216873169, + "num_tokens": 199321416.0, + "step": 7812 + }, + { + "epoch": 0.8580057105205359, + "grad_norm": 2.2724404335021973, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.722100019454956, + "num_tokens": 199344247.0, + "step": 7813 + }, + { + "epoch": 0.8581155282231496, + "grad_norm": 2.2284276485443115, + "learning_rate": 1e-06, + "loss": 1.0421, + "mean_token_accuracy": 0.6878300905227661, + "num_tokens": 199372640.0, + "step": 7814 + }, + { + "epoch": 0.8582253459257633, + "grad_norm": 2.2677435874938965, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.6971851587295532, + "num_tokens": 199395911.0, + "step": 7815 + }, + { + "epoch": 0.8583351636283769, + "grad_norm": 2.0152435302734375, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.6990189552307129, + "num_tokens": 199426833.0, + "step": 7816 + }, + { + "epoch": 0.8584449813309906, + "grad_norm": 2.129082441329956, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.699620246887207, + "num_tokens": 199451977.0, + "step": 7817 + }, + { + "epoch": 0.8585547990336042, + "grad_norm": 2.0731167793273926, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.746147096157074, + "num_tokens": 199476846.0, + "step": 7818 + }, + { + "epoch": 0.8586646167362179, + "grad_norm": 2.1996638774871826, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7145774364471436, + "num_tokens": 199502943.0, + "step": 7819 + }, + { + "epoch": 0.8587744344388315, + "grad_norm": 2.1301016807556152, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7166975736618042, + "num_tokens": 199529778.0, + "step": 7820 + }, + { + "epoch": 0.8588842521414451, + "grad_norm": 2.3457448482513428, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7027072906494141, + "num_tokens": 199552189.0, + "step": 7821 + }, + { + "epoch": 0.8589940698440589, + "grad_norm": 2.1033995151519775, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.685515284538269, + "num_tokens": 199579056.0, + "step": 7822 + }, + { + "epoch": 0.8591038875466726, + "grad_norm": 2.169862985610962, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7176047563552856, + "num_tokens": 199604069.0, + "step": 7823 + }, + { + "epoch": 0.8592137052492862, + "grad_norm": 2.2634761333465576, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7007448673248291, + "num_tokens": 199627035.0, + "step": 7824 + }, + { + "epoch": 0.8593235229518998, + "grad_norm": 2.1428921222686768, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.706255316734314, + "num_tokens": 199652403.0, + "step": 7825 + }, + { + "epoch": 0.8594333406545135, + "grad_norm": 2.256608247756958, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7000691890716553, + "num_tokens": 199677372.0, + "step": 7826 + }, + { + "epoch": 0.8595431583571271, + "grad_norm": 1.991821527481079, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7158594727516174, + "num_tokens": 199704755.0, + "step": 7827 + }, + { + "epoch": 0.8596529760597408, + "grad_norm": 2.047067165374756, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7089754343032837, + "num_tokens": 199730823.0, + "step": 7828 + }, + { + "epoch": 0.8597627937623545, + "grad_norm": 2.351381301879883, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.747025728225708, + "num_tokens": 199751191.0, + "step": 7829 + }, + { + "epoch": 0.8598726114649682, + "grad_norm": 2.257265567779541, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7061927318572998, + "num_tokens": 199775376.0, + "step": 7830 + }, + { + "epoch": 0.8599824291675818, + "grad_norm": 2.110294818878174, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7226996421813965, + "num_tokens": 199801175.0, + "step": 7831 + }, + { + "epoch": 0.8600922468701955, + "grad_norm": 2.018451690673828, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.692430317401886, + "num_tokens": 199831108.0, + "step": 7832 + }, + { + "epoch": 0.8602020645728091, + "grad_norm": 2.1126112937927246, + "learning_rate": 1e-06, + "loss": 1.036, + "mean_token_accuracy": 0.6900829076766968, + "num_tokens": 199858156.0, + "step": 7833 + }, + { + "epoch": 0.8603118822754228, + "grad_norm": 2.2888214588165283, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7061775922775269, + "num_tokens": 199881608.0, + "step": 7834 + }, + { + "epoch": 0.8604216999780364, + "grad_norm": 2.232861042022705, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7009718418121338, + "num_tokens": 199905238.0, + "step": 7835 + }, + { + "epoch": 0.8605315176806502, + "grad_norm": 2.1283843517303467, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7076889276504517, + "num_tokens": 199931314.0, + "step": 7836 + }, + { + "epoch": 0.8606413353832638, + "grad_norm": 2.177374839782715, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.7037111520767212, + "num_tokens": 199956962.0, + "step": 7837 + }, + { + "epoch": 0.8607511530858775, + "grad_norm": 2.1686747074127197, + "learning_rate": 1e-06, + "loss": 1.0657, + "mean_token_accuracy": 0.6762937307357788, + "num_tokens": 199986338.0, + "step": 7838 + }, + { + "epoch": 0.8608609707884911, + "grad_norm": 1.9900197982788086, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7013692259788513, + "num_tokens": 200016835.0, + "step": 7839 + }, + { + "epoch": 0.8609707884911048, + "grad_norm": 1.933593988418579, + "learning_rate": 1e-06, + "loss": 1.0625, + "mean_token_accuracy": 0.677888035774231, + "num_tokens": 200049425.0, + "step": 7840 + }, + { + "epoch": 0.8610806061937184, + "grad_norm": 2.0643720626831055, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.6941717267036438, + "num_tokens": 200078048.0, + "step": 7841 + }, + { + "epoch": 0.861190423896332, + "grad_norm": 2.1553964614868164, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.712033748626709, + "num_tokens": 200105545.0, + "step": 7842 + }, + { + "epoch": 0.8613002415989458, + "grad_norm": 2.369295835494995, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7394664883613586, + "num_tokens": 200126366.0, + "step": 7843 + }, + { + "epoch": 0.8614100593015594, + "grad_norm": 1.9922500848770142, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7017136812210083, + "num_tokens": 200154580.0, + "step": 7844 + }, + { + "epoch": 0.8615198770041731, + "grad_norm": 2.326422691345215, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7115215063095093, + "num_tokens": 200176632.0, + "step": 7845 + }, + { + "epoch": 0.8616296947067867, + "grad_norm": 2.372300386428833, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7202199101448059, + "num_tokens": 200197959.0, + "step": 7846 + }, + { + "epoch": 0.8617395124094004, + "grad_norm": 2.4613325595855713, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7140347957611084, + "num_tokens": 200218389.0, + "step": 7847 + }, + { + "epoch": 0.861849330112014, + "grad_norm": 2.153907299041748, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.6983672976493835, + "num_tokens": 200245735.0, + "step": 7848 + }, + { + "epoch": 0.8619591478146277, + "grad_norm": 2.370378255844116, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7097734808921814, + "num_tokens": 200267560.0, + "step": 7849 + }, + { + "epoch": 0.8620689655172413, + "grad_norm": 1.952285885810852, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.7007207870483398, + "num_tokens": 200298833.0, + "step": 7850 + }, + { + "epoch": 0.8621787832198551, + "grad_norm": 1.928061842918396, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.6978914141654968, + "num_tokens": 200329746.0, + "step": 7851 + }, + { + "epoch": 0.8622886009224687, + "grad_norm": 2.3807146549224854, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7193895578384399, + "num_tokens": 200351614.0, + "step": 7852 + }, + { + "epoch": 0.8623984186250824, + "grad_norm": 2.239736318588257, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7213091254234314, + "num_tokens": 200375033.0, + "step": 7853 + }, + { + "epoch": 0.862508236327696, + "grad_norm": 2.204725503921509, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7064546346664429, + "num_tokens": 200400111.0, + "step": 7854 + }, + { + "epoch": 0.8626180540303097, + "grad_norm": 2.3221678733825684, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7243585586547852, + "num_tokens": 200422021.0, + "step": 7855 + }, + { + "epoch": 0.8627278717329233, + "grad_norm": 2.2881836891174316, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.713948130607605, + "num_tokens": 200444523.0, + "step": 7856 + }, + { + "epoch": 0.862837689435537, + "grad_norm": 2.1761488914489746, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7108843326568604, + "num_tokens": 200470586.0, + "step": 7857 + }, + { + "epoch": 0.8629475071381507, + "grad_norm": 2.1906962394714355, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7095925211906433, + "num_tokens": 200494342.0, + "step": 7858 + }, + { + "epoch": 0.8630573248407644, + "grad_norm": 2.4903340339660645, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7093877792358398, + "num_tokens": 200516665.0, + "step": 7859 + }, + { + "epoch": 0.863167142543378, + "grad_norm": 1.9470120668411255, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7155938148498535, + "num_tokens": 200544138.0, + "step": 7860 + }, + { + "epoch": 0.8632769602459917, + "grad_norm": 2.4562838077545166, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7293232083320618, + "num_tokens": 200564636.0, + "step": 7861 + }, + { + "epoch": 0.8633867779486053, + "grad_norm": 2.5057597160339355, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.6991932392120361, + "num_tokens": 200585597.0, + "step": 7862 + }, + { + "epoch": 0.863496595651219, + "grad_norm": 2.412980079650879, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.725483775138855, + "num_tokens": 200607880.0, + "step": 7863 + }, + { + "epoch": 0.8636064133538326, + "grad_norm": 2.334742307662964, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7278419733047485, + "num_tokens": 200629713.0, + "step": 7864 + }, + { + "epoch": 0.8637162310564463, + "grad_norm": 2.166109561920166, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.6947644948959351, + "num_tokens": 200655712.0, + "step": 7865 + }, + { + "epoch": 0.86382604875906, + "grad_norm": 2.150578737258911, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.695196270942688, + "num_tokens": 200682190.0, + "step": 7866 + }, + { + "epoch": 0.8639358664616736, + "grad_norm": 2.233783006668091, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.71809321641922, + "num_tokens": 200706035.0, + "step": 7867 + }, + { + "epoch": 0.8640456841642873, + "grad_norm": 2.724971294403076, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.6947512030601501, + "num_tokens": 200723380.0, + "step": 7868 + }, + { + "epoch": 0.8641555018669009, + "grad_norm": 2.211895227432251, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.6944448351860046, + "num_tokens": 200750405.0, + "step": 7869 + }, + { + "epoch": 0.8642653195695146, + "grad_norm": 2.157210111618042, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6896675825119019, + "num_tokens": 200777015.0, + "step": 7870 + }, + { + "epoch": 0.8643751372721282, + "grad_norm": 2.1220571994781494, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7323969602584839, + "num_tokens": 200802219.0, + "step": 7871 + }, + { + "epoch": 0.864484954974742, + "grad_norm": 2.1669020652770996, + "learning_rate": 1e-06, + "loss": 0.7704, + "mean_token_accuracy": 0.7469149827957153, + "num_tokens": 200824972.0, + "step": 7872 + }, + { + "epoch": 0.8645947726773556, + "grad_norm": 2.08019757270813, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.6935701370239258, + "num_tokens": 200852286.0, + "step": 7873 + }, + { + "epoch": 0.8647045903799693, + "grad_norm": 2.0333409309387207, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7216597199440002, + "num_tokens": 200878181.0, + "step": 7874 + }, + { + "epoch": 0.8648144080825829, + "grad_norm": 2.4615163803100586, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6969479918479919, + "num_tokens": 200897817.0, + "step": 7875 + }, + { + "epoch": 0.8649242257851966, + "grad_norm": 2.2837514877319336, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7129555940628052, + "num_tokens": 200921051.0, + "step": 7876 + }, + { + "epoch": 0.8650340434878102, + "grad_norm": 2.1616363525390625, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7253565192222595, + "num_tokens": 200945961.0, + "step": 7877 + }, + { + "epoch": 0.8651438611904239, + "grad_norm": 2.2314698696136475, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7273374795913696, + "num_tokens": 200969133.0, + "step": 7878 + }, + { + "epoch": 0.8652536788930375, + "grad_norm": 2.210142135620117, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.6867222189903259, + "num_tokens": 200995998.0, + "step": 7879 + }, + { + "epoch": 0.8653634965956513, + "grad_norm": 2.130028247833252, + "learning_rate": 1e-06, + "loss": 1.012, + "mean_token_accuracy": 0.6911063194274902, + "num_tokens": 201024624.0, + "step": 7880 + }, + { + "epoch": 0.8654733142982649, + "grad_norm": 2.251756191253662, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7031601071357727, + "num_tokens": 201048776.0, + "step": 7881 + }, + { + "epoch": 0.8655831320008786, + "grad_norm": 2.049809455871582, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.705049455165863, + "num_tokens": 201077123.0, + "step": 7882 + }, + { + "epoch": 0.8656929497034922, + "grad_norm": 2.147214651107788, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.7000768780708313, + "num_tokens": 201103478.0, + "step": 7883 + }, + { + "epoch": 0.8658027674061058, + "grad_norm": 2.495359182357788, + "learning_rate": 1e-06, + "loss": 1.0398, + "mean_token_accuracy": 0.6834635734558105, + "num_tokens": 201124483.0, + "step": 7884 + }, + { + "epoch": 0.8659125851087195, + "grad_norm": 1.877842903137207, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7088254690170288, + "num_tokens": 201156402.0, + "step": 7885 + }, + { + "epoch": 0.8660224028113331, + "grad_norm": 2.050291061401367, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6952838897705078, + "num_tokens": 201185229.0, + "step": 7886 + }, + { + "epoch": 0.8661322205139469, + "grad_norm": 2.110435724258423, + "learning_rate": 1e-06, + "loss": 1.0561, + "mean_token_accuracy": 0.6795121431350708, + "num_tokens": 201214856.0, + "step": 7887 + }, + { + "epoch": 0.8662420382165605, + "grad_norm": 1.9752190113067627, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.6986508369445801, + "num_tokens": 201247070.0, + "step": 7888 + }, + { + "epoch": 0.8663518559191742, + "grad_norm": 1.9528038501739502, + "learning_rate": 1e-06, + "loss": 1.0867, + "mean_token_accuracy": 0.6778383255004883, + "num_tokens": 201277557.0, + "step": 7889 + }, + { + "epoch": 0.8664616736217878, + "grad_norm": 2.2414252758026123, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7281879186630249, + "num_tokens": 201300396.0, + "step": 7890 + }, + { + "epoch": 0.8665714913244015, + "grad_norm": 2.25799822807312, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6948837041854858, + "num_tokens": 201325384.0, + "step": 7891 + }, + { + "epoch": 0.8666813090270151, + "grad_norm": 1.9731942415237427, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.729939341545105, + "num_tokens": 201353339.0, + "step": 7892 + }, + { + "epoch": 0.8667911267296288, + "grad_norm": 2.210448741912842, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7109742760658264, + "num_tokens": 201376369.0, + "step": 7893 + }, + { + "epoch": 0.8669009444322425, + "grad_norm": 2.047168731689453, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.7018083930015564, + "num_tokens": 201406495.0, + "step": 7894 + }, + { + "epoch": 0.8670107621348562, + "grad_norm": 2.262244939804077, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7241637706756592, + "num_tokens": 201430705.0, + "step": 7895 + }, + { + "epoch": 0.8671205798374698, + "grad_norm": 2.25311279296875, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6947285532951355, + "num_tokens": 201455334.0, + "step": 7896 + }, + { + "epoch": 0.8672303975400835, + "grad_norm": 1.976812720298767, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6944063901901245, + "num_tokens": 201485789.0, + "step": 7897 + }, + { + "epoch": 0.8673402152426971, + "grad_norm": 1.9741859436035156, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6968631744384766, + "num_tokens": 201519967.0, + "step": 7898 + }, + { + "epoch": 0.8674500329453108, + "grad_norm": 2.2693541049957275, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.6814660429954529, + "num_tokens": 201544815.0, + "step": 7899 + }, + { + "epoch": 0.8675598506479244, + "grad_norm": 2.193316698074341, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7175503373146057, + "num_tokens": 201569174.0, + "step": 7900 + }, + { + "epoch": 0.8676696683505382, + "grad_norm": 2.269806385040283, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7202938199043274, + "num_tokens": 201593498.0, + "step": 7901 + }, + { + "epoch": 0.8677794860531518, + "grad_norm": 2.362652540206909, + "learning_rate": 1e-06, + "loss": 1.0326, + "mean_token_accuracy": 0.7000084519386292, + "num_tokens": 201616275.0, + "step": 7902 + }, + { + "epoch": 0.8678893037557655, + "grad_norm": 2.2625560760498047, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7230719923973083, + "num_tokens": 201639709.0, + "step": 7903 + }, + { + "epoch": 0.8679991214583791, + "grad_norm": 1.937464714050293, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.6948651075363159, + "num_tokens": 201671245.0, + "step": 7904 + }, + { + "epoch": 0.8681089391609927, + "grad_norm": 2.3453497886657715, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.6939038038253784, + "num_tokens": 201695077.0, + "step": 7905 + }, + { + "epoch": 0.8682187568636064, + "grad_norm": 2.3879647254943848, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.7004984617233276, + "num_tokens": 201717525.0, + "step": 7906 + }, + { + "epoch": 0.86832857456622, + "grad_norm": 2.0157968997955322, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.687389612197876, + "num_tokens": 201747663.0, + "step": 7907 + }, + { + "epoch": 0.8684383922688337, + "grad_norm": 1.8673814535140991, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7115805745124817, + "num_tokens": 201779756.0, + "step": 7908 + }, + { + "epoch": 0.8685482099714474, + "grad_norm": 2.1684460639953613, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6901285648345947, + "num_tokens": 201805540.0, + "step": 7909 + }, + { + "epoch": 0.8686580276740611, + "grad_norm": 2.6767938137054443, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.7049725651741028, + "num_tokens": 201826087.0, + "step": 7910 + }, + { + "epoch": 0.8687678453766747, + "grad_norm": 2.0998876094818115, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7154169678688049, + "num_tokens": 201851934.0, + "step": 7911 + }, + { + "epoch": 0.8688776630792884, + "grad_norm": 2.441528797149658, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7117379903793335, + "num_tokens": 201872563.0, + "step": 7912 + }, + { + "epoch": 0.868987480781902, + "grad_norm": 1.975342869758606, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.7009268403053284, + "num_tokens": 201902971.0, + "step": 7913 + }, + { + "epoch": 0.8690972984845157, + "grad_norm": 2.266695499420166, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7066924571990967, + "num_tokens": 201927757.0, + "step": 7914 + }, + { + "epoch": 0.8692071161871293, + "grad_norm": 2.3077542781829834, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7247339487075806, + "num_tokens": 201952467.0, + "step": 7915 + }, + { + "epoch": 0.8693169338897431, + "grad_norm": 1.970302939414978, + "learning_rate": 1e-06, + "loss": 1.0749, + "mean_token_accuracy": 0.6737773418426514, + "num_tokens": 201983343.0, + "step": 7916 + }, + { + "epoch": 0.8694267515923567, + "grad_norm": 2.41807222366333, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.7019875049591064, + "num_tokens": 202008313.0, + "step": 7917 + }, + { + "epoch": 0.8695365692949704, + "grad_norm": 1.9523608684539795, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.6982380151748657, + "num_tokens": 202039464.0, + "step": 7918 + }, + { + "epoch": 0.869646386997584, + "grad_norm": 2.112077236175537, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.7054691314697266, + "num_tokens": 202067294.0, + "step": 7919 + }, + { + "epoch": 0.8697562047001977, + "grad_norm": 2.2029190063476562, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6940513849258423, + "num_tokens": 202093860.0, + "step": 7920 + }, + { + "epoch": 0.8698660224028113, + "grad_norm": 2.0762245655059814, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7270869016647339, + "num_tokens": 202120334.0, + "step": 7921 + }, + { + "epoch": 0.869975840105425, + "grad_norm": 2.0766026973724365, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.7032304406166077, + "num_tokens": 202147613.0, + "step": 7922 + }, + { + "epoch": 0.8700856578080387, + "grad_norm": 2.2553136348724365, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7118722200393677, + "num_tokens": 202170838.0, + "step": 7923 + }, + { + "epoch": 0.8701954755106523, + "grad_norm": 2.34086537361145, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7214025855064392, + "num_tokens": 202192679.0, + "step": 7924 + }, + { + "epoch": 0.870305293213266, + "grad_norm": 2.101001739501953, + "learning_rate": 1e-06, + "loss": 1.0569, + "mean_token_accuracy": 0.6862114667892456, + "num_tokens": 202219486.0, + "step": 7925 + }, + { + "epoch": 0.8704151109158796, + "grad_norm": 2.3461062908172607, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7271394729614258, + "num_tokens": 202243073.0, + "step": 7926 + }, + { + "epoch": 0.8705249286184933, + "grad_norm": 2.3808696269989014, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.7029719352722168, + "num_tokens": 202267193.0, + "step": 7927 + }, + { + "epoch": 0.8706347463211069, + "grad_norm": 1.992277979850769, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6968734264373779, + "num_tokens": 202296659.0, + "step": 7928 + }, + { + "epoch": 0.8707445640237206, + "grad_norm": 2.31917667388916, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7059164047241211, + "num_tokens": 202318102.0, + "step": 7929 + }, + { + "epoch": 0.8708543817263343, + "grad_norm": 2.3267226219177246, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7222285270690918, + "num_tokens": 202341230.0, + "step": 7930 + }, + { + "epoch": 0.870964199428948, + "grad_norm": 2.0643672943115234, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7109466195106506, + "num_tokens": 202368624.0, + "step": 7931 + }, + { + "epoch": 0.8710740171315616, + "grad_norm": 2.4650557041168213, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7213543653488159, + "num_tokens": 202388640.0, + "step": 7932 + }, + { + "epoch": 0.8711838348341753, + "grad_norm": 2.006955623626709, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7098337411880493, + "num_tokens": 202421460.0, + "step": 7933 + }, + { + "epoch": 0.8712936525367889, + "grad_norm": 2.373164653778076, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.729975700378418, + "num_tokens": 202442419.0, + "step": 7934 + }, + { + "epoch": 0.8714034702394026, + "grad_norm": 2.141042947769165, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7157401442527771, + "num_tokens": 202469542.0, + "step": 7935 + }, + { + "epoch": 0.8715132879420162, + "grad_norm": 2.055309772491455, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6974619030952454, + "num_tokens": 202499653.0, + "step": 7936 + }, + { + "epoch": 0.87162310564463, + "grad_norm": 2.375532627105713, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7141023874282837, + "num_tokens": 202521773.0, + "step": 7937 + }, + { + "epoch": 0.8717329233472436, + "grad_norm": 2.5712790489196777, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7224460244178772, + "num_tokens": 202541122.0, + "step": 7938 + }, + { + "epoch": 0.8718427410498573, + "grad_norm": 1.9813008308410645, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7170594930648804, + "num_tokens": 202570825.0, + "step": 7939 + }, + { + "epoch": 0.8719525587524709, + "grad_norm": 2.1177942752838135, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7226395606994629, + "num_tokens": 202596204.0, + "step": 7940 + }, + { + "epoch": 0.8720623764550846, + "grad_norm": 2.1939773559570312, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.6966198682785034, + "num_tokens": 202622090.0, + "step": 7941 + }, + { + "epoch": 0.8721721941576982, + "grad_norm": 2.2800028324127197, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6949162483215332, + "num_tokens": 202647369.0, + "step": 7942 + }, + { + "epoch": 0.8722820118603118, + "grad_norm": 2.3528990745544434, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7103798985481262, + "num_tokens": 202668399.0, + "step": 7943 + }, + { + "epoch": 0.8723918295629255, + "grad_norm": 2.0557305812835693, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7223270535469055, + "num_tokens": 202693658.0, + "step": 7944 + }, + { + "epoch": 0.8725016472655392, + "grad_norm": 2.087568759918213, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.6973812580108643, + "num_tokens": 202722508.0, + "step": 7945 + }, + { + "epoch": 0.8726114649681529, + "grad_norm": 2.017064094543457, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.697279691696167, + "num_tokens": 202752204.0, + "step": 7946 + }, + { + "epoch": 0.8727212826707665, + "grad_norm": 2.1012704372406006, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.6856257915496826, + "num_tokens": 202780349.0, + "step": 7947 + }, + { + "epoch": 0.8728311003733802, + "grad_norm": 2.3258414268493652, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.715941846370697, + "num_tokens": 202802358.0, + "step": 7948 + }, + { + "epoch": 0.8729409180759938, + "grad_norm": 2.164151430130005, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7131833434104919, + "num_tokens": 202828715.0, + "step": 7949 + }, + { + "epoch": 0.8730507357786075, + "grad_norm": 2.1251702308654785, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.699523389339447, + "num_tokens": 202857763.0, + "step": 7950 + }, + { + "epoch": 0.8731605534812211, + "grad_norm": 2.405646562576294, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7133781313896179, + "num_tokens": 202878898.0, + "step": 7951 + }, + { + "epoch": 0.8732703711838349, + "grad_norm": 1.9777263402938843, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7044060230255127, + "num_tokens": 202908473.0, + "step": 7952 + }, + { + "epoch": 0.8733801888864485, + "grad_norm": 2.0067079067230225, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.6998171210289001, + "num_tokens": 202939692.0, + "step": 7953 + }, + { + "epoch": 0.8734900065890622, + "grad_norm": 2.2835121154785156, + "learning_rate": 1e-06, + "loss": 1.0771, + "mean_token_accuracy": 0.6914718151092529, + "num_tokens": 202964134.0, + "step": 7954 + }, + { + "epoch": 0.8735998242916758, + "grad_norm": 2.2088167667388916, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7017960548400879, + "num_tokens": 202989029.0, + "step": 7955 + }, + { + "epoch": 0.8737096419942895, + "grad_norm": 2.0256617069244385, + "learning_rate": 1e-06, + "loss": 1.0508, + "mean_token_accuracy": 0.6845166683197021, + "num_tokens": 203019759.0, + "step": 7956 + }, + { + "epoch": 0.8738194596969031, + "grad_norm": 2.3620355129241943, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6922596096992493, + "num_tokens": 203043945.0, + "step": 7957 + }, + { + "epoch": 0.8739292773995168, + "grad_norm": 1.995462417602539, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7138462066650391, + "num_tokens": 203071873.0, + "step": 7958 + }, + { + "epoch": 0.8740390951021305, + "grad_norm": 1.9007186889648438, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7094964385032654, + "num_tokens": 203102815.0, + "step": 7959 + }, + { + "epoch": 0.8741489128047442, + "grad_norm": 2.0815021991729736, + "learning_rate": 1e-06, + "loss": 1.0844, + "mean_token_accuracy": 0.678687334060669, + "num_tokens": 203132683.0, + "step": 7960 + }, + { + "epoch": 0.8742587305073578, + "grad_norm": 2.4158847332000732, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7185230255126953, + "num_tokens": 203154222.0, + "step": 7961 + }, + { + "epoch": 0.8743685482099715, + "grad_norm": 2.0972096920013428, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7102921009063721, + "num_tokens": 203182347.0, + "step": 7962 + }, + { + "epoch": 0.8744783659125851, + "grad_norm": 2.035592794418335, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7035647630691528, + "num_tokens": 203209211.0, + "step": 7963 + }, + { + "epoch": 0.8745881836151987, + "grad_norm": 2.595669984817505, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7200378775596619, + "num_tokens": 203229997.0, + "step": 7964 + }, + { + "epoch": 0.8746980013178124, + "grad_norm": 2.148833990097046, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.725374698638916, + "num_tokens": 203254363.0, + "step": 7965 + }, + { + "epoch": 0.8748078190204261, + "grad_norm": 2.0715625286102295, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7125968933105469, + "num_tokens": 203281719.0, + "step": 7966 + }, + { + "epoch": 0.8749176367230398, + "grad_norm": 2.1454710960388184, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7033751606941223, + "num_tokens": 203306778.0, + "step": 7967 + }, + { + "epoch": 0.8750274544256534, + "grad_norm": 1.9602669477462769, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7122848033905029, + "num_tokens": 203339571.0, + "step": 7968 + }, + { + "epoch": 0.8751372721282671, + "grad_norm": 2.3675146102905273, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7187446355819702, + "num_tokens": 203360724.0, + "step": 7969 + }, + { + "epoch": 0.8752470898308807, + "grad_norm": 2.446458339691162, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.724134087562561, + "num_tokens": 203381006.0, + "step": 7970 + }, + { + "epoch": 0.8753569075334944, + "grad_norm": 2.131544589996338, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7224130034446716, + "num_tokens": 203408637.0, + "step": 7971 + }, + { + "epoch": 0.875466725236108, + "grad_norm": 2.2530524730682373, + "learning_rate": 1e-06, + "loss": 1.0523, + "mean_token_accuracy": 0.6952478289604187, + "num_tokens": 203436135.0, + "step": 7972 + }, + { + "epoch": 0.8755765429387217, + "grad_norm": 1.9718375205993652, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.706920325756073, + "num_tokens": 203466562.0, + "step": 7973 + }, + { + "epoch": 0.8756863606413354, + "grad_norm": 2.16756010055542, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.7042247653007507, + "num_tokens": 203494035.0, + "step": 7974 + }, + { + "epoch": 0.8757961783439491, + "grad_norm": 2.28667950630188, + "learning_rate": 1e-06, + "loss": 1.0334, + "mean_token_accuracy": 0.6859590411186218, + "num_tokens": 203517744.0, + "step": 7975 + }, + { + "epoch": 0.8759059960465627, + "grad_norm": 2.0340542793273926, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7213394045829773, + "num_tokens": 203545753.0, + "step": 7976 + }, + { + "epoch": 0.8760158137491764, + "grad_norm": 2.317312002182007, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7101621031761169, + "num_tokens": 203569871.0, + "step": 7977 + }, + { + "epoch": 0.87612563145179, + "grad_norm": 2.5317018032073975, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7137007713317871, + "num_tokens": 203588561.0, + "step": 7978 + }, + { + "epoch": 0.8762354491544037, + "grad_norm": 2.2979960441589355, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6920514702796936, + "num_tokens": 203612610.0, + "step": 7979 + }, + { + "epoch": 0.8763452668570173, + "grad_norm": 2.3187127113342285, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7196485996246338, + "num_tokens": 203639345.0, + "step": 7980 + }, + { + "epoch": 0.8764550845596311, + "grad_norm": 2.033320188522339, + "learning_rate": 1e-06, + "loss": 1.0501, + "mean_token_accuracy": 0.6867993474006653, + "num_tokens": 203668305.0, + "step": 7981 + }, + { + "epoch": 0.8765649022622447, + "grad_norm": 2.184077501296997, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7244601249694824, + "num_tokens": 203692722.0, + "step": 7982 + }, + { + "epoch": 0.8766747199648584, + "grad_norm": 2.0635793209075928, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7116989493370056, + "num_tokens": 203721698.0, + "step": 7983 + }, + { + "epoch": 0.876784537667472, + "grad_norm": 2.2933952808380127, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7004612684249878, + "num_tokens": 203744409.0, + "step": 7984 + }, + { + "epoch": 0.8768943553700856, + "grad_norm": 1.9839826822280884, + "learning_rate": 1e-06, + "loss": 1.0293, + "mean_token_accuracy": 0.6936679482460022, + "num_tokens": 203774894.0, + "step": 7985 + }, + { + "epoch": 0.8770041730726993, + "grad_norm": 2.089015245437622, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7072489857673645, + "num_tokens": 203800488.0, + "step": 7986 + }, + { + "epoch": 0.8771139907753129, + "grad_norm": 2.4008827209472656, + "learning_rate": 1e-06, + "loss": 1.067, + "mean_token_accuracy": 0.6908090710639954, + "num_tokens": 203824518.0, + "step": 7987 + }, + { + "epoch": 0.8772238084779267, + "grad_norm": 2.0131685733795166, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6941380500793457, + "num_tokens": 203854565.0, + "step": 7988 + }, + { + "epoch": 0.8773336261805403, + "grad_norm": 1.950657606124878, + "learning_rate": 1e-06, + "loss": 1.0538, + "mean_token_accuracy": 0.6839550733566284, + "num_tokens": 203885976.0, + "step": 7989 + }, + { + "epoch": 0.877443443883154, + "grad_norm": 2.301525115966797, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7153968214988708, + "num_tokens": 203908005.0, + "step": 7990 + }, + { + "epoch": 0.8775532615857676, + "grad_norm": 2.2673513889312744, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.69688880443573, + "num_tokens": 203930695.0, + "step": 7991 + }, + { + "epoch": 0.8776630792883813, + "grad_norm": 2.2368552684783936, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7114635705947876, + "num_tokens": 203956093.0, + "step": 7992 + }, + { + "epoch": 0.8777728969909949, + "grad_norm": 2.104231595993042, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7277116179466248, + "num_tokens": 203980585.0, + "step": 7993 + }, + { + "epoch": 0.8778827146936086, + "grad_norm": 1.947912573814392, + "learning_rate": 1e-06, + "loss": 1.0439, + "mean_token_accuracy": 0.6789056062698364, + "num_tokens": 204015507.0, + "step": 7994 + }, + { + "epoch": 0.8779925323962223, + "grad_norm": 2.236356258392334, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6953130960464478, + "num_tokens": 204040570.0, + "step": 7995 + }, + { + "epoch": 0.878102350098836, + "grad_norm": 1.9483082294464111, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6904900074005127, + "num_tokens": 204074389.0, + "step": 7996 + }, + { + "epoch": 0.8782121678014496, + "grad_norm": 2.2047431468963623, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7172991037368774, + "num_tokens": 204098590.0, + "step": 7997 + }, + { + "epoch": 0.8783219855040633, + "grad_norm": 2.1351373195648193, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.6936337947845459, + "num_tokens": 204123683.0, + "step": 7998 + }, + { + "epoch": 0.8784318032066769, + "grad_norm": 2.1590898036956787, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.691631555557251, + "num_tokens": 204153117.0, + "step": 7999 + }, + { + "epoch": 0.8785416209092906, + "grad_norm": 2.4162158966064453, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7223069667816162, + "num_tokens": 204173217.0, + "step": 8000 + }, + { + "epoch": 0.8786514386119042, + "grad_norm": 2.0171217918395996, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.6978655457496643, + "num_tokens": 204202994.0, + "step": 8001 + }, + { + "epoch": 0.8787612563145178, + "grad_norm": 2.3484833240509033, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7175041437149048, + "num_tokens": 204225085.0, + "step": 8002 + }, + { + "epoch": 0.8788710740171316, + "grad_norm": 2.621508836746216, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.715596079826355, + "num_tokens": 204243573.0, + "step": 8003 + }, + { + "epoch": 0.8789808917197452, + "grad_norm": 2.0180232524871826, + "learning_rate": 1e-06, + "loss": 1.0439, + "mean_token_accuracy": 0.6907529234886169, + "num_tokens": 204273583.0, + "step": 8004 + }, + { + "epoch": 0.8790907094223589, + "grad_norm": 2.165769338607788, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7214549779891968, + "num_tokens": 204298392.0, + "step": 8005 + }, + { + "epoch": 0.8792005271249725, + "grad_norm": 2.6047184467315674, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.736566424369812, + "num_tokens": 204316204.0, + "step": 8006 + }, + { + "epoch": 0.8793103448275862, + "grad_norm": 2.274144411087036, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6977031826972961, + "num_tokens": 204340122.0, + "step": 8007 + }, + { + "epoch": 0.8794201625301998, + "grad_norm": 2.118435859680176, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7123415470123291, + "num_tokens": 204366421.0, + "step": 8008 + }, + { + "epoch": 0.8795299802328135, + "grad_norm": 2.1779706478118896, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7433385848999023, + "num_tokens": 204389446.0, + "step": 8009 + }, + { + "epoch": 0.8796397979354272, + "grad_norm": 2.074962854385376, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7117140293121338, + "num_tokens": 204415286.0, + "step": 8010 + }, + { + "epoch": 0.8797496156380409, + "grad_norm": 2.3281948566436768, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7337989807128906, + "num_tokens": 204438245.0, + "step": 8011 + }, + { + "epoch": 0.8798594333406545, + "grad_norm": 2.1658220291137695, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7277705669403076, + "num_tokens": 204466098.0, + "step": 8012 + }, + { + "epoch": 0.8799692510432682, + "grad_norm": 2.207230806350708, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7121142148971558, + "num_tokens": 204490462.0, + "step": 8013 + }, + { + "epoch": 0.8800790687458818, + "grad_norm": 2.140312671661377, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7089897394180298, + "num_tokens": 204518709.0, + "step": 8014 + }, + { + "epoch": 0.8801888864484955, + "grad_norm": 2.221684217453003, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.7011297345161438, + "num_tokens": 204543374.0, + "step": 8015 + }, + { + "epoch": 0.8802987041511091, + "grad_norm": 2.0195045471191406, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7036921977996826, + "num_tokens": 204572033.0, + "step": 8016 + }, + { + "epoch": 0.8804085218537229, + "grad_norm": 2.5473814010620117, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7174760699272156, + "num_tokens": 204593043.0, + "step": 8017 + }, + { + "epoch": 0.8805183395563365, + "grad_norm": 2.181873321533203, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.6935386657714844, + "num_tokens": 204617944.0, + "step": 8018 + }, + { + "epoch": 0.8806281572589502, + "grad_norm": 2.2061634063720703, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.705897867679596, + "num_tokens": 204643326.0, + "step": 8019 + }, + { + "epoch": 0.8807379749615638, + "grad_norm": 2.0203042030334473, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7122536897659302, + "num_tokens": 204672452.0, + "step": 8020 + }, + { + "epoch": 0.8808477926641775, + "grad_norm": 2.182025194168091, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7114469408988953, + "num_tokens": 204696062.0, + "step": 8021 + }, + { + "epoch": 0.8809576103667911, + "grad_norm": 2.4689621925354004, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7319353818893433, + "num_tokens": 204716431.0, + "step": 8022 + }, + { + "epoch": 0.8810674280694047, + "grad_norm": 2.123652696609497, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7081596851348877, + "num_tokens": 204743953.0, + "step": 8023 + }, + { + "epoch": 0.8811772457720185, + "grad_norm": 2.0044360160827637, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.7011209726333618, + "num_tokens": 204773579.0, + "step": 8024 + }, + { + "epoch": 0.8812870634746321, + "grad_norm": 2.0336506366729736, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7165552973747253, + "num_tokens": 204802467.0, + "step": 8025 + }, + { + "epoch": 0.8813968811772458, + "grad_norm": 2.24883770942688, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7143993377685547, + "num_tokens": 204826689.0, + "step": 8026 + }, + { + "epoch": 0.8815066988798594, + "grad_norm": 1.8769735097885132, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6980506181716919, + "num_tokens": 204856709.0, + "step": 8027 + }, + { + "epoch": 0.8816165165824731, + "grad_norm": 2.0598366260528564, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7202149629592896, + "num_tokens": 204882905.0, + "step": 8028 + }, + { + "epoch": 0.8817263342850867, + "grad_norm": 2.2955591678619385, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7108569145202637, + "num_tokens": 204907243.0, + "step": 8029 + }, + { + "epoch": 0.8818361519877004, + "grad_norm": 1.948150634765625, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7163519859313965, + "num_tokens": 204936784.0, + "step": 8030 + }, + { + "epoch": 0.881945969690314, + "grad_norm": 2.2913401126861572, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7214452624320984, + "num_tokens": 204958744.0, + "step": 8031 + }, + { + "epoch": 0.8820557873929278, + "grad_norm": 2.7196762561798096, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7228288054466248, + "num_tokens": 204977432.0, + "step": 8032 + }, + { + "epoch": 0.8821656050955414, + "grad_norm": 2.106872320175171, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7193052768707275, + "num_tokens": 205003872.0, + "step": 8033 + }, + { + "epoch": 0.8822754227981551, + "grad_norm": 2.292564868927002, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7214735746383667, + "num_tokens": 205027054.0, + "step": 8034 + }, + { + "epoch": 0.8823852405007687, + "grad_norm": 2.2052829265594482, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7174067497253418, + "num_tokens": 205050606.0, + "step": 8035 + }, + { + "epoch": 0.8824950582033824, + "grad_norm": 2.366847515106201, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7051504850387573, + "num_tokens": 205071657.0, + "step": 8036 + }, + { + "epoch": 0.882604875905996, + "grad_norm": 1.885490894317627, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7077773809432983, + "num_tokens": 205103062.0, + "step": 8037 + }, + { + "epoch": 0.8827146936086097, + "grad_norm": 2.2685818672180176, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7058013677597046, + "num_tokens": 205125538.0, + "step": 8038 + }, + { + "epoch": 0.8828245113112234, + "grad_norm": 2.0084831714630127, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7024474143981934, + "num_tokens": 205155226.0, + "step": 8039 + }, + { + "epoch": 0.8829343290138371, + "grad_norm": 2.102931022644043, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.711627721786499, + "num_tokens": 205182652.0, + "step": 8040 + }, + { + "epoch": 0.8830441467164507, + "grad_norm": 2.0312023162841797, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7144737243652344, + "num_tokens": 205210116.0, + "step": 8041 + }, + { + "epoch": 0.8831539644190644, + "grad_norm": 2.0939009189605713, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6911118030548096, + "num_tokens": 205238749.0, + "step": 8042 + }, + { + "epoch": 0.883263782121678, + "grad_norm": 2.3589799404144287, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7224993705749512, + "num_tokens": 205260398.0, + "step": 8043 + }, + { + "epoch": 0.8833735998242916, + "grad_norm": 2.1007280349731445, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7230955958366394, + "num_tokens": 205285887.0, + "step": 8044 + }, + { + "epoch": 0.8834834175269053, + "grad_norm": 2.261913537979126, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.6933706998825073, + "num_tokens": 205309866.0, + "step": 8045 + }, + { + "epoch": 0.883593235229519, + "grad_norm": 2.4043431282043457, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.7091021537780762, + "num_tokens": 205333668.0, + "step": 8046 + }, + { + "epoch": 0.8837030529321327, + "grad_norm": 2.1130197048187256, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7032884955406189, + "num_tokens": 205358237.0, + "step": 8047 + }, + { + "epoch": 0.8838128706347463, + "grad_norm": 1.9482591152191162, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7234877347946167, + "num_tokens": 205388427.0, + "step": 8048 + }, + { + "epoch": 0.88392268833736, + "grad_norm": 2.3426477909088135, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7014952898025513, + "num_tokens": 205412610.0, + "step": 8049 + }, + { + "epoch": 0.8840325060399736, + "grad_norm": 2.0780487060546875, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7092633247375488, + "num_tokens": 205438928.0, + "step": 8050 + }, + { + "epoch": 0.8841423237425873, + "grad_norm": 2.42753529548645, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7429357767105103, + "num_tokens": 205458348.0, + "step": 8051 + }, + { + "epoch": 0.8842521414452009, + "grad_norm": 2.0726265907287598, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7178983688354492, + "num_tokens": 205484749.0, + "step": 8052 + }, + { + "epoch": 0.8843619591478147, + "grad_norm": 2.1755881309509277, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6997328996658325, + "num_tokens": 205510611.0, + "step": 8053 + }, + { + "epoch": 0.8844717768504283, + "grad_norm": 2.2772011756896973, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7105982303619385, + "num_tokens": 205535565.0, + "step": 8054 + }, + { + "epoch": 0.884581594553042, + "grad_norm": 2.3084800243377686, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7040679454803467, + "num_tokens": 205558689.0, + "step": 8055 + }, + { + "epoch": 0.8846914122556556, + "grad_norm": 2.199660062789917, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7170381546020508, + "num_tokens": 205583229.0, + "step": 8056 + }, + { + "epoch": 0.8848012299582693, + "grad_norm": 2.2150356769561768, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.7043685913085938, + "num_tokens": 205609094.0, + "step": 8057 + }, + { + "epoch": 0.8849110476608829, + "grad_norm": 1.7767324447631836, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.6995483636856079, + "num_tokens": 205645281.0, + "step": 8058 + }, + { + "epoch": 0.8850208653634966, + "grad_norm": 2.234102964401245, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7091266512870789, + "num_tokens": 205669935.0, + "step": 8059 + }, + { + "epoch": 0.8851306830661102, + "grad_norm": 1.9779695272445679, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.6894171237945557, + "num_tokens": 205700644.0, + "step": 8060 + }, + { + "epoch": 0.885240500768724, + "grad_norm": 2.3235182762145996, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.693942666053772, + "num_tokens": 205725550.0, + "step": 8061 + }, + { + "epoch": 0.8853503184713376, + "grad_norm": 2.1230380535125732, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7071337699890137, + "num_tokens": 205752774.0, + "step": 8062 + }, + { + "epoch": 0.8854601361739513, + "grad_norm": 1.9323312044143677, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7143452763557434, + "num_tokens": 205781646.0, + "step": 8063 + }, + { + "epoch": 0.8855699538765649, + "grad_norm": 2.172914743423462, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7124403715133667, + "num_tokens": 205807799.0, + "step": 8064 + }, + { + "epoch": 0.8856797715791785, + "grad_norm": 2.165093421936035, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.6862002015113831, + "num_tokens": 205834796.0, + "step": 8065 + }, + { + "epoch": 0.8857895892817922, + "grad_norm": 2.097378730773926, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7125340700149536, + "num_tokens": 205861341.0, + "step": 8066 + }, + { + "epoch": 0.8858994069844058, + "grad_norm": 2.0352721214294434, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7035176753997803, + "num_tokens": 205890257.0, + "step": 8067 + }, + { + "epoch": 0.8860092246870196, + "grad_norm": 2.2594070434570312, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7339224815368652, + "num_tokens": 205913985.0, + "step": 8068 + }, + { + "epoch": 0.8861190423896332, + "grad_norm": 2.8870673179626465, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.692770779132843, + "num_tokens": 205932527.0, + "step": 8069 + }, + { + "epoch": 0.8862288600922469, + "grad_norm": 2.340907335281372, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7326557040214539, + "num_tokens": 205954695.0, + "step": 8070 + }, + { + "epoch": 0.8863386777948605, + "grad_norm": 2.186817169189453, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.736504852771759, + "num_tokens": 205981255.0, + "step": 8071 + }, + { + "epoch": 0.8864484954974742, + "grad_norm": 2.46962571144104, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7113313674926758, + "num_tokens": 206001865.0, + "step": 8072 + }, + { + "epoch": 0.8865583132000878, + "grad_norm": 2.0750725269317627, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7221308350563049, + "num_tokens": 206029339.0, + "step": 8073 + }, + { + "epoch": 0.8866681309027015, + "grad_norm": 2.1456148624420166, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.73305344581604, + "num_tokens": 206056192.0, + "step": 8074 + }, + { + "epoch": 0.8867779486053152, + "grad_norm": 2.1381120681762695, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7032700777053833, + "num_tokens": 206083198.0, + "step": 8075 + }, + { + "epoch": 0.8868877663079289, + "grad_norm": 2.0688655376434326, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7284166812896729, + "num_tokens": 206108369.0, + "step": 8076 + }, + { + "epoch": 0.8869975840105425, + "grad_norm": 2.3058221340179443, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7040089964866638, + "num_tokens": 206132833.0, + "step": 8077 + }, + { + "epoch": 0.8871074017131562, + "grad_norm": 1.9792070388793945, + "learning_rate": 1e-06, + "loss": 1.0195, + "mean_token_accuracy": 0.696129560470581, + "num_tokens": 206163608.0, + "step": 8078 + }, + { + "epoch": 0.8872172194157698, + "grad_norm": 2.452145576477051, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7072368264198303, + "num_tokens": 206184873.0, + "step": 8079 + }, + { + "epoch": 0.8873270371183835, + "grad_norm": 2.391979694366455, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.7007300853729248, + "num_tokens": 206205322.0, + "step": 8080 + }, + { + "epoch": 0.8874368548209971, + "grad_norm": 2.303009510040283, + "learning_rate": 1e-06, + "loss": 1.0441, + "mean_token_accuracy": 0.6828712224960327, + "num_tokens": 206228116.0, + "step": 8081 + }, + { + "epoch": 0.8875466725236109, + "grad_norm": 2.4907288551330566, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.710440456867218, + "num_tokens": 206248391.0, + "step": 8082 + }, + { + "epoch": 0.8876564902262245, + "grad_norm": 2.316429853439331, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.708078145980835, + "num_tokens": 206271176.0, + "step": 8083 + }, + { + "epoch": 0.8877663079288381, + "grad_norm": 2.1106889247894287, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7063971161842346, + "num_tokens": 206297935.0, + "step": 8084 + }, + { + "epoch": 0.8878761256314518, + "grad_norm": 2.2537920475006104, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7075222730636597, + "num_tokens": 206321338.0, + "step": 8085 + }, + { + "epoch": 0.8879859433340654, + "grad_norm": 2.212817668914795, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7106171250343323, + "num_tokens": 206347036.0, + "step": 8086 + }, + { + "epoch": 0.8880957610366791, + "grad_norm": 2.122941732406616, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.6924371719360352, + "num_tokens": 206375496.0, + "step": 8087 + }, + { + "epoch": 0.8882055787392927, + "grad_norm": 2.3644185066223145, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7126332521438599, + "num_tokens": 206399920.0, + "step": 8088 + }, + { + "epoch": 0.8883153964419065, + "grad_norm": 2.4197628498077393, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7204573154449463, + "num_tokens": 206421421.0, + "step": 8089 + }, + { + "epoch": 0.8884252141445201, + "grad_norm": 1.996775507926941, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.6990528106689453, + "num_tokens": 206452978.0, + "step": 8090 + }, + { + "epoch": 0.8885350318471338, + "grad_norm": 1.8042982816696167, + "learning_rate": 1e-06, + "loss": 1.045, + "mean_token_accuracy": 0.6849826574325562, + "num_tokens": 206488451.0, + "step": 8091 + }, + { + "epoch": 0.8886448495497474, + "grad_norm": 2.1196727752685547, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7026354074478149, + "num_tokens": 206514850.0, + "step": 8092 + }, + { + "epoch": 0.8887546672523611, + "grad_norm": 2.1581852436065674, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7023165225982666, + "num_tokens": 206542776.0, + "step": 8093 + }, + { + "epoch": 0.8888644849549747, + "grad_norm": 1.9235841035842896, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7133387327194214, + "num_tokens": 206573699.0, + "step": 8094 + }, + { + "epoch": 0.8889743026575884, + "grad_norm": 2.024003028869629, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7041728496551514, + "num_tokens": 206602712.0, + "step": 8095 + }, + { + "epoch": 0.889084120360202, + "grad_norm": 2.3845951557159424, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7136967182159424, + "num_tokens": 206625992.0, + "step": 8096 + }, + { + "epoch": 0.8891939380628158, + "grad_norm": 2.1271703243255615, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7183996438980103, + "num_tokens": 206652222.0, + "step": 8097 + }, + { + "epoch": 0.8893037557654294, + "grad_norm": 2.16776967048645, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6920185089111328, + "num_tokens": 206678631.0, + "step": 8098 + }, + { + "epoch": 0.8894135734680431, + "grad_norm": 2.182600736618042, + "learning_rate": 1e-06, + "loss": 0.798, + "mean_token_accuracy": 0.7523041367530823, + "num_tokens": 206701727.0, + "step": 8099 + }, + { + "epoch": 0.8895233911706567, + "grad_norm": 2.0134191513061523, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6836556196212769, + "num_tokens": 206736078.0, + "step": 8100 + }, + { + "epoch": 0.8896332088732704, + "grad_norm": 2.0302040576934814, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.7014714479446411, + "num_tokens": 206763956.0, + "step": 8101 + }, + { + "epoch": 0.889743026575884, + "grad_norm": 2.346517324447632, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6950737237930298, + "num_tokens": 206787756.0, + "step": 8102 + }, + { + "epoch": 0.8898528442784976, + "grad_norm": 1.8988317251205444, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.693649172782898, + "num_tokens": 206821211.0, + "step": 8103 + }, + { + "epoch": 0.8899626619811114, + "grad_norm": 2.2481720447540283, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7020895481109619, + "num_tokens": 206845687.0, + "step": 8104 + }, + { + "epoch": 0.890072479683725, + "grad_norm": 2.0671732425689697, + "learning_rate": 1e-06, + "loss": 1.0803, + "mean_token_accuracy": 0.676762044429779, + "num_tokens": 206876427.0, + "step": 8105 + }, + { + "epoch": 0.8901822973863387, + "grad_norm": 1.943385124206543, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.6982173919677734, + "num_tokens": 206904752.0, + "step": 8106 + }, + { + "epoch": 0.8902921150889523, + "grad_norm": 2.1870362758636475, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7129062414169312, + "num_tokens": 206930064.0, + "step": 8107 + }, + { + "epoch": 0.890401932791566, + "grad_norm": 2.5749807357788086, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7229448556900024, + "num_tokens": 206948411.0, + "step": 8108 + }, + { + "epoch": 0.8905117504941796, + "grad_norm": 1.9367761611938477, + "learning_rate": 1e-06, + "loss": 1.0486, + "mean_token_accuracy": 0.6900073289871216, + "num_tokens": 206978968.0, + "step": 8109 + }, + { + "epoch": 0.8906215681967933, + "grad_norm": 2.1072304248809814, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7080546617507935, + "num_tokens": 207005327.0, + "step": 8110 + }, + { + "epoch": 0.890731385899407, + "grad_norm": 2.352297067642212, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7029086351394653, + "num_tokens": 207030508.0, + "step": 8111 + }, + { + "epoch": 0.8908412036020207, + "grad_norm": 2.070324659347534, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7264240384101868, + "num_tokens": 207056899.0, + "step": 8112 + }, + { + "epoch": 0.8909510213046343, + "grad_norm": 2.56591534614563, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7274676561355591, + "num_tokens": 207075843.0, + "step": 8113 + }, + { + "epoch": 0.891060839007248, + "grad_norm": 2.5731210708618164, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7176763415336609, + "num_tokens": 207094938.0, + "step": 8114 + }, + { + "epoch": 0.8911706567098616, + "grad_norm": 1.8527865409851074, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.696052074432373, + "num_tokens": 207127056.0, + "step": 8115 + }, + { + "epoch": 0.8912804744124753, + "grad_norm": 1.9842281341552734, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7328097820281982, + "num_tokens": 207154804.0, + "step": 8116 + }, + { + "epoch": 0.8913902921150889, + "grad_norm": 2.300697088241577, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7300243377685547, + "num_tokens": 207176591.0, + "step": 8117 + }, + { + "epoch": 0.8915001098177027, + "grad_norm": 2.2508349418640137, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.695685625076294, + "num_tokens": 207200400.0, + "step": 8118 + }, + { + "epoch": 0.8916099275203163, + "grad_norm": 2.1758246421813965, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7162708044052124, + "num_tokens": 207226715.0, + "step": 8119 + }, + { + "epoch": 0.89171974522293, + "grad_norm": 2.1845719814300537, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7101333141326904, + "num_tokens": 207252902.0, + "step": 8120 + }, + { + "epoch": 0.8918295629255436, + "grad_norm": 2.028402090072632, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.7023823261260986, + "num_tokens": 207282355.0, + "step": 8121 + }, + { + "epoch": 0.8919393806281573, + "grad_norm": 2.0588176250457764, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.7045276165008545, + "num_tokens": 207309832.0, + "step": 8122 + }, + { + "epoch": 0.8920491983307709, + "grad_norm": 2.17726731300354, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.7023817300796509, + "num_tokens": 207334147.0, + "step": 8123 + }, + { + "epoch": 0.8921590160333845, + "grad_norm": 2.5369808673858643, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7159106731414795, + "num_tokens": 207352569.0, + "step": 8124 + }, + { + "epoch": 0.8922688337359982, + "grad_norm": 2.131450653076172, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7278897762298584, + "num_tokens": 207378566.0, + "step": 8125 + }, + { + "epoch": 0.892378651438612, + "grad_norm": 2.2866594791412354, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.712964653968811, + "num_tokens": 207400608.0, + "step": 8126 + }, + { + "epoch": 0.8924884691412256, + "grad_norm": 2.2342045307159424, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7218716144561768, + "num_tokens": 207423227.0, + "step": 8127 + }, + { + "epoch": 0.8925982868438392, + "grad_norm": 2.408637762069702, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7173826694488525, + "num_tokens": 207444763.0, + "step": 8128 + }, + { + "epoch": 0.8927081045464529, + "grad_norm": 2.461836814880371, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7401597499847412, + "num_tokens": 207464277.0, + "step": 8129 + }, + { + "epoch": 0.8928179222490665, + "grad_norm": 2.0348687171936035, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7127262949943542, + "num_tokens": 207491347.0, + "step": 8130 + }, + { + "epoch": 0.8929277399516802, + "grad_norm": 2.591243028640747, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.702373206615448, + "num_tokens": 207512458.0, + "step": 8131 + }, + { + "epoch": 0.8930375576542938, + "grad_norm": 1.7369784116744995, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.7035571932792664, + "num_tokens": 207551405.0, + "step": 8132 + }, + { + "epoch": 0.8931473753569076, + "grad_norm": 2.3520164489746094, + "learning_rate": 1e-06, + "loss": 1.0816, + "mean_token_accuracy": 0.6781793236732483, + "num_tokens": 207578380.0, + "step": 8133 + }, + { + "epoch": 0.8932571930595212, + "grad_norm": 2.0722768306732178, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6822837591171265, + "num_tokens": 207607610.0, + "step": 8134 + }, + { + "epoch": 0.8933670107621349, + "grad_norm": 2.0780889987945557, + "learning_rate": 1e-06, + "loss": 1.0413, + "mean_token_accuracy": 0.6837605237960815, + "num_tokens": 207635487.0, + "step": 8135 + }, + { + "epoch": 0.8934768284647485, + "grad_norm": 2.2104365825653076, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7171766757965088, + "num_tokens": 207660793.0, + "step": 8136 + }, + { + "epoch": 0.8935866461673622, + "grad_norm": 2.046804666519165, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7134707570075989, + "num_tokens": 207689467.0, + "step": 8137 + }, + { + "epoch": 0.8936964638699758, + "grad_norm": 1.9579980373382568, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7191202044487, + "num_tokens": 207719137.0, + "step": 8138 + }, + { + "epoch": 0.8938062815725895, + "grad_norm": 2.096395492553711, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7028322815895081, + "num_tokens": 207747106.0, + "step": 8139 + }, + { + "epoch": 0.8939160992752032, + "grad_norm": 2.3557021617889404, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.710523247718811, + "num_tokens": 207770159.0, + "step": 8140 + }, + { + "epoch": 0.8940259169778169, + "grad_norm": 2.256887674331665, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.6984013319015503, + "num_tokens": 207793677.0, + "step": 8141 + }, + { + "epoch": 0.8941357346804305, + "grad_norm": 2.425215244293213, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7092025876045227, + "num_tokens": 207814349.0, + "step": 8142 + }, + { + "epoch": 0.8942455523830442, + "grad_norm": 2.3955423831939697, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.7067289352416992, + "num_tokens": 207836904.0, + "step": 8143 + }, + { + "epoch": 0.8943553700856578, + "grad_norm": 1.838095784187317, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.6916681528091431, + "num_tokens": 207873181.0, + "step": 8144 + }, + { + "epoch": 0.8944651877882714, + "grad_norm": 1.8907183408737183, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6953459978103638, + "num_tokens": 207905613.0, + "step": 8145 + }, + { + "epoch": 0.8945750054908851, + "grad_norm": 1.9717977046966553, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.715842604637146, + "num_tokens": 207934349.0, + "step": 8146 + }, + { + "epoch": 0.8946848231934988, + "grad_norm": 2.1978819370269775, + "learning_rate": 1e-06, + "loss": 1.0683, + "mean_token_accuracy": 0.6833866834640503, + "num_tokens": 207963158.0, + "step": 8147 + }, + { + "epoch": 0.8947946408961125, + "grad_norm": 2.421788215637207, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7150449752807617, + "num_tokens": 207984394.0, + "step": 8148 + }, + { + "epoch": 0.8949044585987261, + "grad_norm": 2.6382522583007812, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7035849690437317, + "num_tokens": 208002675.0, + "step": 8149 + }, + { + "epoch": 0.8950142763013398, + "grad_norm": 2.243023157119751, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7132771611213684, + "num_tokens": 208027320.0, + "step": 8150 + }, + { + "epoch": 0.8951240940039534, + "grad_norm": 2.0607874393463135, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7063799500465393, + "num_tokens": 208054896.0, + "step": 8151 + }, + { + "epoch": 0.8952339117065671, + "grad_norm": 2.1611263751983643, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7077221870422363, + "num_tokens": 208082136.0, + "step": 8152 + }, + { + "epoch": 0.8953437294091807, + "grad_norm": 2.0695509910583496, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7110941410064697, + "num_tokens": 208110835.0, + "step": 8153 + }, + { + "epoch": 0.8954535471117944, + "grad_norm": 2.073334217071533, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7112303972244263, + "num_tokens": 208137498.0, + "step": 8154 + }, + { + "epoch": 0.8955633648144081, + "grad_norm": 2.066505193710327, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7115767002105713, + "num_tokens": 208162914.0, + "step": 8155 + }, + { + "epoch": 0.8956731825170218, + "grad_norm": 2.1054797172546387, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7243220806121826, + "num_tokens": 208187222.0, + "step": 8156 + }, + { + "epoch": 0.8957830002196354, + "grad_norm": 2.123638391494751, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7150706052780151, + "num_tokens": 208213427.0, + "step": 8157 + }, + { + "epoch": 0.8958928179222491, + "grad_norm": 2.112565755844116, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7200005054473877, + "num_tokens": 208237589.0, + "step": 8158 + }, + { + "epoch": 0.8960026356248627, + "grad_norm": 2.33008074760437, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7161738276481628, + "num_tokens": 208261419.0, + "step": 8159 + }, + { + "epoch": 0.8961124533274764, + "grad_norm": 2.094921112060547, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.7020004987716675, + "num_tokens": 208288453.0, + "step": 8160 + }, + { + "epoch": 0.89622227103009, + "grad_norm": 2.3519318103790283, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7039912939071655, + "num_tokens": 208309595.0, + "step": 8161 + }, + { + "epoch": 0.8963320887327038, + "grad_norm": 2.0799500942230225, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7106013298034668, + "num_tokens": 208337072.0, + "step": 8162 + }, + { + "epoch": 0.8964419064353174, + "grad_norm": 2.5361225605010986, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7241424322128296, + "num_tokens": 208356570.0, + "step": 8163 + }, + { + "epoch": 0.896551724137931, + "grad_norm": 2.2795450687408447, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7116364240646362, + "num_tokens": 208378570.0, + "step": 8164 + }, + { + "epoch": 0.8966615418405447, + "grad_norm": 2.0645058155059814, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.6955856680870056, + "num_tokens": 208405369.0, + "step": 8165 + }, + { + "epoch": 0.8967713595431583, + "grad_norm": 2.2083215713500977, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7245143055915833, + "num_tokens": 208430992.0, + "step": 8166 + }, + { + "epoch": 0.896881177245772, + "grad_norm": 2.0142414569854736, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7119238376617432, + "num_tokens": 208459249.0, + "step": 8167 + }, + { + "epoch": 0.8969909949483856, + "grad_norm": 1.9236258268356323, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.7024449110031128, + "num_tokens": 208491139.0, + "step": 8168 + }, + { + "epoch": 0.8971008126509994, + "grad_norm": 2.3072311878204346, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7119112014770508, + "num_tokens": 208514568.0, + "step": 8169 + }, + { + "epoch": 0.897210630353613, + "grad_norm": 2.0993571281433105, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7137969136238098, + "num_tokens": 208539419.0, + "step": 8170 + }, + { + "epoch": 0.8973204480562267, + "grad_norm": 2.1041698455810547, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6880246996879578, + "num_tokens": 208569371.0, + "step": 8171 + }, + { + "epoch": 0.8974302657588403, + "grad_norm": 2.2329494953155518, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7310115098953247, + "num_tokens": 208592390.0, + "step": 8172 + }, + { + "epoch": 0.897540083461454, + "grad_norm": 2.447042465209961, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.698722243309021, + "num_tokens": 208613236.0, + "step": 8173 + }, + { + "epoch": 0.8976499011640676, + "grad_norm": 2.221576690673828, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7170082330703735, + "num_tokens": 208637998.0, + "step": 8174 + }, + { + "epoch": 0.8977597188666813, + "grad_norm": 2.1318347454071045, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6907804608345032, + "num_tokens": 208664565.0, + "step": 8175 + }, + { + "epoch": 0.897869536569295, + "grad_norm": 2.3087522983551025, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6990033984184265, + "num_tokens": 208687643.0, + "step": 8176 + }, + { + "epoch": 0.8979793542719087, + "grad_norm": 2.0804526805877686, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7072123289108276, + "num_tokens": 208714037.0, + "step": 8177 + }, + { + "epoch": 0.8980891719745223, + "grad_norm": 2.1933529376983643, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.6891457438468933, + "num_tokens": 208741007.0, + "step": 8178 + }, + { + "epoch": 0.898198989677136, + "grad_norm": 2.417901039123535, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7103731036186218, + "num_tokens": 208761840.0, + "step": 8179 + }, + { + "epoch": 0.8983088073797496, + "grad_norm": 2.079425573348999, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7042207717895508, + "num_tokens": 208789094.0, + "step": 8180 + }, + { + "epoch": 0.8984186250823633, + "grad_norm": 2.3457400798797607, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7369875907897949, + "num_tokens": 208812000.0, + "step": 8181 + }, + { + "epoch": 0.8985284427849769, + "grad_norm": 2.2131285667419434, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.6995183229446411, + "num_tokens": 208837394.0, + "step": 8182 + }, + { + "epoch": 0.8986382604875905, + "grad_norm": 2.060746192932129, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6935306787490845, + "num_tokens": 208866437.0, + "step": 8183 + }, + { + "epoch": 0.8987480781902043, + "grad_norm": 2.0057640075683594, + "learning_rate": 1e-06, + "loss": 1.0575, + "mean_token_accuracy": 0.6909931898117065, + "num_tokens": 208897156.0, + "step": 8184 + }, + { + "epoch": 0.898857895892818, + "grad_norm": 2.392537832260132, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.7047427296638489, + "num_tokens": 208920195.0, + "step": 8185 + }, + { + "epoch": 0.8989677135954316, + "grad_norm": 1.9834046363830566, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6876018047332764, + "num_tokens": 208954411.0, + "step": 8186 + }, + { + "epoch": 0.8990775312980452, + "grad_norm": 2.446153402328491, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7078350782394409, + "num_tokens": 208976492.0, + "step": 8187 + }, + { + "epoch": 0.8991873490006589, + "grad_norm": 2.0284016132354736, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.7001134157180786, + "num_tokens": 209006608.0, + "step": 8188 + }, + { + "epoch": 0.8992971667032725, + "grad_norm": 2.137441396713257, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.7015140056610107, + "num_tokens": 209034936.0, + "step": 8189 + }, + { + "epoch": 0.8994069844058862, + "grad_norm": 2.1444668769836426, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.6984111070632935, + "num_tokens": 209062745.0, + "step": 8190 + }, + { + "epoch": 0.8995168021084999, + "grad_norm": 2.2657670974731445, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7293143272399902, + "num_tokens": 209086094.0, + "step": 8191 + }, + { + "epoch": 0.8996266198111136, + "grad_norm": 2.0969014167785645, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.6931700706481934, + "num_tokens": 209113978.0, + "step": 8192 + }, + { + "epoch": 0.8997364375137272, + "grad_norm": 2.103283405303955, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7387771606445312, + "num_tokens": 209138996.0, + "step": 8193 + }, + { + "epoch": 0.8998462552163409, + "grad_norm": 2.104464054107666, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.6978615522384644, + "num_tokens": 209167343.0, + "step": 8194 + }, + { + "epoch": 0.8999560729189545, + "grad_norm": 2.102968692779541, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7091192007064819, + "num_tokens": 209194215.0, + "step": 8195 + }, + { + "epoch": 0.9000658906215682, + "grad_norm": 2.2868616580963135, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6961884498596191, + "num_tokens": 209221228.0, + "step": 8196 + }, + { + "epoch": 0.9001757083241818, + "grad_norm": 2.086684226989746, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.7145974040031433, + "num_tokens": 209248153.0, + "step": 8197 + }, + { + "epoch": 0.9002855260267956, + "grad_norm": 2.270866632461548, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6945072412490845, + "num_tokens": 209273221.0, + "step": 8198 + }, + { + "epoch": 0.9003953437294092, + "grad_norm": 1.9699279069900513, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7086246013641357, + "num_tokens": 209303009.0, + "step": 8199 + }, + { + "epoch": 0.9005051614320229, + "grad_norm": 2.1093428134918213, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6975051164627075, + "num_tokens": 209330460.0, + "step": 8200 + }, + { + "epoch": 0.9006149791346365, + "grad_norm": 2.284337043762207, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6992518305778503, + "num_tokens": 209354711.0, + "step": 8201 + }, + { + "epoch": 0.9007247968372502, + "grad_norm": 2.303065538406372, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7098444700241089, + "num_tokens": 209378854.0, + "step": 8202 + }, + { + "epoch": 0.9008346145398638, + "grad_norm": 2.1029791831970215, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6936413049697876, + "num_tokens": 209406967.0, + "step": 8203 + }, + { + "epoch": 0.9009444322424774, + "grad_norm": 2.077446937561035, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7145816087722778, + "num_tokens": 209434978.0, + "step": 8204 + }, + { + "epoch": 0.9010542499450912, + "grad_norm": 2.4574825763702393, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7323477864265442, + "num_tokens": 209454437.0, + "step": 8205 + }, + { + "epoch": 0.9011640676477048, + "grad_norm": 2.1407997608184814, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7041203379631042, + "num_tokens": 209479251.0, + "step": 8206 + }, + { + "epoch": 0.9012738853503185, + "grad_norm": 2.1040139198303223, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.726842999458313, + "num_tokens": 209506747.0, + "step": 8207 + }, + { + "epoch": 0.9013837030529321, + "grad_norm": 2.337153911590576, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7195035815238953, + "num_tokens": 209529196.0, + "step": 8208 + }, + { + "epoch": 0.9014935207555458, + "grad_norm": 1.9493293762207031, + "learning_rate": 1e-06, + "loss": 1.0676, + "mean_token_accuracy": 0.6865408420562744, + "num_tokens": 209561691.0, + "step": 8209 + }, + { + "epoch": 0.9016033384581594, + "grad_norm": 2.083122730255127, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.7093666791915894, + "num_tokens": 209589862.0, + "step": 8210 + }, + { + "epoch": 0.9017131561607731, + "grad_norm": 1.965116262435913, + "learning_rate": 1e-06, + "loss": 1.0351, + "mean_token_accuracy": 0.6941131353378296, + "num_tokens": 209619565.0, + "step": 8211 + }, + { + "epoch": 0.9018229738633867, + "grad_norm": 2.3992998600006104, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7217390537261963, + "num_tokens": 209639852.0, + "step": 8212 + }, + { + "epoch": 0.9019327915660005, + "grad_norm": 3.156799793243408, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7520861625671387, + "num_tokens": 209652890.0, + "step": 8213 + }, + { + "epoch": 0.9020426092686141, + "grad_norm": 2.49477219581604, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.6865535974502563, + "num_tokens": 209675328.0, + "step": 8214 + }, + { + "epoch": 0.9021524269712278, + "grad_norm": 2.4625470638275146, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7248571515083313, + "num_tokens": 209694943.0, + "step": 8215 + }, + { + "epoch": 0.9022622446738414, + "grad_norm": 2.164482593536377, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7131797075271606, + "num_tokens": 209720349.0, + "step": 8216 + }, + { + "epoch": 0.9023720623764551, + "grad_norm": 2.1537246704101562, + "learning_rate": 1e-06, + "loss": 1.0238, + "mean_token_accuracy": 0.6878260374069214, + "num_tokens": 209746023.0, + "step": 8217 + }, + { + "epoch": 0.9024818800790687, + "grad_norm": 2.084418773651123, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6923834681510925, + "num_tokens": 209773597.0, + "step": 8218 + }, + { + "epoch": 0.9025916977816824, + "grad_norm": 2.383103847503662, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7290114164352417, + "num_tokens": 209794225.0, + "step": 8219 + }, + { + "epoch": 0.9027015154842961, + "grad_norm": 2.119530200958252, + "learning_rate": 1e-06, + "loss": 1.0567, + "mean_token_accuracy": 0.6849199533462524, + "num_tokens": 209820944.0, + "step": 8220 + }, + { + "epoch": 0.9028113331869098, + "grad_norm": 2.186028003692627, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7317938208580017, + "num_tokens": 209845918.0, + "step": 8221 + }, + { + "epoch": 0.9029211508895234, + "grad_norm": 2.398470163345337, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7235792875289917, + "num_tokens": 209866564.0, + "step": 8222 + }, + { + "epoch": 0.903030968592137, + "grad_norm": 1.9000766277313232, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.6999328136444092, + "num_tokens": 209895224.0, + "step": 8223 + }, + { + "epoch": 0.9031407862947507, + "grad_norm": 2.1799635887145996, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7263394594192505, + "num_tokens": 209919127.0, + "step": 8224 + }, + { + "epoch": 0.9032506039973643, + "grad_norm": 2.3838412761688232, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7160475254058838, + "num_tokens": 209940479.0, + "step": 8225 + }, + { + "epoch": 0.903360421699978, + "grad_norm": 2.3337912559509277, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7055763006210327, + "num_tokens": 209963870.0, + "step": 8226 + }, + { + "epoch": 0.9034702394025917, + "grad_norm": 2.1042730808258057, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6930480599403381, + "num_tokens": 209992907.0, + "step": 8227 + }, + { + "epoch": 0.9035800571052054, + "grad_norm": 2.1997077465057373, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7044655680656433, + "num_tokens": 210019318.0, + "step": 8228 + }, + { + "epoch": 0.903689874807819, + "grad_norm": 2.268958806991577, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7115179300308228, + "num_tokens": 210043300.0, + "step": 8229 + }, + { + "epoch": 0.9037996925104327, + "grad_norm": 2.115872621536255, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7037508487701416, + "num_tokens": 210071285.0, + "step": 8230 + }, + { + "epoch": 0.9039095102130463, + "grad_norm": 2.441955089569092, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7029836773872375, + "num_tokens": 210092643.0, + "step": 8231 + }, + { + "epoch": 0.90401932791566, + "grad_norm": 2.023045539855957, + "learning_rate": 1e-06, + "loss": 1.0472, + "mean_token_accuracy": 0.6844881772994995, + "num_tokens": 210122149.0, + "step": 8232 + }, + { + "epoch": 0.9041291456182736, + "grad_norm": 2.086890935897827, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7087714076042175, + "num_tokens": 210148317.0, + "step": 8233 + }, + { + "epoch": 0.9042389633208874, + "grad_norm": 2.289519786834717, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7342684268951416, + "num_tokens": 210169864.0, + "step": 8234 + }, + { + "epoch": 0.904348781023501, + "grad_norm": 2.1271088123321533, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6959245800971985, + "num_tokens": 210194653.0, + "step": 8235 + }, + { + "epoch": 0.9044585987261147, + "grad_norm": 2.2537999153137207, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.716649055480957, + "num_tokens": 210218082.0, + "step": 8236 + }, + { + "epoch": 0.9045684164287283, + "grad_norm": 2.3785550594329834, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7107816934585571, + "num_tokens": 210241722.0, + "step": 8237 + }, + { + "epoch": 0.904678234131342, + "grad_norm": 2.0812220573425293, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.6976780891418457, + "num_tokens": 210269584.0, + "step": 8238 + }, + { + "epoch": 0.9047880518339556, + "grad_norm": 2.3189194202423096, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7114961743354797, + "num_tokens": 210293855.0, + "step": 8239 + }, + { + "epoch": 0.9048978695365693, + "grad_norm": 1.9349243640899658, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7012312412261963, + "num_tokens": 210327494.0, + "step": 8240 + }, + { + "epoch": 0.9050076872391829, + "grad_norm": 2.138049840927124, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7028136849403381, + "num_tokens": 210355913.0, + "step": 8241 + }, + { + "epoch": 0.9051175049417967, + "grad_norm": 2.2546379566192627, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6886201500892639, + "num_tokens": 210382602.0, + "step": 8242 + }, + { + "epoch": 0.9052273226444103, + "grad_norm": 2.304539918899536, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7180201411247253, + "num_tokens": 210407114.0, + "step": 8243 + }, + { + "epoch": 0.905337140347024, + "grad_norm": 2.500370502471924, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7210990786552429, + "num_tokens": 210427443.0, + "step": 8244 + }, + { + "epoch": 0.9054469580496376, + "grad_norm": 2.5456173419952393, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.6988140344619751, + "num_tokens": 210448525.0, + "step": 8245 + }, + { + "epoch": 0.9055567757522512, + "grad_norm": 2.097316026687622, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.6961987018585205, + "num_tokens": 210475484.0, + "step": 8246 + }, + { + "epoch": 0.9056665934548649, + "grad_norm": 2.12465238571167, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7416906356811523, + "num_tokens": 210501566.0, + "step": 8247 + }, + { + "epoch": 0.9057764111574785, + "grad_norm": 2.182295799255371, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7153162956237793, + "num_tokens": 210526753.0, + "step": 8248 + }, + { + "epoch": 0.9058862288600923, + "grad_norm": 2.213397741317749, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7171356678009033, + "num_tokens": 210549880.0, + "step": 8249 + }, + { + "epoch": 0.9059960465627059, + "grad_norm": 2.3343427181243896, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7016216516494751, + "num_tokens": 210572097.0, + "step": 8250 + }, + { + "epoch": 0.9061058642653196, + "grad_norm": 2.078770637512207, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7167640924453735, + "num_tokens": 210599294.0, + "step": 8251 + }, + { + "epoch": 0.9062156819679332, + "grad_norm": 2.232945442199707, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.7100313901901245, + "num_tokens": 210622831.0, + "step": 8252 + }, + { + "epoch": 0.9063254996705469, + "grad_norm": 2.054152011871338, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.7003246545791626, + "num_tokens": 210652459.0, + "step": 8253 + }, + { + "epoch": 0.9064353173731605, + "grad_norm": 2.2054684162139893, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.6897026300430298, + "num_tokens": 210678157.0, + "step": 8254 + }, + { + "epoch": 0.9065451350757742, + "grad_norm": 1.9898271560668945, + "learning_rate": 1e-06, + "loss": 1.0506, + "mean_token_accuracy": 0.6857411861419678, + "num_tokens": 210707794.0, + "step": 8255 + }, + { + "epoch": 0.9066549527783879, + "grad_norm": 2.0693671703338623, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7106510400772095, + "num_tokens": 210732719.0, + "step": 8256 + }, + { + "epoch": 0.9067647704810016, + "grad_norm": 2.371004581451416, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7230894565582275, + "num_tokens": 210754791.0, + "step": 8257 + }, + { + "epoch": 0.9068745881836152, + "grad_norm": 1.9515793323516846, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7074423432350159, + "num_tokens": 210784289.0, + "step": 8258 + }, + { + "epoch": 0.9069844058862289, + "grad_norm": 2.4540040493011475, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7136307954788208, + "num_tokens": 210806185.0, + "step": 8259 + }, + { + "epoch": 0.9070942235888425, + "grad_norm": 2.565291166305542, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7305831909179688, + "num_tokens": 210825876.0, + "step": 8260 + }, + { + "epoch": 0.9072040412914562, + "grad_norm": 2.2623140811920166, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.721362829208374, + "num_tokens": 210846917.0, + "step": 8261 + }, + { + "epoch": 0.9073138589940698, + "grad_norm": 2.3595340251922607, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7337254881858826, + "num_tokens": 210868820.0, + "step": 8262 + }, + { + "epoch": 0.9074236766966836, + "grad_norm": 2.0953445434570312, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.710159957408905, + "num_tokens": 210894745.0, + "step": 8263 + }, + { + "epoch": 0.9075334943992972, + "grad_norm": 2.2737982273101807, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.750799834728241, + "num_tokens": 210916863.0, + "step": 8264 + }, + { + "epoch": 0.9076433121019108, + "grad_norm": 1.9948264360427856, + "learning_rate": 1e-06, + "loss": 1.148, + "mean_token_accuracy": 0.668470025062561, + "num_tokens": 210946961.0, + "step": 8265 + }, + { + "epoch": 0.9077531298045245, + "grad_norm": 2.2884836196899414, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7018813490867615, + "num_tokens": 210970636.0, + "step": 8266 + }, + { + "epoch": 0.9078629475071381, + "grad_norm": 1.8627077341079712, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6938990354537964, + "num_tokens": 211004821.0, + "step": 8267 + }, + { + "epoch": 0.9079727652097518, + "grad_norm": 2.3468496799468994, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7038485407829285, + "num_tokens": 211026990.0, + "step": 8268 + }, + { + "epoch": 0.9080825829123654, + "grad_norm": 2.264582395553589, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7174032926559448, + "num_tokens": 211051092.0, + "step": 8269 + }, + { + "epoch": 0.9081924006149792, + "grad_norm": 2.3363239765167236, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7107518911361694, + "num_tokens": 211073681.0, + "step": 8270 + }, + { + "epoch": 0.9083022183175928, + "grad_norm": 2.0720787048339844, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.7002225518226624, + "num_tokens": 211101921.0, + "step": 8271 + }, + { + "epoch": 0.9084120360202065, + "grad_norm": 2.143157720565796, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.6970149874687195, + "num_tokens": 211128730.0, + "step": 8272 + }, + { + "epoch": 0.9085218537228201, + "grad_norm": 2.0928680896759033, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7075484991073608, + "num_tokens": 211154927.0, + "step": 8273 + }, + { + "epoch": 0.9086316714254338, + "grad_norm": 2.202009439468384, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7214025259017944, + "num_tokens": 211178569.0, + "step": 8274 + }, + { + "epoch": 0.9087414891280474, + "grad_norm": 2.0602667331695557, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7113645076751709, + "num_tokens": 211207520.0, + "step": 8275 + }, + { + "epoch": 0.9088513068306611, + "grad_norm": 2.030757427215576, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6968877911567688, + "num_tokens": 211238990.0, + "step": 8276 + }, + { + "epoch": 0.9089611245332747, + "grad_norm": 2.3570539951324463, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7204061150550842, + "num_tokens": 211262000.0, + "step": 8277 + }, + { + "epoch": 0.9090709422358885, + "grad_norm": 2.043424367904663, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7060802578926086, + "num_tokens": 211289847.0, + "step": 8278 + }, + { + "epoch": 0.9091807599385021, + "grad_norm": 2.5305988788604736, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7093907594680786, + "num_tokens": 211311196.0, + "step": 8279 + }, + { + "epoch": 0.9092905776411158, + "grad_norm": 2.3051421642303467, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.7072254419326782, + "num_tokens": 211334694.0, + "step": 8280 + }, + { + "epoch": 0.9094003953437294, + "grad_norm": 2.138296127319336, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.704027533531189, + "num_tokens": 211362421.0, + "step": 8281 + }, + { + "epoch": 0.909510213046343, + "grad_norm": 1.9305921792984009, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7035975456237793, + "num_tokens": 211391122.0, + "step": 8282 + }, + { + "epoch": 0.9096200307489567, + "grad_norm": 2.552616596221924, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7110788226127625, + "num_tokens": 211410706.0, + "step": 8283 + }, + { + "epoch": 0.9097298484515703, + "grad_norm": 2.158454418182373, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.695012629032135, + "num_tokens": 211436905.0, + "step": 8284 + }, + { + "epoch": 0.9098396661541841, + "grad_norm": 2.4659271240234375, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7237289547920227, + "num_tokens": 211457444.0, + "step": 8285 + }, + { + "epoch": 0.9099494838567977, + "grad_norm": 2.5032968521118164, + "learning_rate": 1e-06, + "loss": 0.7698, + "mean_token_accuracy": 0.7516409158706665, + "num_tokens": 211475837.0, + "step": 8286 + }, + { + "epoch": 0.9100593015594114, + "grad_norm": 1.8707871437072754, + "learning_rate": 1e-06, + "loss": 1.0242, + "mean_token_accuracy": 0.6970337629318237, + "num_tokens": 211508584.0, + "step": 8287 + }, + { + "epoch": 0.910169119262025, + "grad_norm": 2.0626208782196045, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.6934213042259216, + "num_tokens": 211535706.0, + "step": 8288 + }, + { + "epoch": 0.9102789369646387, + "grad_norm": 2.2822117805480957, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7229098081588745, + "num_tokens": 211558734.0, + "step": 8289 + }, + { + "epoch": 0.9103887546672523, + "grad_norm": 2.088759422302246, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7195278406143188, + "num_tokens": 211585927.0, + "step": 8290 + }, + { + "epoch": 0.910498572369866, + "grad_norm": 2.3477277755737305, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7159441709518433, + "num_tokens": 211608020.0, + "step": 8291 + }, + { + "epoch": 0.9106083900724797, + "grad_norm": 2.062196731567383, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7149256467819214, + "num_tokens": 211634794.0, + "step": 8292 + }, + { + "epoch": 0.9107182077750934, + "grad_norm": 2.258047580718994, + "learning_rate": 1e-06, + "loss": 1.0708, + "mean_token_accuracy": 0.6740620136260986, + "num_tokens": 211661850.0, + "step": 8293 + }, + { + "epoch": 0.910828025477707, + "grad_norm": 2.2878997325897217, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7131130695343018, + "num_tokens": 211686284.0, + "step": 8294 + }, + { + "epoch": 0.9109378431803207, + "grad_norm": 1.995981216430664, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6942430734634399, + "num_tokens": 211714903.0, + "step": 8295 + }, + { + "epoch": 0.9110476608829343, + "grad_norm": 2.093993902206421, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7052192091941833, + "num_tokens": 211740526.0, + "step": 8296 + }, + { + "epoch": 0.911157478585548, + "grad_norm": 2.439347267150879, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7202556729316711, + "num_tokens": 211760646.0, + "step": 8297 + }, + { + "epoch": 0.9112672962881616, + "grad_norm": 3.0642614364624023, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7107282876968384, + "num_tokens": 211775464.0, + "step": 8298 + }, + { + "epoch": 0.9113771139907754, + "grad_norm": 2.1763505935668945, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7215341329574585, + "num_tokens": 211798874.0, + "step": 8299 + }, + { + "epoch": 0.911486931693389, + "grad_norm": 2.67916202545166, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7115668058395386, + "num_tokens": 211818892.0, + "step": 8300 + }, + { + "epoch": 0.9115967493960027, + "grad_norm": 2.233166456222534, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7123439311981201, + "num_tokens": 211845222.0, + "step": 8301 + }, + { + "epoch": 0.9117065670986163, + "grad_norm": 2.1533823013305664, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.6986585855484009, + "num_tokens": 211871690.0, + "step": 8302 + }, + { + "epoch": 0.91181638480123, + "grad_norm": 2.2991130352020264, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7161545157432556, + "num_tokens": 211893481.0, + "step": 8303 + }, + { + "epoch": 0.9119262025038436, + "grad_norm": 2.0210049152374268, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.6822459697723389, + "num_tokens": 211924406.0, + "step": 8304 + }, + { + "epoch": 0.9120360202064572, + "grad_norm": 2.145620822906494, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7047017812728882, + "num_tokens": 211952570.0, + "step": 8305 + }, + { + "epoch": 0.9121458379090709, + "grad_norm": 2.1543831825256348, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7218958139419556, + "num_tokens": 211977590.0, + "step": 8306 + }, + { + "epoch": 0.9122556556116846, + "grad_norm": 2.154322862625122, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7239116430282593, + "num_tokens": 212001218.0, + "step": 8307 + }, + { + "epoch": 0.9123654733142983, + "grad_norm": 2.105801582336426, + "learning_rate": 1e-06, + "loss": 1.0176, + "mean_token_accuracy": 0.6955063343048096, + "num_tokens": 212029686.0, + "step": 8308 + }, + { + "epoch": 0.9124752910169119, + "grad_norm": 1.9158053398132324, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7156819105148315, + "num_tokens": 212061173.0, + "step": 8309 + }, + { + "epoch": 0.9125851087195256, + "grad_norm": 2.5303850173950195, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7095681428909302, + "num_tokens": 212082039.0, + "step": 8310 + }, + { + "epoch": 0.9126949264221392, + "grad_norm": 2.0585827827453613, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.6970314979553223, + "num_tokens": 212111108.0, + "step": 8311 + }, + { + "epoch": 0.9128047441247529, + "grad_norm": 2.088921546936035, + "learning_rate": 1e-06, + "loss": 1.0215, + "mean_token_accuracy": 0.6918976306915283, + "num_tokens": 212141171.0, + "step": 8312 + }, + { + "epoch": 0.9129145618273665, + "grad_norm": 2.330498218536377, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7263288497924805, + "num_tokens": 212163263.0, + "step": 8313 + }, + { + "epoch": 0.9130243795299803, + "grad_norm": 2.1455421447753906, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.709229588508606, + "num_tokens": 212188650.0, + "step": 8314 + }, + { + "epoch": 0.9131341972325939, + "grad_norm": 2.1459527015686035, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.707878828048706, + "num_tokens": 212215395.0, + "step": 8315 + }, + { + "epoch": 0.9132440149352076, + "grad_norm": 2.206350564956665, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.686601459980011, + "num_tokens": 212242579.0, + "step": 8316 + }, + { + "epoch": 0.9133538326378212, + "grad_norm": 2.126461982727051, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6903077363967896, + "num_tokens": 212270528.0, + "step": 8317 + }, + { + "epoch": 0.9134636503404349, + "grad_norm": 2.005154609680176, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7110902667045593, + "num_tokens": 212298229.0, + "step": 8318 + }, + { + "epoch": 0.9135734680430485, + "grad_norm": 2.05237078666687, + "learning_rate": 1e-06, + "loss": 1.0455, + "mean_token_accuracy": 0.6811587810516357, + "num_tokens": 212326689.0, + "step": 8319 + }, + { + "epoch": 0.9136832857456622, + "grad_norm": 2.3620922565460205, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7106525897979736, + "num_tokens": 212349939.0, + "step": 8320 + }, + { + "epoch": 0.9137931034482759, + "grad_norm": 2.2350656986236572, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7080297470092773, + "num_tokens": 212373807.0, + "step": 8321 + }, + { + "epoch": 0.9139029211508896, + "grad_norm": 2.0043601989746094, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.718502938747406, + "num_tokens": 212402228.0, + "step": 8322 + }, + { + "epoch": 0.9140127388535032, + "grad_norm": 2.264307975769043, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7297515869140625, + "num_tokens": 212426186.0, + "step": 8323 + }, + { + "epoch": 0.9141225565561168, + "grad_norm": 2.2721264362335205, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7353257536888123, + "num_tokens": 212448653.0, + "step": 8324 + }, + { + "epoch": 0.9142323742587305, + "grad_norm": 2.083437442779541, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7199556827545166, + "num_tokens": 212476824.0, + "step": 8325 + }, + { + "epoch": 0.9143421919613441, + "grad_norm": 2.1071503162384033, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.6922178268432617, + "num_tokens": 212505176.0, + "step": 8326 + }, + { + "epoch": 0.9144520096639578, + "grad_norm": 2.08384370803833, + "learning_rate": 1e-06, + "loss": 1.0558, + "mean_token_accuracy": 0.681524395942688, + "num_tokens": 212532673.0, + "step": 8327 + }, + { + "epoch": 0.9145618273665715, + "grad_norm": 2.584014892578125, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7178802490234375, + "num_tokens": 212552401.0, + "step": 8328 + }, + { + "epoch": 0.9146716450691852, + "grad_norm": 1.999557375907898, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6900140643119812, + "num_tokens": 212581849.0, + "step": 8329 + }, + { + "epoch": 0.9147814627717988, + "grad_norm": 2.1086769104003906, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7047367691993713, + "num_tokens": 212608946.0, + "step": 8330 + }, + { + "epoch": 0.9148912804744125, + "grad_norm": 2.2333033084869385, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.702931821346283, + "num_tokens": 212635032.0, + "step": 8331 + }, + { + "epoch": 0.9150010981770261, + "grad_norm": 1.9287890195846558, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.7023352980613708, + "num_tokens": 212666129.0, + "step": 8332 + }, + { + "epoch": 0.9151109158796398, + "grad_norm": 1.9141526222229004, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.713402509689331, + "num_tokens": 212695287.0, + "step": 8333 + }, + { + "epoch": 0.9152207335822534, + "grad_norm": 2.1459028720855713, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7189053297042847, + "num_tokens": 212719583.0, + "step": 8334 + }, + { + "epoch": 0.9153305512848671, + "grad_norm": 1.9047327041625977, + "learning_rate": 1e-06, + "loss": 1.0336, + "mean_token_accuracy": 0.6852747201919556, + "num_tokens": 212752017.0, + "step": 8335 + }, + { + "epoch": 0.9154403689874808, + "grad_norm": 2.2665789127349854, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6964379549026489, + "num_tokens": 212776100.0, + "step": 8336 + }, + { + "epoch": 0.9155501866900945, + "grad_norm": 2.438495635986328, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.702163577079773, + "num_tokens": 212797715.0, + "step": 8337 + }, + { + "epoch": 0.9156600043927081, + "grad_norm": 1.9202629327774048, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7035583257675171, + "num_tokens": 212829490.0, + "step": 8338 + }, + { + "epoch": 0.9157698220953218, + "grad_norm": 2.382387161254883, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7004901170730591, + "num_tokens": 212852205.0, + "step": 8339 + }, + { + "epoch": 0.9158796397979354, + "grad_norm": 1.8204982280731201, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6993878483772278, + "num_tokens": 212888144.0, + "step": 8340 + }, + { + "epoch": 0.915989457500549, + "grad_norm": 2.0290629863739014, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7055354714393616, + "num_tokens": 212917961.0, + "step": 8341 + }, + { + "epoch": 0.9160992752031627, + "grad_norm": 1.9798518419265747, + "learning_rate": 1e-06, + "loss": 1.0171, + "mean_token_accuracy": 0.692610502243042, + "num_tokens": 212947265.0, + "step": 8342 + }, + { + "epoch": 0.9162090929057765, + "grad_norm": 2.1265933513641357, + "learning_rate": 1e-06, + "loss": 1.0419, + "mean_token_accuracy": 0.6871283054351807, + "num_tokens": 212975399.0, + "step": 8343 + }, + { + "epoch": 0.9163189106083901, + "grad_norm": 2.0387980937957764, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7059251070022583, + "num_tokens": 213004585.0, + "step": 8344 + }, + { + "epoch": 0.9164287283110037, + "grad_norm": 2.213420867919922, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7017802000045776, + "num_tokens": 213028679.0, + "step": 8345 + }, + { + "epoch": 0.9165385460136174, + "grad_norm": 2.2989397048950195, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7219126224517822, + "num_tokens": 213050657.0, + "step": 8346 + }, + { + "epoch": 0.916648363716231, + "grad_norm": 2.2938578128814697, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6908536553382874, + "num_tokens": 213076087.0, + "step": 8347 + }, + { + "epoch": 0.9167581814188447, + "grad_norm": 2.1278350353240967, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7129684686660767, + "num_tokens": 213101551.0, + "step": 8348 + }, + { + "epoch": 0.9168679991214583, + "grad_norm": 2.1336967945098877, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7137104272842407, + "num_tokens": 213128919.0, + "step": 8349 + }, + { + "epoch": 0.9169778168240721, + "grad_norm": 2.3610286712646484, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7066936492919922, + "num_tokens": 213150900.0, + "step": 8350 + }, + { + "epoch": 0.9170876345266857, + "grad_norm": 2.0390634536743164, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.7126575112342834, + "num_tokens": 213182278.0, + "step": 8351 + }, + { + "epoch": 0.9171974522292994, + "grad_norm": 2.400953769683838, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6956955790519714, + "num_tokens": 213203856.0, + "step": 8352 + }, + { + "epoch": 0.917307269931913, + "grad_norm": 2.530879259109497, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7186636924743652, + "num_tokens": 213224313.0, + "step": 8353 + }, + { + "epoch": 0.9174170876345267, + "grad_norm": 2.3844339847564697, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.6958911418914795, + "num_tokens": 213246845.0, + "step": 8354 + }, + { + "epoch": 0.9175269053371403, + "grad_norm": 2.4080331325531006, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7367621064186096, + "num_tokens": 213266390.0, + "step": 8355 + }, + { + "epoch": 0.917636723039754, + "grad_norm": 2.1789896488189697, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7152777910232544, + "num_tokens": 213292893.0, + "step": 8356 + }, + { + "epoch": 0.9177465407423677, + "grad_norm": 1.894437551498413, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7072444558143616, + "num_tokens": 213324580.0, + "step": 8357 + }, + { + "epoch": 0.9178563584449814, + "grad_norm": 2.2877750396728516, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7224216461181641, + "num_tokens": 213347642.0, + "step": 8358 + }, + { + "epoch": 0.917966176147595, + "grad_norm": 1.9569581747055054, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.6984306573867798, + "num_tokens": 213376605.0, + "step": 8359 + }, + { + "epoch": 0.9180759938502087, + "grad_norm": 2.519984722137451, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7199904918670654, + "num_tokens": 213398438.0, + "step": 8360 + }, + { + "epoch": 0.9181858115528223, + "grad_norm": 2.2204742431640625, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7075408697128296, + "num_tokens": 213424162.0, + "step": 8361 + }, + { + "epoch": 0.918295629255436, + "grad_norm": 2.2426414489746094, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.6991697549819946, + "num_tokens": 213447405.0, + "step": 8362 + }, + { + "epoch": 0.9184054469580496, + "grad_norm": 2.414438486099243, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7349652051925659, + "num_tokens": 213467353.0, + "step": 8363 + }, + { + "epoch": 0.9185152646606632, + "grad_norm": 2.4705445766448975, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7126070261001587, + "num_tokens": 213487502.0, + "step": 8364 + }, + { + "epoch": 0.918625082363277, + "grad_norm": 2.2577059268951416, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7091524600982666, + "num_tokens": 213509723.0, + "step": 8365 + }, + { + "epoch": 0.9187349000658906, + "grad_norm": 2.2623350620269775, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.711449146270752, + "num_tokens": 213533832.0, + "step": 8366 + }, + { + "epoch": 0.9188447177685043, + "grad_norm": 2.5818898677825928, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7223228216171265, + "num_tokens": 213552521.0, + "step": 8367 + }, + { + "epoch": 0.9189545354711179, + "grad_norm": 1.9770272970199585, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7008070945739746, + "num_tokens": 213581454.0, + "step": 8368 + }, + { + "epoch": 0.9190643531737316, + "grad_norm": 2.169977903366089, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7332228422164917, + "num_tokens": 213605509.0, + "step": 8369 + }, + { + "epoch": 0.9191741708763452, + "grad_norm": 1.885619878768921, + "learning_rate": 1e-06, + "loss": 1.0809, + "mean_token_accuracy": 0.6774609088897705, + "num_tokens": 213637361.0, + "step": 8370 + }, + { + "epoch": 0.9192839885789589, + "grad_norm": 2.300452470779419, + "learning_rate": 1e-06, + "loss": 0.7791, + "mean_token_accuracy": 0.7466049194335938, + "num_tokens": 213657788.0, + "step": 8371 + }, + { + "epoch": 0.9193938062815726, + "grad_norm": 2.1686782836914062, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7064914107322693, + "num_tokens": 213682085.0, + "step": 8372 + }, + { + "epoch": 0.9195036239841863, + "grad_norm": 2.1052358150482178, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7350763082504272, + "num_tokens": 213708330.0, + "step": 8373 + }, + { + "epoch": 0.9196134416867999, + "grad_norm": 2.43979811668396, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7132563591003418, + "num_tokens": 213728488.0, + "step": 8374 + }, + { + "epoch": 0.9197232593894136, + "grad_norm": 2.220099449157715, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7254796028137207, + "num_tokens": 213752748.0, + "step": 8375 + }, + { + "epoch": 0.9198330770920272, + "grad_norm": 1.904819130897522, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6930553913116455, + "num_tokens": 213785347.0, + "step": 8376 + }, + { + "epoch": 0.9199428947946409, + "grad_norm": 2.4257593154907227, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6995053291320801, + "num_tokens": 213807507.0, + "step": 8377 + }, + { + "epoch": 0.9200527124972545, + "grad_norm": 1.9609894752502441, + "learning_rate": 1e-06, + "loss": 1.0383, + "mean_token_accuracy": 0.6885578632354736, + "num_tokens": 213840365.0, + "step": 8378 + }, + { + "epoch": 0.9201625301998683, + "grad_norm": 2.521724224090576, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7279043793678284, + "num_tokens": 213859845.0, + "step": 8379 + }, + { + "epoch": 0.9202723479024819, + "grad_norm": 2.08194637298584, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.694718599319458, + "num_tokens": 213885738.0, + "step": 8380 + }, + { + "epoch": 0.9203821656050956, + "grad_norm": 2.085777759552002, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7227410078048706, + "num_tokens": 213911154.0, + "step": 8381 + }, + { + "epoch": 0.9204919833077092, + "grad_norm": 2.3452670574188232, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7229097485542297, + "num_tokens": 213934463.0, + "step": 8382 + }, + { + "epoch": 0.9206018010103229, + "grad_norm": 2.3641695976257324, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.7016350626945496, + "num_tokens": 213957465.0, + "step": 8383 + }, + { + "epoch": 0.9207116187129365, + "grad_norm": 2.162703275680542, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7221775650978088, + "num_tokens": 213981981.0, + "step": 8384 + }, + { + "epoch": 0.9208214364155501, + "grad_norm": 2.0190422534942627, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7291320562362671, + "num_tokens": 214010226.0, + "step": 8385 + }, + { + "epoch": 0.9209312541181639, + "grad_norm": 2.2506895065307617, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7062283754348755, + "num_tokens": 214034740.0, + "step": 8386 + }, + { + "epoch": 0.9210410718207775, + "grad_norm": 2.108816623687744, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7166247963905334, + "num_tokens": 214062435.0, + "step": 8387 + }, + { + "epoch": 0.9211508895233912, + "grad_norm": 2.422699451446533, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.7020070552825928, + "num_tokens": 214084161.0, + "step": 8388 + }, + { + "epoch": 0.9212607072260048, + "grad_norm": 1.9576612710952759, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7363048791885376, + "num_tokens": 214112324.0, + "step": 8389 + }, + { + "epoch": 0.9213705249286185, + "grad_norm": 2.218428134918213, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7276781797409058, + "num_tokens": 214136439.0, + "step": 8390 + }, + { + "epoch": 0.9214803426312321, + "grad_norm": 2.076856851577759, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7110897302627563, + "num_tokens": 214163840.0, + "step": 8391 + }, + { + "epoch": 0.9215901603338458, + "grad_norm": 2.158221960067749, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7331667542457581, + "num_tokens": 214188114.0, + "step": 8392 + }, + { + "epoch": 0.9216999780364594, + "grad_norm": 2.182833671569824, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7288022041320801, + "num_tokens": 214212363.0, + "step": 8393 + }, + { + "epoch": 0.9218097957390732, + "grad_norm": 2.12890887260437, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7096186280250549, + "num_tokens": 214240614.0, + "step": 8394 + }, + { + "epoch": 0.9219196134416868, + "grad_norm": 2.239081382751465, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7205668687820435, + "num_tokens": 214263868.0, + "step": 8395 + }, + { + "epoch": 0.9220294311443005, + "grad_norm": 2.1529603004455566, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7111189365386963, + "num_tokens": 214291519.0, + "step": 8396 + }, + { + "epoch": 0.9221392488469141, + "grad_norm": 2.0769145488739014, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7071716785430908, + "num_tokens": 214319747.0, + "step": 8397 + }, + { + "epoch": 0.9222490665495278, + "grad_norm": 2.298482656478882, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7280474901199341, + "num_tokens": 214341318.0, + "step": 8398 + }, + { + "epoch": 0.9223588842521414, + "grad_norm": 2.2542808055877686, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.6963822841644287, + "num_tokens": 214366997.0, + "step": 8399 + }, + { + "epoch": 0.922468701954755, + "grad_norm": 1.8898707628250122, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.694068431854248, + "num_tokens": 214399531.0, + "step": 8400 + }, + { + "epoch": 0.9225785196573688, + "grad_norm": 1.9823060035705566, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6894912123680115, + "num_tokens": 214428616.0, + "step": 8401 + }, + { + "epoch": 0.9226883373599825, + "grad_norm": 2.0869667530059814, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.6999845504760742, + "num_tokens": 214458795.0, + "step": 8402 + }, + { + "epoch": 0.9227981550625961, + "grad_norm": 2.1160330772399902, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7160391807556152, + "num_tokens": 214483024.0, + "step": 8403 + }, + { + "epoch": 0.9229079727652097, + "grad_norm": 2.079893112182617, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.7029410004615784, + "num_tokens": 214510821.0, + "step": 8404 + }, + { + "epoch": 0.9230177904678234, + "grad_norm": 2.271484375, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7037338018417358, + "num_tokens": 214535846.0, + "step": 8405 + }, + { + "epoch": 0.923127608170437, + "grad_norm": 2.096660614013672, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7199727296829224, + "num_tokens": 214561912.0, + "step": 8406 + }, + { + "epoch": 0.9232374258730507, + "grad_norm": 2.091057538986206, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7038453817367554, + "num_tokens": 214587549.0, + "step": 8407 + }, + { + "epoch": 0.9233472435756644, + "grad_norm": 2.2305173873901367, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.715096116065979, + "num_tokens": 214612345.0, + "step": 8408 + }, + { + "epoch": 0.9234570612782781, + "grad_norm": 2.3199660778045654, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.730501651763916, + "num_tokens": 214634176.0, + "step": 8409 + }, + { + "epoch": 0.9235668789808917, + "grad_norm": 2.035203456878662, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.6974825859069824, + "num_tokens": 214662299.0, + "step": 8410 + }, + { + "epoch": 0.9236766966835054, + "grad_norm": 2.1140406131744385, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.6931149363517761, + "num_tokens": 214690475.0, + "step": 8411 + }, + { + "epoch": 0.923786514386119, + "grad_norm": 2.1121318340301514, + "learning_rate": 1e-06, + "loss": 1.0379, + "mean_token_accuracy": 0.6871031522750854, + "num_tokens": 214721864.0, + "step": 8412 + }, + { + "epoch": 0.9238963320887327, + "grad_norm": 2.2516379356384277, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7305277585983276, + "num_tokens": 214745784.0, + "step": 8413 + }, + { + "epoch": 0.9240061497913463, + "grad_norm": 2.026318073272705, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7153361439704895, + "num_tokens": 214774358.0, + "step": 8414 + }, + { + "epoch": 0.9241159674939601, + "grad_norm": 1.9069476127624512, + "learning_rate": 1e-06, + "loss": 1.0512, + "mean_token_accuracy": 0.6803654432296753, + "num_tokens": 214807537.0, + "step": 8415 + }, + { + "epoch": 0.9242257851965737, + "grad_norm": 1.9247806072235107, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7061808705329895, + "num_tokens": 214839209.0, + "step": 8416 + }, + { + "epoch": 0.9243356028991874, + "grad_norm": 2.398928165435791, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7246302962303162, + "num_tokens": 214860685.0, + "step": 8417 + }, + { + "epoch": 0.924445420601801, + "grad_norm": 2.210447072982788, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6877113580703735, + "num_tokens": 214886777.0, + "step": 8418 + }, + { + "epoch": 0.9245552383044147, + "grad_norm": 2.3614020347595215, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7151522040367126, + "num_tokens": 214908974.0, + "step": 8419 + }, + { + "epoch": 0.9246650560070283, + "grad_norm": 2.3001439571380615, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7100983262062073, + "num_tokens": 214931744.0, + "step": 8420 + }, + { + "epoch": 0.924774873709642, + "grad_norm": 2.7166144847869873, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7181712985038757, + "num_tokens": 214949605.0, + "step": 8421 + }, + { + "epoch": 0.9248846914122557, + "grad_norm": 2.2087254524230957, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7045843601226807, + "num_tokens": 214974742.0, + "step": 8422 + }, + { + "epoch": 0.9249945091148694, + "grad_norm": 2.129824638366699, + "learning_rate": 1e-06, + "loss": 1.1031, + "mean_token_accuracy": 0.6841108798980713, + "num_tokens": 215004179.0, + "step": 8423 + }, + { + "epoch": 0.925104326817483, + "grad_norm": 2.069216728210449, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6914982795715332, + "num_tokens": 215035493.0, + "step": 8424 + }, + { + "epoch": 0.9252141445200966, + "grad_norm": 2.365708112716675, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7073568105697632, + "num_tokens": 215057714.0, + "step": 8425 + }, + { + "epoch": 0.9253239622227103, + "grad_norm": 2.255955696105957, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7043967247009277, + "num_tokens": 215081410.0, + "step": 8426 + }, + { + "epoch": 0.9254337799253239, + "grad_norm": 2.09295654296875, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7047778367996216, + "num_tokens": 215107047.0, + "step": 8427 + }, + { + "epoch": 0.9255435976279376, + "grad_norm": 1.9669623374938965, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7119672298431396, + "num_tokens": 215136876.0, + "step": 8428 + }, + { + "epoch": 0.9256534153305512, + "grad_norm": 2.2480990886688232, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6998227834701538, + "num_tokens": 215161306.0, + "step": 8429 + }, + { + "epoch": 0.925763233033165, + "grad_norm": 2.0423099994659424, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7109910249710083, + "num_tokens": 215190480.0, + "step": 8430 + }, + { + "epoch": 0.9258730507357786, + "grad_norm": 2.1395440101623535, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6996259689331055, + "num_tokens": 215218319.0, + "step": 8431 + }, + { + "epoch": 0.9259828684383923, + "grad_norm": 2.0093462467193604, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6957112550735474, + "num_tokens": 215251303.0, + "step": 8432 + }, + { + "epoch": 0.9260926861410059, + "grad_norm": 2.2215311527252197, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.726292073726654, + "num_tokens": 215274890.0, + "step": 8433 + }, + { + "epoch": 0.9262025038436196, + "grad_norm": 2.140413284301758, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7233786582946777, + "num_tokens": 215301806.0, + "step": 8434 + }, + { + "epoch": 0.9263123215462332, + "grad_norm": 2.0406274795532227, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7124922275543213, + "num_tokens": 215330957.0, + "step": 8435 + }, + { + "epoch": 0.9264221392488469, + "grad_norm": 2.141556978225708, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6916356086730957, + "num_tokens": 215356242.0, + "step": 8436 + }, + { + "epoch": 0.9265319569514606, + "grad_norm": 2.2088828086853027, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7098362445831299, + "num_tokens": 215382328.0, + "step": 8437 + }, + { + "epoch": 0.9266417746540743, + "grad_norm": 1.8250919580459595, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.693108320236206, + "num_tokens": 215418383.0, + "step": 8438 + }, + { + "epoch": 0.9267515923566879, + "grad_norm": 2.2147762775421143, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6958614587783813, + "num_tokens": 215442661.0, + "step": 8439 + }, + { + "epoch": 0.9268614100593016, + "grad_norm": 2.3394386768341064, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7411806583404541, + "num_tokens": 215465101.0, + "step": 8440 + }, + { + "epoch": 0.9269712277619152, + "grad_norm": 2.0107624530792236, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6986751556396484, + "num_tokens": 215494565.0, + "step": 8441 + }, + { + "epoch": 0.9270810454645289, + "grad_norm": 2.2781808376312256, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7138856649398804, + "num_tokens": 215516761.0, + "step": 8442 + }, + { + "epoch": 0.9271908631671425, + "grad_norm": 2.080850601196289, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7284931540489197, + "num_tokens": 215539956.0, + "step": 8443 + }, + { + "epoch": 0.9273006808697563, + "grad_norm": 2.1822164058685303, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6906856298446655, + "num_tokens": 215568709.0, + "step": 8444 + }, + { + "epoch": 0.9274104985723699, + "grad_norm": 1.882538080215454, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6907998323440552, + "num_tokens": 215600397.0, + "step": 8445 + }, + { + "epoch": 0.9275203162749835, + "grad_norm": 2.5524768829345703, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7137298583984375, + "num_tokens": 215619554.0, + "step": 8446 + }, + { + "epoch": 0.9276301339775972, + "grad_norm": 2.19663405418396, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.7018147706985474, + "num_tokens": 215647334.0, + "step": 8447 + }, + { + "epoch": 0.9277399516802108, + "grad_norm": 2.429211139678955, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7272450923919678, + "num_tokens": 215667067.0, + "step": 8448 + }, + { + "epoch": 0.9278497693828245, + "grad_norm": 2.011035442352295, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6887767910957336, + "num_tokens": 215696119.0, + "step": 8449 + }, + { + "epoch": 0.9279595870854381, + "grad_norm": 2.2074368000030518, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7279943227767944, + "num_tokens": 215720102.0, + "step": 8450 + }, + { + "epoch": 0.9280694047880519, + "grad_norm": 2.237123727798462, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6861600875854492, + "num_tokens": 215744653.0, + "step": 8451 + }, + { + "epoch": 0.9281792224906655, + "grad_norm": 2.0655179023742676, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7037402987480164, + "num_tokens": 215772943.0, + "step": 8452 + }, + { + "epoch": 0.9282890401932792, + "grad_norm": 2.149576187133789, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7074093222618103, + "num_tokens": 215797930.0, + "step": 8453 + }, + { + "epoch": 0.9283988578958928, + "grad_norm": 2.285159111022949, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7159515023231506, + "num_tokens": 215821391.0, + "step": 8454 + }, + { + "epoch": 0.9285086755985065, + "grad_norm": 2.06793212890625, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7001134753227234, + "num_tokens": 215848095.0, + "step": 8455 + }, + { + "epoch": 0.9286184933011201, + "grad_norm": 1.7461129426956177, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6920055150985718, + "num_tokens": 215886030.0, + "step": 8456 + }, + { + "epoch": 0.9287283110037338, + "grad_norm": 2.0965943336486816, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7101453542709351, + "num_tokens": 215913664.0, + "step": 8457 + }, + { + "epoch": 0.9288381287063474, + "grad_norm": 2.4426865577697754, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7159963846206665, + "num_tokens": 215935343.0, + "step": 8458 + }, + { + "epoch": 0.9289479464089612, + "grad_norm": 1.948893427848816, + "learning_rate": 1e-06, + "loss": 1.06, + "mean_token_accuracy": 0.6773845553398132, + "num_tokens": 215966014.0, + "step": 8459 + }, + { + "epoch": 0.9290577641115748, + "grad_norm": 2.344890594482422, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7185444831848145, + "num_tokens": 215990196.0, + "step": 8460 + }, + { + "epoch": 0.9291675818141885, + "grad_norm": 1.988707423210144, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.721922755241394, + "num_tokens": 216018558.0, + "step": 8461 + }, + { + "epoch": 0.9292773995168021, + "grad_norm": 2.0565989017486572, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7230352163314819, + "num_tokens": 216047607.0, + "step": 8462 + }, + { + "epoch": 0.9293872172194158, + "grad_norm": 2.362830638885498, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7235507965087891, + "num_tokens": 216068972.0, + "step": 8463 + }, + { + "epoch": 0.9294970349220294, + "grad_norm": 2.5663199424743652, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7023080587387085, + "num_tokens": 216088724.0, + "step": 8464 + }, + { + "epoch": 0.929606852624643, + "grad_norm": 1.8272022008895874, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.725956916809082, + "num_tokens": 216123286.0, + "step": 8465 + }, + { + "epoch": 0.9297166703272568, + "grad_norm": 2.0360827445983887, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7075285911560059, + "num_tokens": 216153242.0, + "step": 8466 + }, + { + "epoch": 0.9298264880298704, + "grad_norm": 2.160935878753662, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7157936692237854, + "num_tokens": 216177578.0, + "step": 8467 + }, + { + "epoch": 0.9299363057324841, + "grad_norm": 2.306581497192383, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7011682987213135, + "num_tokens": 216199852.0, + "step": 8468 + }, + { + "epoch": 0.9300461234350977, + "grad_norm": 2.2251474857330322, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7222947478294373, + "num_tokens": 216223446.0, + "step": 8469 + }, + { + "epoch": 0.9301559411377114, + "grad_norm": 2.199587106704712, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7040667533874512, + "num_tokens": 216249632.0, + "step": 8470 + }, + { + "epoch": 0.930265758840325, + "grad_norm": 2.084115982055664, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7154611945152283, + "num_tokens": 216277422.0, + "step": 8471 + }, + { + "epoch": 0.9303755765429387, + "grad_norm": 2.2948644161224365, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7042496204376221, + "num_tokens": 216301902.0, + "step": 8472 + }, + { + "epoch": 0.9304853942455524, + "grad_norm": 2.122004985809326, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.6947088241577148, + "num_tokens": 216328696.0, + "step": 8473 + }, + { + "epoch": 0.9305952119481661, + "grad_norm": 2.532905340194702, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7297479510307312, + "num_tokens": 216347281.0, + "step": 8474 + }, + { + "epoch": 0.9307050296507797, + "grad_norm": 2.132350206375122, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7103763818740845, + "num_tokens": 216372558.0, + "step": 8475 + }, + { + "epoch": 0.9308148473533934, + "grad_norm": 2.4980409145355225, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7329940795898438, + "num_tokens": 216391593.0, + "step": 8476 + }, + { + "epoch": 0.930924665056007, + "grad_norm": 2.5865585803985596, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7116478681564331, + "num_tokens": 216410648.0, + "step": 8477 + }, + { + "epoch": 0.9310344827586207, + "grad_norm": 2.682191848754883, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7184603214263916, + "num_tokens": 216428503.0, + "step": 8478 + }, + { + "epoch": 0.9311443004612343, + "grad_norm": 2.1472599506378174, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7088438868522644, + "num_tokens": 216456126.0, + "step": 8479 + }, + { + "epoch": 0.9312541181638481, + "grad_norm": 2.81563663482666, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7185719609260559, + "num_tokens": 216475691.0, + "step": 8480 + }, + { + "epoch": 0.9313639358664617, + "grad_norm": 2.2150261402130127, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.6950466632843018, + "num_tokens": 216502374.0, + "step": 8481 + }, + { + "epoch": 0.9314737535690754, + "grad_norm": 2.3368821144104004, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7022532224655151, + "num_tokens": 216529327.0, + "step": 8482 + }, + { + "epoch": 0.931583571271689, + "grad_norm": 2.419309139251709, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.6983139514923096, + "num_tokens": 216551175.0, + "step": 8483 + }, + { + "epoch": 0.9316933889743026, + "grad_norm": 2.0082104206085205, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7121033072471619, + "num_tokens": 216578925.0, + "step": 8484 + }, + { + "epoch": 0.9318032066769163, + "grad_norm": 2.0467536449432373, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7126511335372925, + "num_tokens": 216605848.0, + "step": 8485 + }, + { + "epoch": 0.9319130243795299, + "grad_norm": 2.1253068447113037, + "learning_rate": 1e-06, + "loss": 1.0335, + "mean_token_accuracy": 0.6820560693740845, + "num_tokens": 216633235.0, + "step": 8486 + }, + { + "epoch": 0.9320228420821436, + "grad_norm": 2.1095774173736572, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7001311779022217, + "num_tokens": 216660357.0, + "step": 8487 + }, + { + "epoch": 0.9321326597847573, + "grad_norm": 2.143068552017212, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6985708475112915, + "num_tokens": 216685997.0, + "step": 8488 + }, + { + "epoch": 0.932242477487371, + "grad_norm": 2.262014627456665, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7074061632156372, + "num_tokens": 216709501.0, + "step": 8489 + }, + { + "epoch": 0.9323522951899846, + "grad_norm": 2.4179272651672363, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6959177851676941, + "num_tokens": 216731001.0, + "step": 8490 + }, + { + "epoch": 0.9324621128925983, + "grad_norm": 2.1054346561431885, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7107164263725281, + "num_tokens": 216755292.0, + "step": 8491 + }, + { + "epoch": 0.9325719305952119, + "grad_norm": 2.3127048015594482, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7022457122802734, + "num_tokens": 216778456.0, + "step": 8492 + }, + { + "epoch": 0.9326817482978256, + "grad_norm": 2.055546522140503, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7129397392272949, + "num_tokens": 216805942.0, + "step": 8493 + }, + { + "epoch": 0.9327915660004392, + "grad_norm": 2.1090025901794434, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.6974849700927734, + "num_tokens": 216833013.0, + "step": 8494 + }, + { + "epoch": 0.932901383703053, + "grad_norm": 1.9852327108383179, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7311383485794067, + "num_tokens": 216860890.0, + "step": 8495 + }, + { + "epoch": 0.9330112014056666, + "grad_norm": 2.2882537841796875, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7004841566085815, + "num_tokens": 216884560.0, + "step": 8496 + }, + { + "epoch": 0.9331210191082803, + "grad_norm": 2.1535606384277344, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7190954685211182, + "num_tokens": 216909617.0, + "step": 8497 + }, + { + "epoch": 0.9332308368108939, + "grad_norm": 2.3272457122802734, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7283172011375427, + "num_tokens": 216931971.0, + "step": 8498 + }, + { + "epoch": 0.9333406545135076, + "grad_norm": 2.3264200687408447, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7235502004623413, + "num_tokens": 216954516.0, + "step": 8499 + }, + { + "epoch": 0.9334504722161212, + "grad_norm": 2.2515318393707275, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7060353755950928, + "num_tokens": 216979940.0, + "step": 8500 + }, + { + "epoch": 0.9335602899187349, + "grad_norm": 1.9338369369506836, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.6969103217124939, + "num_tokens": 217010342.0, + "step": 8501 + }, + { + "epoch": 0.9336701076213486, + "grad_norm": 2.2543182373046875, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.7003663778305054, + "num_tokens": 217033500.0, + "step": 8502 + }, + { + "epoch": 0.9337799253239623, + "grad_norm": 2.0881340503692627, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7188190221786499, + "num_tokens": 217060033.0, + "step": 8503 + }, + { + "epoch": 0.9338897430265759, + "grad_norm": 2.1840333938598633, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7058807015419006, + "num_tokens": 217086583.0, + "step": 8504 + }, + { + "epoch": 0.9339995607291895, + "grad_norm": 2.065948486328125, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7061682343482971, + "num_tokens": 217112947.0, + "step": 8505 + }, + { + "epoch": 0.9341093784318032, + "grad_norm": 2.363814353942871, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7075064778327942, + "num_tokens": 217135388.0, + "step": 8506 + }, + { + "epoch": 0.9342191961344168, + "grad_norm": 2.128495454788208, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7127020359039307, + "num_tokens": 217161723.0, + "step": 8507 + }, + { + "epoch": 0.9343290138370305, + "grad_norm": 2.5701892375946045, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.711992621421814, + "num_tokens": 217181247.0, + "step": 8508 + }, + { + "epoch": 0.9344388315396442, + "grad_norm": 2.3961386680603027, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.723851203918457, + "num_tokens": 217203703.0, + "step": 8509 + }, + { + "epoch": 0.9345486492422579, + "grad_norm": 2.4741358757019043, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7138334512710571, + "num_tokens": 217223426.0, + "step": 8510 + }, + { + "epoch": 0.9346584669448715, + "grad_norm": 2.206158399581909, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7189136743545532, + "num_tokens": 217246758.0, + "step": 8511 + }, + { + "epoch": 0.9347682846474852, + "grad_norm": 2.2694499492645264, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7099325656890869, + "num_tokens": 217270417.0, + "step": 8512 + }, + { + "epoch": 0.9348781023500988, + "grad_norm": 2.168477773666382, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6968157887458801, + "num_tokens": 217294971.0, + "step": 8513 + }, + { + "epoch": 0.9349879200527125, + "grad_norm": 2.1004726886749268, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7036998271942139, + "num_tokens": 217323559.0, + "step": 8514 + }, + { + "epoch": 0.9350977377553261, + "grad_norm": 2.247040271759033, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.72528475522995, + "num_tokens": 217344054.0, + "step": 8515 + }, + { + "epoch": 0.9352075554579398, + "grad_norm": 1.9054348468780518, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.6877012252807617, + "num_tokens": 217376153.0, + "step": 8516 + }, + { + "epoch": 0.9353173731605535, + "grad_norm": 2.133709669113159, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.702567458152771, + "num_tokens": 217401918.0, + "step": 8517 + }, + { + "epoch": 0.9354271908631672, + "grad_norm": 2.257859230041504, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7191601991653442, + "num_tokens": 217426471.0, + "step": 8518 + }, + { + "epoch": 0.9355370085657808, + "grad_norm": 2.3094170093536377, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7144559025764465, + "num_tokens": 217450863.0, + "step": 8519 + }, + { + "epoch": 0.9356468262683945, + "grad_norm": 2.213419198989868, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7064778804779053, + "num_tokens": 217475719.0, + "step": 8520 + }, + { + "epoch": 0.9357566439710081, + "grad_norm": 2.361616611480713, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7199090719223022, + "num_tokens": 217496547.0, + "step": 8521 + }, + { + "epoch": 0.9358664616736218, + "grad_norm": 2.239036798477173, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7132837176322937, + "num_tokens": 217521141.0, + "step": 8522 + }, + { + "epoch": 0.9359762793762354, + "grad_norm": 2.047525405883789, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.703275740146637, + "num_tokens": 217549363.0, + "step": 8523 + }, + { + "epoch": 0.9360860970788492, + "grad_norm": 2.575857639312744, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7226386070251465, + "num_tokens": 217566493.0, + "step": 8524 + }, + { + "epoch": 0.9361959147814628, + "grad_norm": 1.9040758609771729, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7216005325317383, + "num_tokens": 217596854.0, + "step": 8525 + }, + { + "epoch": 0.9363057324840764, + "grad_norm": 2.0985870361328125, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.708330512046814, + "num_tokens": 217623217.0, + "step": 8526 + }, + { + "epoch": 0.9364155501866901, + "grad_norm": 2.175095796585083, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7270174026489258, + "num_tokens": 217645639.0, + "step": 8527 + }, + { + "epoch": 0.9365253678893037, + "grad_norm": 2.23384690284729, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7367574572563171, + "num_tokens": 217668225.0, + "step": 8528 + }, + { + "epoch": 0.9366351855919174, + "grad_norm": 2.2812061309814453, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7384519577026367, + "num_tokens": 217689322.0, + "step": 8529 + }, + { + "epoch": 0.936745003294531, + "grad_norm": 2.4332053661346436, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7215837240219116, + "num_tokens": 217710539.0, + "step": 8530 + }, + { + "epoch": 0.9368548209971448, + "grad_norm": 2.116989850997925, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.6902544498443604, + "num_tokens": 217739126.0, + "step": 8531 + }, + { + "epoch": 0.9369646386997584, + "grad_norm": 2.0859386920928955, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.701947808265686, + "num_tokens": 217767175.0, + "step": 8532 + }, + { + "epoch": 0.9370744564023721, + "grad_norm": 2.5850298404693604, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7097587585449219, + "num_tokens": 217785840.0, + "step": 8533 + }, + { + "epoch": 0.9371842741049857, + "grad_norm": 2.3098082542419434, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7091278433799744, + "num_tokens": 217808720.0, + "step": 8534 + }, + { + "epoch": 0.9372940918075994, + "grad_norm": 2.1808433532714844, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7055712938308716, + "num_tokens": 217834696.0, + "step": 8535 + }, + { + "epoch": 0.937403909510213, + "grad_norm": 1.9422770738601685, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7198500037193298, + "num_tokens": 217864899.0, + "step": 8536 + }, + { + "epoch": 0.9375137272128267, + "grad_norm": 2.00053071975708, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.6948242783546448, + "num_tokens": 217893651.0, + "step": 8537 + }, + { + "epoch": 0.9376235449154404, + "grad_norm": 2.050565719604492, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7028346061706543, + "num_tokens": 217922287.0, + "step": 8538 + }, + { + "epoch": 0.9377333626180541, + "grad_norm": 2.1513826847076416, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7001310586929321, + "num_tokens": 217949198.0, + "step": 8539 + }, + { + "epoch": 0.9378431803206677, + "grad_norm": 2.0515899658203125, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.721419095993042, + "num_tokens": 217975728.0, + "step": 8540 + }, + { + "epoch": 0.9379529980232814, + "grad_norm": 2.204420566558838, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7138937711715698, + "num_tokens": 217999658.0, + "step": 8541 + }, + { + "epoch": 0.938062815725895, + "grad_norm": 2.415066957473755, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7160628437995911, + "num_tokens": 218022597.0, + "step": 8542 + }, + { + "epoch": 0.9381726334285087, + "grad_norm": 2.331920862197876, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7117179036140442, + "num_tokens": 218046588.0, + "step": 8543 + }, + { + "epoch": 0.9382824511311223, + "grad_norm": 2.033862829208374, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.7002218961715698, + "num_tokens": 218075863.0, + "step": 8544 + }, + { + "epoch": 0.9383922688337359, + "grad_norm": 2.0960896015167236, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7110883593559265, + "num_tokens": 218101555.0, + "step": 8545 + }, + { + "epoch": 0.9385020865363497, + "grad_norm": 2.0982086658477783, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7090096473693848, + "num_tokens": 218128112.0, + "step": 8546 + }, + { + "epoch": 0.9386119042389633, + "grad_norm": 1.8990932703018188, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6931975483894348, + "num_tokens": 218159365.0, + "step": 8547 + }, + { + "epoch": 0.938721721941577, + "grad_norm": 2.2783725261688232, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.7036289572715759, + "num_tokens": 218184272.0, + "step": 8548 + }, + { + "epoch": 0.9388315396441906, + "grad_norm": 2.2880454063415527, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7114766836166382, + "num_tokens": 218207229.0, + "step": 8549 + }, + { + "epoch": 0.9389413573468043, + "grad_norm": 2.1696767807006836, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6965943574905396, + "num_tokens": 218234046.0, + "step": 8550 + }, + { + "epoch": 0.9390511750494179, + "grad_norm": 2.075713872909546, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7348619103431702, + "num_tokens": 218260099.0, + "step": 8551 + }, + { + "epoch": 0.9391609927520316, + "grad_norm": 2.3398756980895996, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.726137638092041, + "num_tokens": 218282782.0, + "step": 8552 + }, + { + "epoch": 0.9392708104546453, + "grad_norm": 2.630725860595703, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.7004582285881042, + "num_tokens": 218303239.0, + "step": 8553 + }, + { + "epoch": 0.939380628157259, + "grad_norm": 2.7157297134399414, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6984829902648926, + "num_tokens": 218321232.0, + "step": 8554 + }, + { + "epoch": 0.9394904458598726, + "grad_norm": 2.227207899093628, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7083302736282349, + "num_tokens": 218347016.0, + "step": 8555 + }, + { + "epoch": 0.9396002635624863, + "grad_norm": 2.166046380996704, + "learning_rate": 1e-06, + "loss": 1.0319, + "mean_token_accuracy": 0.6951406002044678, + "num_tokens": 218372986.0, + "step": 8556 + }, + { + "epoch": 0.9397100812650999, + "grad_norm": 2.092221260070801, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7096916437149048, + "num_tokens": 218401575.0, + "step": 8557 + }, + { + "epoch": 0.9398198989677136, + "grad_norm": 2.143263101577759, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.6904865503311157, + "num_tokens": 218426968.0, + "step": 8558 + }, + { + "epoch": 0.9399297166703272, + "grad_norm": 2.5247998237609863, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7266963720321655, + "num_tokens": 218446119.0, + "step": 8559 + }, + { + "epoch": 0.940039534372941, + "grad_norm": 2.2392115592956543, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.7025919556617737, + "num_tokens": 218470105.0, + "step": 8560 + }, + { + "epoch": 0.9401493520755546, + "grad_norm": 2.1370840072631836, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7167835235595703, + "num_tokens": 218495738.0, + "step": 8561 + }, + { + "epoch": 0.9402591697781683, + "grad_norm": 2.031480312347412, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.6981209516525269, + "num_tokens": 218525729.0, + "step": 8562 + }, + { + "epoch": 0.9403689874807819, + "grad_norm": 2.334139823913574, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7198789119720459, + "num_tokens": 218547392.0, + "step": 8563 + }, + { + "epoch": 0.9404788051833955, + "grad_norm": 2.0052547454833984, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7099617123603821, + "num_tokens": 218575240.0, + "step": 8564 + }, + { + "epoch": 0.9405886228860092, + "grad_norm": 2.034135341644287, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7112017273902893, + "num_tokens": 218605277.0, + "step": 8565 + }, + { + "epoch": 0.9406984405886228, + "grad_norm": 2.322112560272217, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7235931754112244, + "num_tokens": 218627227.0, + "step": 8566 + }, + { + "epoch": 0.9408082582912366, + "grad_norm": 2.2332472801208496, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7236067652702332, + "num_tokens": 218650465.0, + "step": 8567 + }, + { + "epoch": 0.9409180759938502, + "grad_norm": 1.9952629804611206, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7080435752868652, + "num_tokens": 218680351.0, + "step": 8568 + }, + { + "epoch": 0.9410278936964639, + "grad_norm": 2.0826029777526855, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7057170271873474, + "num_tokens": 218708747.0, + "step": 8569 + }, + { + "epoch": 0.9411377113990775, + "grad_norm": 2.3949177265167236, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7219180464744568, + "num_tokens": 218729900.0, + "step": 8570 + }, + { + "epoch": 0.9412475291016912, + "grad_norm": 2.098552942276001, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7255859375, + "num_tokens": 218756684.0, + "step": 8571 + }, + { + "epoch": 0.9413573468043048, + "grad_norm": 2.2514443397521973, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6946207880973816, + "num_tokens": 218782086.0, + "step": 8572 + }, + { + "epoch": 0.9414671645069185, + "grad_norm": 2.2309725284576416, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7339487671852112, + "num_tokens": 218803918.0, + "step": 8573 + }, + { + "epoch": 0.9415769822095322, + "grad_norm": 2.085115909576416, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7141127586364746, + "num_tokens": 218833469.0, + "step": 8574 + }, + { + "epoch": 0.9416867999121459, + "grad_norm": 2.0872364044189453, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.6989383697509766, + "num_tokens": 218861367.0, + "step": 8575 + }, + { + "epoch": 0.9417966176147595, + "grad_norm": 2.060171604156494, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7087597846984863, + "num_tokens": 218890392.0, + "step": 8576 + }, + { + "epoch": 0.9419064353173732, + "grad_norm": 2.087045907974243, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7061073780059814, + "num_tokens": 218917333.0, + "step": 8577 + }, + { + "epoch": 0.9420162530199868, + "grad_norm": 2.455336809158325, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7200405597686768, + "num_tokens": 218937281.0, + "step": 8578 + }, + { + "epoch": 0.9421260707226005, + "grad_norm": 2.166252374649048, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.723686695098877, + "num_tokens": 218963911.0, + "step": 8579 + }, + { + "epoch": 0.9422358884252141, + "grad_norm": 2.3843274116516113, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7135295867919922, + "num_tokens": 218986318.0, + "step": 8580 + }, + { + "epoch": 0.9423457061278278, + "grad_norm": 2.1400368213653564, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7125172019004822, + "num_tokens": 219011684.0, + "step": 8581 + }, + { + "epoch": 0.9424555238304415, + "grad_norm": 2.1554672718048096, + "learning_rate": 1e-06, + "loss": 1.0543, + "mean_token_accuracy": 0.6847252249717712, + "num_tokens": 219038213.0, + "step": 8582 + }, + { + "epoch": 0.9425653415330552, + "grad_norm": 2.6312735080718994, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7161153554916382, + "num_tokens": 219056190.0, + "step": 8583 + }, + { + "epoch": 0.9426751592356688, + "grad_norm": 2.031561851501465, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6913613677024841, + "num_tokens": 219084956.0, + "step": 8584 + }, + { + "epoch": 0.9427849769382824, + "grad_norm": 2.39150333404541, + "learning_rate": 1e-06, + "loss": 0.7909, + "mean_token_accuracy": 0.7502939105033875, + "num_tokens": 219105655.0, + "step": 8585 + }, + { + "epoch": 0.9428947946408961, + "grad_norm": 1.9293363094329834, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.7005786299705505, + "num_tokens": 219138166.0, + "step": 8586 + }, + { + "epoch": 0.9430046123435097, + "grad_norm": 2.1193881034851074, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.711219072341919, + "num_tokens": 219163151.0, + "step": 8587 + }, + { + "epoch": 0.9431144300461234, + "grad_norm": 2.379781723022461, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.722184956073761, + "num_tokens": 219185878.0, + "step": 8588 + }, + { + "epoch": 0.9432242477487371, + "grad_norm": 2.467970132827759, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7399083375930786, + "num_tokens": 219204780.0, + "step": 8589 + }, + { + "epoch": 0.9433340654513508, + "grad_norm": 2.417726755142212, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7188007235527039, + "num_tokens": 219226790.0, + "step": 8590 + }, + { + "epoch": 0.9434438831539644, + "grad_norm": 2.214489221572876, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7060050964355469, + "num_tokens": 219251088.0, + "step": 8591 + }, + { + "epoch": 0.9435537008565781, + "grad_norm": 2.194814920425415, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7074215412139893, + "num_tokens": 219274404.0, + "step": 8592 + }, + { + "epoch": 0.9436635185591917, + "grad_norm": 2.1919009685516357, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7312175035476685, + "num_tokens": 219297866.0, + "step": 8593 + }, + { + "epoch": 0.9437733362618054, + "grad_norm": 1.8411246538162231, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.6956613063812256, + "num_tokens": 219331599.0, + "step": 8594 + }, + { + "epoch": 0.943883153964419, + "grad_norm": 2.051424980163574, + "learning_rate": 1e-06, + "loss": 1.0224, + "mean_token_accuracy": 0.6908859014511108, + "num_tokens": 219358851.0, + "step": 8595 + }, + { + "epoch": 0.9439929716670328, + "grad_norm": 2.2149953842163086, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7168461084365845, + "num_tokens": 219383957.0, + "step": 8596 + }, + { + "epoch": 0.9441027893696464, + "grad_norm": 1.785890817642212, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7229273915290833, + "num_tokens": 219415280.0, + "step": 8597 + }, + { + "epoch": 0.9442126070722601, + "grad_norm": 2.2332701683044434, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7308433055877686, + "num_tokens": 219441121.0, + "step": 8598 + }, + { + "epoch": 0.9443224247748737, + "grad_norm": 2.2713499069213867, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7029282450675964, + "num_tokens": 219464167.0, + "step": 8599 + }, + { + "epoch": 0.9444322424774874, + "grad_norm": 2.2220709323883057, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6957723498344421, + "num_tokens": 219491109.0, + "step": 8600 + }, + { + "epoch": 0.944542060180101, + "grad_norm": 2.264390468597412, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7033125162124634, + "num_tokens": 219514210.0, + "step": 8601 + }, + { + "epoch": 0.9446518778827147, + "grad_norm": 2.081725835800171, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.7083835601806641, + "num_tokens": 219542200.0, + "step": 8602 + }, + { + "epoch": 0.9447616955853284, + "grad_norm": 2.2746267318725586, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7168479561805725, + "num_tokens": 219566288.0, + "step": 8603 + }, + { + "epoch": 0.944871513287942, + "grad_norm": 2.1874566078186035, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7215070128440857, + "num_tokens": 219590278.0, + "step": 8604 + }, + { + "epoch": 0.9449813309905557, + "grad_norm": 2.066490888595581, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6991218328475952, + "num_tokens": 219616881.0, + "step": 8605 + }, + { + "epoch": 0.9450911486931693, + "grad_norm": 2.131430149078369, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.6923836469650269, + "num_tokens": 219645104.0, + "step": 8606 + }, + { + "epoch": 0.945200966395783, + "grad_norm": 2.166863203048706, + "learning_rate": 1e-06, + "loss": 1.0405, + "mean_token_accuracy": 0.6847954392433167, + "num_tokens": 219673357.0, + "step": 8607 + }, + { + "epoch": 0.9453107840983966, + "grad_norm": 1.7741727828979492, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.7008776664733887, + "num_tokens": 219706684.0, + "step": 8608 + }, + { + "epoch": 0.9454206018010103, + "grad_norm": 2.1047279834747314, + "learning_rate": 1e-06, + "loss": 1.0117, + "mean_token_accuracy": 0.6897896528244019, + "num_tokens": 219735097.0, + "step": 8609 + }, + { + "epoch": 0.9455304195036239, + "grad_norm": 2.1550240516662598, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6844642758369446, + "num_tokens": 219761343.0, + "step": 8610 + }, + { + "epoch": 0.9456402372062377, + "grad_norm": 1.8880363702774048, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7114353775978088, + "num_tokens": 219791899.0, + "step": 8611 + }, + { + "epoch": 0.9457500549088513, + "grad_norm": 2.0467581748962402, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7304077744483948, + "num_tokens": 219816324.0, + "step": 8612 + }, + { + "epoch": 0.945859872611465, + "grad_norm": 2.2383244037628174, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7027089595794678, + "num_tokens": 219839842.0, + "step": 8613 + }, + { + "epoch": 0.9459696903140786, + "grad_norm": 2.7685773372650146, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7191560864448547, + "num_tokens": 219856932.0, + "step": 8614 + }, + { + "epoch": 0.9460795080166923, + "grad_norm": 2.1409847736358643, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6958246231079102, + "num_tokens": 219881389.0, + "step": 8615 + }, + { + "epoch": 0.9461893257193059, + "grad_norm": 2.221078872680664, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.694350004196167, + "num_tokens": 219906351.0, + "step": 8616 + }, + { + "epoch": 0.9462991434219196, + "grad_norm": 2.3705828189849854, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7212262153625488, + "num_tokens": 219926412.0, + "step": 8617 + }, + { + "epoch": 0.9464089611245333, + "grad_norm": 2.0502209663391113, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6989743113517761, + "num_tokens": 219953777.0, + "step": 8618 + }, + { + "epoch": 0.946518778827147, + "grad_norm": 2.1820566654205322, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7206276655197144, + "num_tokens": 219977873.0, + "step": 8619 + }, + { + "epoch": 0.9466285965297606, + "grad_norm": 2.0579609870910645, + "learning_rate": 1e-06, + "loss": 1.0316, + "mean_token_accuracy": 0.6820821762084961, + "num_tokens": 220006326.0, + "step": 8620 + }, + { + "epoch": 0.9467384142323743, + "grad_norm": 1.829089879989624, + "learning_rate": 1e-06, + "loss": 1.0584, + "mean_token_accuracy": 0.6858638525009155, + "num_tokens": 220043816.0, + "step": 8621 + }, + { + "epoch": 0.9468482319349879, + "grad_norm": 2.571895122528076, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7016730904579163, + "num_tokens": 220064832.0, + "step": 8622 + }, + { + "epoch": 0.9469580496376016, + "grad_norm": 2.160079002380371, + "learning_rate": 1e-06, + "loss": 1.0339, + "mean_token_accuracy": 0.6879224181175232, + "num_tokens": 220092866.0, + "step": 8623 + }, + { + "epoch": 0.9470678673402152, + "grad_norm": 2.140138626098633, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7142794132232666, + "num_tokens": 220119155.0, + "step": 8624 + }, + { + "epoch": 0.947177685042829, + "grad_norm": 2.186548948287964, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7184346914291382, + "num_tokens": 220144864.0, + "step": 8625 + }, + { + "epoch": 0.9472875027454426, + "grad_norm": 2.122774362564087, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7032655477523804, + "num_tokens": 220172827.0, + "step": 8626 + }, + { + "epoch": 0.9473973204480562, + "grad_norm": 2.319791316986084, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7333937287330627, + "num_tokens": 220195196.0, + "step": 8627 + }, + { + "epoch": 0.9475071381506699, + "grad_norm": 2.2746312618255615, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7201368808746338, + "num_tokens": 220219277.0, + "step": 8628 + }, + { + "epoch": 0.9476169558532835, + "grad_norm": 2.186271905899048, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7156203985214233, + "num_tokens": 220245602.0, + "step": 8629 + }, + { + "epoch": 0.9477267735558972, + "grad_norm": 2.1767024993896484, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7061435580253601, + "num_tokens": 220271863.0, + "step": 8630 + }, + { + "epoch": 0.9478365912585108, + "grad_norm": 2.1609301567077637, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6977181434631348, + "num_tokens": 220298030.0, + "step": 8631 + }, + { + "epoch": 0.9479464089611246, + "grad_norm": 2.5855226516723633, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7116914987564087, + "num_tokens": 220317505.0, + "step": 8632 + }, + { + "epoch": 0.9480562266637382, + "grad_norm": 2.5317063331604004, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.732011079788208, + "num_tokens": 220337968.0, + "step": 8633 + }, + { + "epoch": 0.9481660443663519, + "grad_norm": 2.087895393371582, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7169080972671509, + "num_tokens": 220363783.0, + "step": 8634 + }, + { + "epoch": 0.9482758620689655, + "grad_norm": 2.2208638191223145, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7122008800506592, + "num_tokens": 220387848.0, + "step": 8635 + }, + { + "epoch": 0.9483856797715792, + "grad_norm": 2.3243279457092285, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.718276858329773, + "num_tokens": 220409929.0, + "step": 8636 + }, + { + "epoch": 0.9484954974741928, + "grad_norm": 2.314509153366089, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7344451546669006, + "num_tokens": 220431367.0, + "step": 8637 + }, + { + "epoch": 0.9486053151768065, + "grad_norm": 2.173914670944214, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7046269774436951, + "num_tokens": 220457412.0, + "step": 8638 + }, + { + "epoch": 0.9487151328794201, + "grad_norm": 2.1355364322662354, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7285354137420654, + "num_tokens": 220481197.0, + "step": 8639 + }, + { + "epoch": 0.9488249505820339, + "grad_norm": 2.609722375869751, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7445050477981567, + "num_tokens": 220499754.0, + "step": 8640 + }, + { + "epoch": 0.9489347682846475, + "grad_norm": 2.5899791717529297, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.6996261477470398, + "num_tokens": 220520749.0, + "step": 8641 + }, + { + "epoch": 0.9490445859872612, + "grad_norm": 2.3454861640930176, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7112560272216797, + "num_tokens": 220542085.0, + "step": 8642 + }, + { + "epoch": 0.9491544036898748, + "grad_norm": 2.0716259479522705, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7210484743118286, + "num_tokens": 220567590.0, + "step": 8643 + }, + { + "epoch": 0.9492642213924884, + "grad_norm": 2.0916428565979004, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7067216634750366, + "num_tokens": 220597290.0, + "step": 8644 + }, + { + "epoch": 0.9493740390951021, + "grad_norm": 2.126844882965088, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.6979368925094604, + "num_tokens": 220624272.0, + "step": 8645 + }, + { + "epoch": 0.9494838567977157, + "grad_norm": 2.2002413272857666, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7210560441017151, + "num_tokens": 220647073.0, + "step": 8646 + }, + { + "epoch": 0.9495936745003295, + "grad_norm": 2.145198345184326, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7139451503753662, + "num_tokens": 220673142.0, + "step": 8647 + }, + { + "epoch": 0.9497034922029431, + "grad_norm": 2.36418080329895, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7073875069618225, + "num_tokens": 220694889.0, + "step": 8648 + }, + { + "epoch": 0.9498133099055568, + "grad_norm": 2.0115926265716553, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6877247095108032, + "num_tokens": 220722902.0, + "step": 8649 + }, + { + "epoch": 0.9499231276081704, + "grad_norm": 1.9863147735595703, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.7028337121009827, + "num_tokens": 220752480.0, + "step": 8650 + }, + { + "epoch": 0.9500329453107841, + "grad_norm": 2.2941603660583496, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.6971136927604675, + "num_tokens": 220777228.0, + "step": 8651 + }, + { + "epoch": 0.9501427630133977, + "grad_norm": 2.297391891479492, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7167295217514038, + "num_tokens": 220797604.0, + "step": 8652 + }, + { + "epoch": 0.9502525807160114, + "grad_norm": 2.252345323562622, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.717881441116333, + "num_tokens": 220821418.0, + "step": 8653 + }, + { + "epoch": 0.9503623984186251, + "grad_norm": 1.832460880279541, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.6980973482131958, + "num_tokens": 220853979.0, + "step": 8654 + }, + { + "epoch": 0.9504722161212388, + "grad_norm": 2.0834832191467285, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7074445486068726, + "num_tokens": 220881631.0, + "step": 8655 + }, + { + "epoch": 0.9505820338238524, + "grad_norm": 2.222083806991577, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.6982506513595581, + "num_tokens": 220906992.0, + "step": 8656 + }, + { + "epoch": 0.9506918515264661, + "grad_norm": 1.9207119941711426, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.6971917748451233, + "num_tokens": 220938604.0, + "step": 8657 + }, + { + "epoch": 0.9508016692290797, + "grad_norm": 2.1201529502868652, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.698249101638794, + "num_tokens": 220964484.0, + "step": 8658 + }, + { + "epoch": 0.9509114869316934, + "grad_norm": 2.289429187774658, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7304847836494446, + "num_tokens": 220986548.0, + "step": 8659 + }, + { + "epoch": 0.951021304634307, + "grad_norm": 2.252581834793091, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.720065176486969, + "num_tokens": 221008940.0, + "step": 8660 + }, + { + "epoch": 0.9511311223369208, + "grad_norm": 2.1070053577423096, + "learning_rate": 1e-06, + "loss": 1.0246, + "mean_token_accuracy": 0.6953792572021484, + "num_tokens": 221034766.0, + "step": 8661 + }, + { + "epoch": 0.9512409400395344, + "grad_norm": 2.0852341651916504, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7215209007263184, + "num_tokens": 221058435.0, + "step": 8662 + }, + { + "epoch": 0.9513507577421481, + "grad_norm": 2.186948537826538, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7051090002059937, + "num_tokens": 221083326.0, + "step": 8663 + }, + { + "epoch": 0.9514605754447617, + "grad_norm": 2.5230817794799805, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7317796945571899, + "num_tokens": 221101547.0, + "step": 8664 + }, + { + "epoch": 0.9515703931473753, + "grad_norm": 2.1745145320892334, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6880077123641968, + "num_tokens": 221127742.0, + "step": 8665 + }, + { + "epoch": 0.951680210849989, + "grad_norm": 2.1387572288513184, + "learning_rate": 1e-06, + "loss": 1.0737, + "mean_token_accuracy": 0.6765680313110352, + "num_tokens": 221155242.0, + "step": 8666 + }, + { + "epoch": 0.9517900285526026, + "grad_norm": 2.3719913959503174, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7059853076934814, + "num_tokens": 221179768.0, + "step": 8667 + }, + { + "epoch": 0.9518998462552163, + "grad_norm": 2.189349412918091, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.710357666015625, + "num_tokens": 221204801.0, + "step": 8668 + }, + { + "epoch": 0.95200966395783, + "grad_norm": 2.3179996013641357, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7132331132888794, + "num_tokens": 221228303.0, + "step": 8669 + }, + { + "epoch": 0.9521194816604437, + "grad_norm": 2.1038520336151123, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7094510793685913, + "num_tokens": 221256438.0, + "step": 8670 + }, + { + "epoch": 0.9522292993630573, + "grad_norm": 2.0624279975891113, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7147833108901978, + "num_tokens": 221285448.0, + "step": 8671 + }, + { + "epoch": 0.952339117065671, + "grad_norm": 2.3024520874023438, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7070479393005371, + "num_tokens": 221311057.0, + "step": 8672 + }, + { + "epoch": 0.9524489347682846, + "grad_norm": 2.1735661029815674, + "learning_rate": 1e-06, + "loss": 1.0171, + "mean_token_accuracy": 0.6995929479598999, + "num_tokens": 221338498.0, + "step": 8673 + }, + { + "epoch": 0.9525587524708983, + "grad_norm": 2.2373383045196533, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7096055746078491, + "num_tokens": 221360284.0, + "step": 8674 + }, + { + "epoch": 0.9526685701735119, + "grad_norm": 2.0714645385742188, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7305640578269958, + "num_tokens": 221384532.0, + "step": 8675 + }, + { + "epoch": 0.9527783878761257, + "grad_norm": 1.8942224979400635, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.6969813108444214, + "num_tokens": 221417181.0, + "step": 8676 + }, + { + "epoch": 0.9528882055787393, + "grad_norm": 2.4063262939453125, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7268462777137756, + "num_tokens": 221438410.0, + "step": 8677 + }, + { + "epoch": 0.952998023281353, + "grad_norm": 2.3916878700256348, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7312813997268677, + "num_tokens": 221459942.0, + "step": 8678 + }, + { + "epoch": 0.9531078409839666, + "grad_norm": 1.8212586641311646, + "learning_rate": 1e-06, + "loss": 1.0347, + "mean_token_accuracy": 0.6906419992446899, + "num_tokens": 221494704.0, + "step": 8679 + }, + { + "epoch": 0.9532176586865803, + "grad_norm": 2.1698124408721924, + "learning_rate": 1e-06, + "loss": 1.0837, + "mean_token_accuracy": 0.6696339249610901, + "num_tokens": 221523770.0, + "step": 8680 + }, + { + "epoch": 0.9533274763891939, + "grad_norm": 1.916598916053772, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7058666944503784, + "num_tokens": 221555462.0, + "step": 8681 + }, + { + "epoch": 0.9534372940918076, + "grad_norm": 2.2410590648651123, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7399367690086365, + "num_tokens": 221580761.0, + "step": 8682 + }, + { + "epoch": 0.9535471117944213, + "grad_norm": 2.0407750606536865, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6902109980583191, + "num_tokens": 221608734.0, + "step": 8683 + }, + { + "epoch": 0.953656929497035, + "grad_norm": 2.159227132797241, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7012941837310791, + "num_tokens": 221635346.0, + "step": 8684 + }, + { + "epoch": 0.9537667471996486, + "grad_norm": 2.040581464767456, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7070230841636658, + "num_tokens": 221663156.0, + "step": 8685 + }, + { + "epoch": 0.9538765649022622, + "grad_norm": 2.0907702445983887, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.6956931948661804, + "num_tokens": 221691679.0, + "step": 8686 + }, + { + "epoch": 0.9539863826048759, + "grad_norm": 2.143540143966675, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7115568518638611, + "num_tokens": 221717939.0, + "step": 8687 + }, + { + "epoch": 0.9540962003074895, + "grad_norm": 2.171359062194824, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7015765309333801, + "num_tokens": 221743179.0, + "step": 8688 + }, + { + "epoch": 0.9542060180101032, + "grad_norm": 2.0647950172424316, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.6934472322463989, + "num_tokens": 221771182.0, + "step": 8689 + }, + { + "epoch": 0.9543158357127169, + "grad_norm": 2.5975098609924316, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7108919024467468, + "num_tokens": 221790695.0, + "step": 8690 + }, + { + "epoch": 0.9544256534153306, + "grad_norm": 2.2346351146698, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6970741748809814, + "num_tokens": 221814424.0, + "step": 8691 + }, + { + "epoch": 0.9545354711179442, + "grad_norm": 2.6768133640289307, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7201762199401855, + "num_tokens": 221832403.0, + "step": 8692 + }, + { + "epoch": 0.9546452888205579, + "grad_norm": 2.295077085494995, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7323719263076782, + "num_tokens": 221853854.0, + "step": 8693 + }, + { + "epoch": 0.9547551065231715, + "grad_norm": 2.5267691612243652, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7117346525192261, + "num_tokens": 221875150.0, + "step": 8694 + }, + { + "epoch": 0.9548649242257852, + "grad_norm": 2.2679262161254883, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.711240291595459, + "num_tokens": 221899863.0, + "step": 8695 + }, + { + "epoch": 0.9549747419283988, + "grad_norm": 2.2869508266448975, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7221674919128418, + "num_tokens": 221921990.0, + "step": 8696 + }, + { + "epoch": 0.9550845596310125, + "grad_norm": 2.3959078788757324, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7244336605072021, + "num_tokens": 221944202.0, + "step": 8697 + }, + { + "epoch": 0.9551943773336262, + "grad_norm": 2.3625998497009277, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7274023294448853, + "num_tokens": 221966632.0, + "step": 8698 + }, + { + "epoch": 0.9553041950362399, + "grad_norm": 2.178860664367676, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6904670000076294, + "num_tokens": 221994000.0, + "step": 8699 + }, + { + "epoch": 0.9554140127388535, + "grad_norm": 2.330095052719116, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7322397232055664, + "num_tokens": 222017562.0, + "step": 8700 + }, + { + "epoch": 0.9555238304414672, + "grad_norm": 2.2458767890930176, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7146336436271667, + "num_tokens": 222039993.0, + "step": 8701 + }, + { + "epoch": 0.9556336481440808, + "grad_norm": 2.3335812091827393, + "learning_rate": 1e-06, + "loss": 1.0779, + "mean_token_accuracy": 0.6853237152099609, + "num_tokens": 222064579.0, + "step": 8702 + }, + { + "epoch": 0.9557434658466945, + "grad_norm": 2.0768487453460693, + "learning_rate": 1e-06, + "loss": 1.0224, + "mean_token_accuracy": 0.684894323348999, + "num_tokens": 222091758.0, + "step": 8703 + }, + { + "epoch": 0.9558532835493081, + "grad_norm": 2.112977981567383, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6920530796051025, + "num_tokens": 222119960.0, + "step": 8704 + }, + { + "epoch": 0.9559631012519219, + "grad_norm": 2.105860471725464, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7187336683273315, + "num_tokens": 222147608.0, + "step": 8705 + }, + { + "epoch": 0.9560729189545355, + "grad_norm": 2.0502350330352783, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.716922402381897, + "num_tokens": 222174159.0, + "step": 8706 + }, + { + "epoch": 0.9561827366571491, + "grad_norm": 1.985117793083191, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7231689691543579, + "num_tokens": 222202518.0, + "step": 8707 + }, + { + "epoch": 0.9562925543597628, + "grad_norm": 2.1959118843078613, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7191447615623474, + "num_tokens": 222226087.0, + "step": 8708 + }, + { + "epoch": 0.9564023720623764, + "grad_norm": 1.9569138288497925, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.714411735534668, + "num_tokens": 222256094.0, + "step": 8709 + }, + { + "epoch": 0.9565121897649901, + "grad_norm": 2.1546711921691895, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7161803841590881, + "num_tokens": 222280941.0, + "step": 8710 + }, + { + "epoch": 0.9566220074676037, + "grad_norm": 1.8857953548431396, + "learning_rate": 1e-06, + "loss": 1.0904, + "mean_token_accuracy": 0.6735670566558838, + "num_tokens": 222315596.0, + "step": 8711 + }, + { + "epoch": 0.9567318251702175, + "grad_norm": 2.214221715927124, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7463635206222534, + "num_tokens": 222338194.0, + "step": 8712 + }, + { + "epoch": 0.9568416428728311, + "grad_norm": 2.4550158977508545, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7147763967514038, + "num_tokens": 222360193.0, + "step": 8713 + }, + { + "epoch": 0.9569514605754448, + "grad_norm": 2.1768198013305664, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.704016923904419, + "num_tokens": 222386906.0, + "step": 8714 + }, + { + "epoch": 0.9570612782780584, + "grad_norm": 2.2659013271331787, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.725431501865387, + "num_tokens": 222409965.0, + "step": 8715 + }, + { + "epoch": 0.9571710959806721, + "grad_norm": 2.0191991329193115, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7411296963691711, + "num_tokens": 222437289.0, + "step": 8716 + }, + { + "epoch": 0.9572809136832857, + "grad_norm": 2.1801228523254395, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7222452759742737, + "num_tokens": 222462345.0, + "step": 8717 + }, + { + "epoch": 0.9573907313858994, + "grad_norm": 2.455744743347168, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7092323899269104, + "num_tokens": 222485049.0, + "step": 8718 + }, + { + "epoch": 0.9575005490885131, + "grad_norm": 2.2453157901763916, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6929717063903809, + "num_tokens": 222509470.0, + "step": 8719 + }, + { + "epoch": 0.9576103667911268, + "grad_norm": 1.9729368686676025, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7096437215805054, + "num_tokens": 222536903.0, + "step": 8720 + }, + { + "epoch": 0.9577201844937404, + "grad_norm": 1.9552587270736694, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7115012407302856, + "num_tokens": 222564187.0, + "step": 8721 + }, + { + "epoch": 0.9578300021963541, + "grad_norm": 2.148049831390381, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7264915704727173, + "num_tokens": 222587578.0, + "step": 8722 + }, + { + "epoch": 0.9579398198989677, + "grad_norm": 2.1945488452911377, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.7039288282394409, + "num_tokens": 222614436.0, + "step": 8723 + }, + { + "epoch": 0.9580496376015813, + "grad_norm": 1.9283500909805298, + "learning_rate": 1e-06, + "loss": 1.0486, + "mean_token_accuracy": 0.6822260618209839, + "num_tokens": 222648140.0, + "step": 8724 + }, + { + "epoch": 0.958159455304195, + "grad_norm": 2.2720143795013428, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7240278720855713, + "num_tokens": 222670473.0, + "step": 8725 + }, + { + "epoch": 0.9582692730068088, + "grad_norm": 2.240154981613159, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6901217103004456, + "num_tokens": 222698048.0, + "step": 8726 + }, + { + "epoch": 0.9583790907094224, + "grad_norm": 2.133117437362671, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6851299405097961, + "num_tokens": 222726360.0, + "step": 8727 + }, + { + "epoch": 0.958488908412036, + "grad_norm": 2.139941453933716, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7254366278648376, + "num_tokens": 222750652.0, + "step": 8728 + }, + { + "epoch": 0.9585987261146497, + "grad_norm": 2.0863709449768066, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7201766967773438, + "num_tokens": 222777658.0, + "step": 8729 + }, + { + "epoch": 0.9587085438172633, + "grad_norm": 2.166703701019287, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7043581008911133, + "num_tokens": 222805667.0, + "step": 8730 + }, + { + "epoch": 0.958818361519877, + "grad_norm": 2.089047431945801, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7063390016555786, + "num_tokens": 222832299.0, + "step": 8731 + }, + { + "epoch": 0.9589281792224906, + "grad_norm": 2.264293909072876, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7106234431266785, + "num_tokens": 222857480.0, + "step": 8732 + }, + { + "epoch": 0.9590379969251043, + "grad_norm": 2.061842203140259, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7152190208435059, + "num_tokens": 222886101.0, + "step": 8733 + }, + { + "epoch": 0.959147814627718, + "grad_norm": 2.091866970062256, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7102278470993042, + "num_tokens": 222911921.0, + "step": 8734 + }, + { + "epoch": 0.9592576323303317, + "grad_norm": 2.197356939315796, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7064248323440552, + "num_tokens": 222936357.0, + "step": 8735 + }, + { + "epoch": 0.9593674500329453, + "grad_norm": 1.8812527656555176, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.6942832469940186, + "num_tokens": 222971798.0, + "step": 8736 + }, + { + "epoch": 0.959477267735559, + "grad_norm": 1.9226658344268799, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.6985350847244263, + "num_tokens": 223003685.0, + "step": 8737 + }, + { + "epoch": 0.9595870854381726, + "grad_norm": 1.9449069499969482, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7285248637199402, + "num_tokens": 223033454.0, + "step": 8738 + }, + { + "epoch": 0.9596969031407863, + "grad_norm": 1.8906855583190918, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.6855433583259583, + "num_tokens": 223065101.0, + "step": 8739 + }, + { + "epoch": 0.9598067208433999, + "grad_norm": 2.3386943340301514, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7187973260879517, + "num_tokens": 223088836.0, + "step": 8740 + }, + { + "epoch": 0.9599165385460137, + "grad_norm": 2.4538767337799072, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7226240038871765, + "num_tokens": 223110285.0, + "step": 8741 + }, + { + "epoch": 0.9600263562486273, + "grad_norm": 2.2358014583587646, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7078943252563477, + "num_tokens": 223134400.0, + "step": 8742 + }, + { + "epoch": 0.960136173951241, + "grad_norm": 2.2501182556152344, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.7020338177680969, + "num_tokens": 223161256.0, + "step": 8743 + }, + { + "epoch": 0.9602459916538546, + "grad_norm": 2.3083677291870117, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7266069650650024, + "num_tokens": 223183125.0, + "step": 8744 + }, + { + "epoch": 0.9603558093564682, + "grad_norm": 2.179347515106201, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.732505202293396, + "num_tokens": 223207548.0, + "step": 8745 + }, + { + "epoch": 0.9604656270590819, + "grad_norm": 2.432936668395996, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7271538376808167, + "num_tokens": 223226314.0, + "step": 8746 + }, + { + "epoch": 0.9605754447616955, + "grad_norm": 2.292675018310547, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7181584239006042, + "num_tokens": 223249544.0, + "step": 8747 + }, + { + "epoch": 0.9606852624643093, + "grad_norm": 2.207514762878418, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.701622724533081, + "num_tokens": 223275628.0, + "step": 8748 + }, + { + "epoch": 0.9607950801669229, + "grad_norm": 2.1720380783081055, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7041014432907104, + "num_tokens": 223301846.0, + "step": 8749 + }, + { + "epoch": 0.9609048978695366, + "grad_norm": 2.2914772033691406, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7063391208648682, + "num_tokens": 223324495.0, + "step": 8750 + }, + { + "epoch": 0.9610147155721502, + "grad_norm": 2.2645621299743652, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7204249501228333, + "num_tokens": 223346584.0, + "step": 8751 + }, + { + "epoch": 0.9611245332747639, + "grad_norm": 2.3457391262054443, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6970192193984985, + "num_tokens": 223369861.0, + "step": 8752 + }, + { + "epoch": 0.9612343509773775, + "grad_norm": 2.311511754989624, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.727501630783081, + "num_tokens": 223392536.0, + "step": 8753 + }, + { + "epoch": 0.9613441686799912, + "grad_norm": 2.4187939167022705, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7155158519744873, + "num_tokens": 223414058.0, + "step": 8754 + }, + { + "epoch": 0.9614539863826049, + "grad_norm": 2.21565842628479, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.711663007736206, + "num_tokens": 223437525.0, + "step": 8755 + }, + { + "epoch": 0.9615638040852186, + "grad_norm": 2.0078506469726562, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7486205697059631, + "num_tokens": 223463691.0, + "step": 8756 + }, + { + "epoch": 0.9616736217878322, + "grad_norm": 1.9946677684783936, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7103163003921509, + "num_tokens": 223493914.0, + "step": 8757 + }, + { + "epoch": 0.9617834394904459, + "grad_norm": 2.1050615310668945, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.696605920791626, + "num_tokens": 223518351.0, + "step": 8758 + }, + { + "epoch": 0.9618932571930595, + "grad_norm": 2.184901714324951, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6918042898178101, + "num_tokens": 223544377.0, + "step": 8759 + }, + { + "epoch": 0.9620030748956732, + "grad_norm": 1.9173613786697388, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.724083423614502, + "num_tokens": 223572366.0, + "step": 8760 + }, + { + "epoch": 0.9621128925982868, + "grad_norm": 2.2556748390197754, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7123781442642212, + "num_tokens": 223595939.0, + "step": 8761 + }, + { + "epoch": 0.9622227103009005, + "grad_norm": 1.9239479303359985, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7044776678085327, + "num_tokens": 223628638.0, + "step": 8762 + }, + { + "epoch": 0.9623325280035142, + "grad_norm": 2.1373002529144287, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.691106915473938, + "num_tokens": 223654394.0, + "step": 8763 + }, + { + "epoch": 0.9624423457061279, + "grad_norm": 2.238028049468994, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7062580585479736, + "num_tokens": 223678614.0, + "step": 8764 + }, + { + "epoch": 0.9625521634087415, + "grad_norm": 2.163996934890747, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7302609086036682, + "num_tokens": 223702813.0, + "step": 8765 + }, + { + "epoch": 0.9626619811113551, + "grad_norm": 2.346323251724243, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.6977870464324951, + "num_tokens": 223726006.0, + "step": 8766 + }, + { + "epoch": 0.9627717988139688, + "grad_norm": 2.3775861263275146, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.723991334438324, + "num_tokens": 223747327.0, + "step": 8767 + }, + { + "epoch": 0.9628816165165824, + "grad_norm": 2.114952802658081, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7049672603607178, + "num_tokens": 223774320.0, + "step": 8768 + }, + { + "epoch": 0.9629914342191961, + "grad_norm": 1.8997056484222412, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7087074518203735, + "num_tokens": 223806585.0, + "step": 8769 + }, + { + "epoch": 0.9631012519218098, + "grad_norm": 2.101803779602051, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7160564661026001, + "num_tokens": 223833534.0, + "step": 8770 + }, + { + "epoch": 0.9632110696244235, + "grad_norm": 2.2920827865600586, + "learning_rate": 1e-06, + "loss": 1.0603, + "mean_token_accuracy": 0.6844518184661865, + "num_tokens": 223859474.0, + "step": 8771 + }, + { + "epoch": 0.9633208873270371, + "grad_norm": 2.047243595123291, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7077826261520386, + "num_tokens": 223888730.0, + "step": 8772 + }, + { + "epoch": 0.9634307050296508, + "grad_norm": 2.446911334991455, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7114867568016052, + "num_tokens": 223911368.0, + "step": 8773 + }, + { + "epoch": 0.9635405227322644, + "grad_norm": 2.060242176055908, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7059764266014099, + "num_tokens": 223938741.0, + "step": 8774 + }, + { + "epoch": 0.9636503404348781, + "grad_norm": 2.049753427505493, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7029115557670593, + "num_tokens": 223966158.0, + "step": 8775 + }, + { + "epoch": 0.9637601581374917, + "grad_norm": 2.2044432163238525, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7159008383750916, + "num_tokens": 223988961.0, + "step": 8776 + }, + { + "epoch": 0.9638699758401055, + "grad_norm": 2.356393814086914, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7139480113983154, + "num_tokens": 224011242.0, + "step": 8777 + }, + { + "epoch": 0.9639797935427191, + "grad_norm": 2.398921489715576, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.7052769660949707, + "num_tokens": 224035059.0, + "step": 8778 + }, + { + "epoch": 0.9640896112453328, + "grad_norm": 2.3675711154937744, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7183719873428345, + "num_tokens": 224056356.0, + "step": 8779 + }, + { + "epoch": 0.9641994289479464, + "grad_norm": 1.9411187171936035, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6870124340057373, + "num_tokens": 224088631.0, + "step": 8780 + }, + { + "epoch": 0.9643092466505601, + "grad_norm": 1.8878360986709595, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7115926742553711, + "num_tokens": 224119623.0, + "step": 8781 + }, + { + "epoch": 0.9644190643531737, + "grad_norm": 2.419243097305298, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7059050798416138, + "num_tokens": 224141823.0, + "step": 8782 + }, + { + "epoch": 0.9645288820557874, + "grad_norm": 2.5755937099456787, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7021119594573975, + "num_tokens": 224162772.0, + "step": 8783 + }, + { + "epoch": 0.9646386997584011, + "grad_norm": 2.4341070652008057, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7151646018028259, + "num_tokens": 224184587.0, + "step": 8784 + }, + { + "epoch": 0.9647485174610148, + "grad_norm": 2.238171100616455, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.700937807559967, + "num_tokens": 224209135.0, + "step": 8785 + }, + { + "epoch": 0.9648583351636284, + "grad_norm": 2.1168808937072754, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7023172378540039, + "num_tokens": 224236879.0, + "step": 8786 + }, + { + "epoch": 0.964968152866242, + "grad_norm": 2.3130362033843994, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.704724907875061, + "num_tokens": 224260526.0, + "step": 8787 + }, + { + "epoch": 0.9650779705688557, + "grad_norm": 2.2762773036956787, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7152626514434814, + "num_tokens": 224282440.0, + "step": 8788 + }, + { + "epoch": 0.9651877882714693, + "grad_norm": 2.149376630783081, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7150497436523438, + "num_tokens": 224307149.0, + "step": 8789 + }, + { + "epoch": 0.965297605974083, + "grad_norm": 2.2654426097869873, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7120020389556885, + "num_tokens": 224331098.0, + "step": 8790 + }, + { + "epoch": 0.9654074236766966, + "grad_norm": 2.4110701084136963, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7076810598373413, + "num_tokens": 224352744.0, + "step": 8791 + }, + { + "epoch": 0.9655172413793104, + "grad_norm": 2.5073142051696777, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7486207485198975, + "num_tokens": 224371889.0, + "step": 8792 + }, + { + "epoch": 0.965627059081924, + "grad_norm": 2.3499574661254883, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7086659073829651, + "num_tokens": 224394704.0, + "step": 8793 + }, + { + "epoch": 0.9657368767845377, + "grad_norm": 2.0898845195770264, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7133837938308716, + "num_tokens": 224423579.0, + "step": 8794 + }, + { + "epoch": 0.9658466944871513, + "grad_norm": 2.0651800632476807, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7104299664497375, + "num_tokens": 224451381.0, + "step": 8795 + }, + { + "epoch": 0.965956512189765, + "grad_norm": 2.199711322784424, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.7031186819076538, + "num_tokens": 224478217.0, + "step": 8796 + }, + { + "epoch": 0.9660663298923786, + "grad_norm": 2.3287854194641113, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7335699796676636, + "num_tokens": 224499355.0, + "step": 8797 + }, + { + "epoch": 0.9661761475949923, + "grad_norm": 2.4783730506896973, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.726894736289978, + "num_tokens": 224518322.0, + "step": 8798 + }, + { + "epoch": 0.966285965297606, + "grad_norm": 2.1117467880249023, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.704355001449585, + "num_tokens": 224545518.0, + "step": 8799 + }, + { + "epoch": 0.9663957830002197, + "grad_norm": 2.0937325954437256, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7328989505767822, + "num_tokens": 224569878.0, + "step": 8800 + }, + { + "epoch": 0.9665056007028333, + "grad_norm": 2.4059512615203857, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.706152081489563, + "num_tokens": 224592428.0, + "step": 8801 + }, + { + "epoch": 0.966615418405447, + "grad_norm": 2.1822946071624756, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7110288739204407, + "num_tokens": 224616739.0, + "step": 8802 + }, + { + "epoch": 0.9667252361080606, + "grad_norm": 2.35896897315979, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7249908447265625, + "num_tokens": 224638218.0, + "step": 8803 + }, + { + "epoch": 0.9668350538106742, + "grad_norm": 2.5220532417297363, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7202727198600769, + "num_tokens": 224657457.0, + "step": 8804 + }, + { + "epoch": 0.9669448715132879, + "grad_norm": 2.261714220046997, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7294132709503174, + "num_tokens": 224679350.0, + "step": 8805 + }, + { + "epoch": 0.9670546892159017, + "grad_norm": 2.175816297531128, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7337331771850586, + "num_tokens": 224702437.0, + "step": 8806 + }, + { + "epoch": 0.9671645069185153, + "grad_norm": 2.113255500793457, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7065937519073486, + "num_tokens": 224729376.0, + "step": 8807 + }, + { + "epoch": 0.967274324621129, + "grad_norm": 2.072258234024048, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6888872385025024, + "num_tokens": 224759672.0, + "step": 8808 + }, + { + "epoch": 0.9673841423237426, + "grad_norm": 2.275689125061035, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7120188474655151, + "num_tokens": 224783135.0, + "step": 8809 + }, + { + "epoch": 0.9674939600263562, + "grad_norm": 2.1709299087524414, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7162169814109802, + "num_tokens": 224806584.0, + "step": 8810 + }, + { + "epoch": 0.9676037777289699, + "grad_norm": 2.020390510559082, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7302460670471191, + "num_tokens": 224833862.0, + "step": 8811 + }, + { + "epoch": 0.9677135954315835, + "grad_norm": 2.6229586601257324, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7172834873199463, + "num_tokens": 224853867.0, + "step": 8812 + }, + { + "epoch": 0.9678234131341973, + "grad_norm": 2.3062753677368164, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7155189514160156, + "num_tokens": 224882048.0, + "step": 8813 + }, + { + "epoch": 0.9679332308368109, + "grad_norm": 2.1570239067077637, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7290294766426086, + "num_tokens": 224908574.0, + "step": 8814 + }, + { + "epoch": 0.9680430485394246, + "grad_norm": 2.273789882659912, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7097485065460205, + "num_tokens": 224933601.0, + "step": 8815 + }, + { + "epoch": 0.9681528662420382, + "grad_norm": 2.2326009273529053, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.726150631904602, + "num_tokens": 224956605.0, + "step": 8816 + }, + { + "epoch": 0.9682626839446519, + "grad_norm": 2.298288345336914, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7030508518218994, + "num_tokens": 224979010.0, + "step": 8817 + }, + { + "epoch": 0.9683725016472655, + "grad_norm": 2.1162126064300537, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7227720022201538, + "num_tokens": 225006406.0, + "step": 8818 + }, + { + "epoch": 0.9684823193498792, + "grad_norm": 2.3164687156677246, + "learning_rate": 1e-06, + "loss": 1.0554, + "mean_token_accuracy": 0.6897067427635193, + "num_tokens": 225030987.0, + "step": 8819 + }, + { + "epoch": 0.9685921370524928, + "grad_norm": 1.9969755411148071, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7172549962997437, + "num_tokens": 225058414.0, + "step": 8820 + }, + { + "epoch": 0.9687019547551066, + "grad_norm": 2.2415964603424072, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.6900731325149536, + "num_tokens": 225081653.0, + "step": 8821 + }, + { + "epoch": 0.9688117724577202, + "grad_norm": 2.132535934448242, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7167706489562988, + "num_tokens": 225105546.0, + "step": 8822 + }, + { + "epoch": 0.9689215901603339, + "grad_norm": 2.2343008518218994, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7318208813667297, + "num_tokens": 225128312.0, + "step": 8823 + }, + { + "epoch": 0.9690314078629475, + "grad_norm": 2.4852075576782227, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7200303077697754, + "num_tokens": 225149155.0, + "step": 8824 + }, + { + "epoch": 0.9691412255655611, + "grad_norm": 2.2130753993988037, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.693559467792511, + "num_tokens": 225174517.0, + "step": 8825 + }, + { + "epoch": 0.9692510432681748, + "grad_norm": 2.1034696102142334, + "learning_rate": 1e-06, + "loss": 1.0296, + "mean_token_accuracy": 0.6837944984436035, + "num_tokens": 225202196.0, + "step": 8826 + }, + { + "epoch": 0.9693608609707884, + "grad_norm": 2.213386058807373, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7146249413490295, + "num_tokens": 225226650.0, + "step": 8827 + }, + { + "epoch": 0.9694706786734022, + "grad_norm": 2.111008882522583, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7427529096603394, + "num_tokens": 225251345.0, + "step": 8828 + }, + { + "epoch": 0.9695804963760158, + "grad_norm": 2.015850782394409, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7254289984703064, + "num_tokens": 225279955.0, + "step": 8829 + }, + { + "epoch": 0.9696903140786295, + "grad_norm": 2.1009364128112793, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7110057473182678, + "num_tokens": 225306737.0, + "step": 8830 + }, + { + "epoch": 0.9698001317812431, + "grad_norm": 2.2789885997772217, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6946383714675903, + "num_tokens": 225332748.0, + "step": 8831 + }, + { + "epoch": 0.9699099494838568, + "grad_norm": 2.0600883960723877, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.6945744752883911, + "num_tokens": 225364024.0, + "step": 8832 + }, + { + "epoch": 0.9700197671864704, + "grad_norm": 2.5112109184265137, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7203537225723267, + "num_tokens": 225384563.0, + "step": 8833 + }, + { + "epoch": 0.9701295848890841, + "grad_norm": 2.1420884132385254, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7172427177429199, + "num_tokens": 225409104.0, + "step": 8834 + }, + { + "epoch": 0.9702394025916978, + "grad_norm": 2.3134024143218994, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.733547568321228, + "num_tokens": 225431482.0, + "step": 8835 + }, + { + "epoch": 0.9703492202943115, + "grad_norm": 2.0716331005096436, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.7037073969841003, + "num_tokens": 225460211.0, + "step": 8836 + }, + { + "epoch": 0.9704590379969251, + "grad_norm": 1.916390299797058, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.7052648663520813, + "num_tokens": 225491282.0, + "step": 8837 + }, + { + "epoch": 0.9705688556995388, + "grad_norm": 2.068817138671875, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7264791131019592, + "num_tokens": 225517487.0, + "step": 8838 + }, + { + "epoch": 0.9706786734021524, + "grad_norm": 2.3034212589263916, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6913597583770752, + "num_tokens": 225541350.0, + "step": 8839 + }, + { + "epoch": 0.9707884911047661, + "grad_norm": 2.064664840698242, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.708617627620697, + "num_tokens": 225568658.0, + "step": 8840 + }, + { + "epoch": 0.9708983088073797, + "grad_norm": 2.1988158226013184, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7112350463867188, + "num_tokens": 225594242.0, + "step": 8841 + }, + { + "epoch": 0.9710081265099935, + "grad_norm": 2.1551783084869385, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7023996114730835, + "num_tokens": 225619215.0, + "step": 8842 + }, + { + "epoch": 0.9711179442126071, + "grad_norm": 1.8845882415771484, + "learning_rate": 1e-06, + "loss": 1.0821, + "mean_token_accuracy": 0.6879448890686035, + "num_tokens": 225653214.0, + "step": 8843 + }, + { + "epoch": 0.9712277619152208, + "grad_norm": 1.8923667669296265, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.6930217742919922, + "num_tokens": 225686607.0, + "step": 8844 + }, + { + "epoch": 0.9713375796178344, + "grad_norm": 2.262484312057495, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7141721248626709, + "num_tokens": 225709046.0, + "step": 8845 + }, + { + "epoch": 0.971447397320448, + "grad_norm": 2.2736918926239014, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7024065256118774, + "num_tokens": 225733260.0, + "step": 8846 + }, + { + "epoch": 0.9715572150230617, + "grad_norm": 2.0565874576568604, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6952545642852783, + "num_tokens": 225760431.0, + "step": 8847 + }, + { + "epoch": 0.9716670327256753, + "grad_norm": 2.4319353103637695, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7148758172988892, + "num_tokens": 225781852.0, + "step": 8848 + }, + { + "epoch": 0.971776850428289, + "grad_norm": 1.9107627868652344, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7219078540802002, + "num_tokens": 225811047.0, + "step": 8849 + }, + { + "epoch": 0.9718866681309027, + "grad_norm": 2.4704511165618896, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7264654636383057, + "num_tokens": 225830037.0, + "step": 8850 + }, + { + "epoch": 0.9719964858335164, + "grad_norm": 1.9763922691345215, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7053936719894409, + "num_tokens": 225862377.0, + "step": 8851 + }, + { + "epoch": 0.97210630353613, + "grad_norm": 2.033024549484253, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.6949479579925537, + "num_tokens": 225887933.0, + "step": 8852 + }, + { + "epoch": 0.9722161212387437, + "grad_norm": 1.76400887966156, + "learning_rate": 1e-06, + "loss": 1.0853, + "mean_token_accuracy": 0.6796644926071167, + "num_tokens": 225924209.0, + "step": 8853 + }, + { + "epoch": 0.9723259389413573, + "grad_norm": 2.389103651046753, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7168325185775757, + "num_tokens": 225945056.0, + "step": 8854 + }, + { + "epoch": 0.972435756643971, + "grad_norm": 2.175703763961792, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7424330115318298, + "num_tokens": 225969992.0, + "step": 8855 + }, + { + "epoch": 0.9725455743465846, + "grad_norm": 2.1564037799835205, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7360090613365173, + "num_tokens": 225993931.0, + "step": 8856 + }, + { + "epoch": 0.9726553920491984, + "grad_norm": 1.8177310228347778, + "learning_rate": 1e-06, + "loss": 1.0542, + "mean_token_accuracy": 0.6895084977149963, + "num_tokens": 226028426.0, + "step": 8857 + }, + { + "epoch": 0.972765209751812, + "grad_norm": 2.1910595893859863, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7060559988021851, + "num_tokens": 226054594.0, + "step": 8858 + }, + { + "epoch": 0.9728750274544257, + "grad_norm": 2.2699739933013916, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.7008866667747498, + "num_tokens": 226078842.0, + "step": 8859 + }, + { + "epoch": 0.9729848451570393, + "grad_norm": 2.1238479614257812, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7074357271194458, + "num_tokens": 226105367.0, + "step": 8860 + }, + { + "epoch": 0.973094662859653, + "grad_norm": 2.208268642425537, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7203601598739624, + "num_tokens": 226128661.0, + "step": 8861 + }, + { + "epoch": 0.9732044805622666, + "grad_norm": 2.1111278533935547, + "learning_rate": 1e-06, + "loss": 1.0587, + "mean_token_accuracy": 0.6799628734588623, + "num_tokens": 226156862.0, + "step": 8862 + }, + { + "epoch": 0.9733142982648803, + "grad_norm": 2.449352741241455, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7296561002731323, + "num_tokens": 226176863.0, + "step": 8863 + }, + { + "epoch": 0.973424115967494, + "grad_norm": 2.4120638370513916, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7020860910415649, + "num_tokens": 226199960.0, + "step": 8864 + }, + { + "epoch": 0.9735339336701077, + "grad_norm": 2.1652133464813232, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.6991934776306152, + "num_tokens": 226225778.0, + "step": 8865 + }, + { + "epoch": 0.9736437513727213, + "grad_norm": 2.090898036956787, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7042055130004883, + "num_tokens": 226251593.0, + "step": 8866 + }, + { + "epoch": 0.973753569075335, + "grad_norm": 2.1281776428222656, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6956311464309692, + "num_tokens": 226279536.0, + "step": 8867 + }, + { + "epoch": 0.9738633867779486, + "grad_norm": 2.1799960136413574, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6841012835502625, + "num_tokens": 226305675.0, + "step": 8868 + }, + { + "epoch": 0.9739732044805622, + "grad_norm": 2.2628047466278076, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7122443914413452, + "num_tokens": 226330014.0, + "step": 8869 + }, + { + "epoch": 0.9740830221831759, + "grad_norm": 2.3543763160705566, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7051588892936707, + "num_tokens": 226350606.0, + "step": 8870 + }, + { + "epoch": 0.9741928398857896, + "grad_norm": 2.3125014305114746, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7068805694580078, + "num_tokens": 226374467.0, + "step": 8871 + }, + { + "epoch": 0.9743026575884033, + "grad_norm": 2.2984063625335693, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7406762838363647, + "num_tokens": 226398441.0, + "step": 8872 + }, + { + "epoch": 0.9744124752910169, + "grad_norm": 2.116070032119751, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.692264199256897, + "num_tokens": 226425272.0, + "step": 8873 + }, + { + "epoch": 0.9745222929936306, + "grad_norm": 2.233381509780884, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7194314002990723, + "num_tokens": 226448288.0, + "step": 8874 + }, + { + "epoch": 0.9746321106962442, + "grad_norm": 2.368293285369873, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7331780195236206, + "num_tokens": 226469509.0, + "step": 8875 + }, + { + "epoch": 0.9747419283988579, + "grad_norm": 2.201807737350464, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6972229480743408, + "num_tokens": 226495143.0, + "step": 8876 + }, + { + "epoch": 0.9748517461014715, + "grad_norm": 2.046334981918335, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6895974278450012, + "num_tokens": 226523091.0, + "step": 8877 + }, + { + "epoch": 0.9749615638040853, + "grad_norm": 2.401599645614624, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6976445913314819, + "num_tokens": 226545057.0, + "step": 8878 + }, + { + "epoch": 0.9750713815066989, + "grad_norm": 2.474745035171509, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7128620147705078, + "num_tokens": 226565248.0, + "step": 8879 + }, + { + "epoch": 0.9751811992093126, + "grad_norm": 2.3426625728607178, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.715309202671051, + "num_tokens": 226585917.0, + "step": 8880 + }, + { + "epoch": 0.9752910169119262, + "grad_norm": 2.0557003021240234, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7152202129364014, + "num_tokens": 226613834.0, + "step": 8881 + }, + { + "epoch": 0.9754008346145399, + "grad_norm": 2.3620643615722656, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7025323510169983, + "num_tokens": 226637597.0, + "step": 8882 + }, + { + "epoch": 0.9755106523171535, + "grad_norm": 2.1021950244903564, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7108062505722046, + "num_tokens": 226664583.0, + "step": 8883 + }, + { + "epoch": 0.9756204700197671, + "grad_norm": 2.4713311195373535, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7086178064346313, + "num_tokens": 226685110.0, + "step": 8884 + }, + { + "epoch": 0.9757302877223808, + "grad_norm": 2.325251579284668, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7166780233383179, + "num_tokens": 226709289.0, + "step": 8885 + }, + { + "epoch": 0.9758401054249946, + "grad_norm": 2.2436163425445557, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6922391057014465, + "num_tokens": 226732884.0, + "step": 8886 + }, + { + "epoch": 0.9759499231276082, + "grad_norm": 1.9248137474060059, + "learning_rate": 1e-06, + "loss": 1.0325, + "mean_token_accuracy": 0.6853342056274414, + "num_tokens": 226766811.0, + "step": 8887 + }, + { + "epoch": 0.9760597408302218, + "grad_norm": 1.989101529121399, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.7159591913223267, + "num_tokens": 226797633.0, + "step": 8888 + }, + { + "epoch": 0.9761695585328355, + "grad_norm": 1.9585680961608887, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.6977243423461914, + "num_tokens": 226828502.0, + "step": 8889 + }, + { + "epoch": 0.9762793762354491, + "grad_norm": 2.0614590644836426, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.6965252161026001, + "num_tokens": 226858738.0, + "step": 8890 + }, + { + "epoch": 0.9763891939380628, + "grad_norm": 2.2736032009124756, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7286379337310791, + "num_tokens": 226882099.0, + "step": 8891 + }, + { + "epoch": 0.9764990116406764, + "grad_norm": 2.0299408435821533, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7161410450935364, + "num_tokens": 226909921.0, + "step": 8892 + }, + { + "epoch": 0.9766088293432902, + "grad_norm": 2.4083187580108643, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7272051572799683, + "num_tokens": 226932533.0, + "step": 8893 + }, + { + "epoch": 0.9767186470459038, + "grad_norm": 2.1449506282806396, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7149868011474609, + "num_tokens": 226958878.0, + "step": 8894 + }, + { + "epoch": 0.9768284647485175, + "grad_norm": 2.3052003383636475, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7303911447525024, + "num_tokens": 226981241.0, + "step": 8895 + }, + { + "epoch": 0.9769382824511311, + "grad_norm": 2.1630544662475586, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7162383198738098, + "num_tokens": 227007345.0, + "step": 8896 + }, + { + "epoch": 0.9770481001537448, + "grad_norm": 2.0409164428710938, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7248920798301697, + "num_tokens": 227035779.0, + "step": 8897 + }, + { + "epoch": 0.9771579178563584, + "grad_norm": 1.9949591159820557, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7004367113113403, + "num_tokens": 227063864.0, + "step": 8898 + }, + { + "epoch": 0.9772677355589721, + "grad_norm": 2.130450487136841, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7002661228179932, + "num_tokens": 227090469.0, + "step": 8899 + }, + { + "epoch": 0.9773775532615858, + "grad_norm": 2.2477076053619385, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7119339108467102, + "num_tokens": 227114482.0, + "step": 8900 + }, + { + "epoch": 0.9774873709641995, + "grad_norm": 2.2152416706085205, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7164815664291382, + "num_tokens": 227138266.0, + "step": 8901 + }, + { + "epoch": 0.9775971886668131, + "grad_norm": 2.1997125148773193, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7137807607650757, + "num_tokens": 227164537.0, + "step": 8902 + }, + { + "epoch": 0.9777070063694268, + "grad_norm": 2.248244047164917, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.7032163739204407, + "num_tokens": 227188826.0, + "step": 8903 + }, + { + "epoch": 0.9778168240720404, + "grad_norm": 1.9223638772964478, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7183570861816406, + "num_tokens": 227219227.0, + "step": 8904 + }, + { + "epoch": 0.977926641774654, + "grad_norm": 2.37060546875, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7213539481163025, + "num_tokens": 227242070.0, + "step": 8905 + }, + { + "epoch": 0.9780364594772677, + "grad_norm": 2.2987112998962402, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7087430953979492, + "num_tokens": 227265333.0, + "step": 8906 + }, + { + "epoch": 0.9781462771798815, + "grad_norm": 1.9327720403671265, + "learning_rate": 1e-06, + "loss": 1.0515, + "mean_token_accuracy": 0.6814310550689697, + "num_tokens": 227297761.0, + "step": 8907 + }, + { + "epoch": 0.9782560948824951, + "grad_norm": 2.830815553665161, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7156025767326355, + "num_tokens": 227313567.0, + "step": 8908 + }, + { + "epoch": 0.9783659125851087, + "grad_norm": 2.0056710243225098, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7235168218612671, + "num_tokens": 227342557.0, + "step": 8909 + }, + { + "epoch": 0.9784757302877224, + "grad_norm": 2.2174158096313477, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7228012084960938, + "num_tokens": 227366162.0, + "step": 8910 + }, + { + "epoch": 0.978585547990336, + "grad_norm": 1.9315205812454224, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6946706175804138, + "num_tokens": 227400064.0, + "step": 8911 + }, + { + "epoch": 0.9786953656929497, + "grad_norm": 2.122091293334961, + "learning_rate": 1e-06, + "loss": 1.0209, + "mean_token_accuracy": 0.6844873428344727, + "num_tokens": 227428000.0, + "step": 8912 + }, + { + "epoch": 0.9788051833955633, + "grad_norm": 2.150306463241577, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7144132852554321, + "num_tokens": 227453540.0, + "step": 8913 + }, + { + "epoch": 0.978915001098177, + "grad_norm": 2.186426877975464, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7349134683609009, + "num_tokens": 227477064.0, + "step": 8914 + }, + { + "epoch": 0.9790248188007907, + "grad_norm": 2.0081675052642822, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7044274806976318, + "num_tokens": 227506170.0, + "step": 8915 + }, + { + "epoch": 0.9791346365034044, + "grad_norm": 2.232234477996826, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.689341127872467, + "num_tokens": 227530729.0, + "step": 8916 + }, + { + "epoch": 0.979244454206018, + "grad_norm": 2.224911689758301, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7007961273193359, + "num_tokens": 227556302.0, + "step": 8917 + }, + { + "epoch": 0.9793542719086317, + "grad_norm": 1.8601350784301758, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7330191135406494, + "num_tokens": 227589071.0, + "step": 8918 + }, + { + "epoch": 0.9794640896112453, + "grad_norm": 1.9328477382659912, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7019473910331726, + "num_tokens": 227621493.0, + "step": 8919 + }, + { + "epoch": 0.979573907313859, + "grad_norm": 2.0596113204956055, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7193470597267151, + "num_tokens": 227647706.0, + "step": 8920 + }, + { + "epoch": 0.9796837250164726, + "grad_norm": 2.2135043144226074, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7025060057640076, + "num_tokens": 227673551.0, + "step": 8921 + }, + { + "epoch": 0.9797935427190864, + "grad_norm": 2.142666816711426, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.7021343111991882, + "num_tokens": 227699385.0, + "step": 8922 + }, + { + "epoch": 0.9799033604217, + "grad_norm": 2.0820581912994385, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7000857591629028, + "num_tokens": 227728183.0, + "step": 8923 + }, + { + "epoch": 0.9800131781243137, + "grad_norm": 2.021592140197754, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.722038745880127, + "num_tokens": 227754893.0, + "step": 8924 + }, + { + "epoch": 0.9801229958269273, + "grad_norm": 2.2930257320404053, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7259994149208069, + "num_tokens": 227777024.0, + "step": 8925 + }, + { + "epoch": 0.980232813529541, + "grad_norm": 1.9595366716384888, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7118586897850037, + "num_tokens": 227803476.0, + "step": 8926 + }, + { + "epoch": 0.9803426312321546, + "grad_norm": 2.1075453758239746, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7121137380599976, + "num_tokens": 227827950.0, + "step": 8927 + }, + { + "epoch": 0.9804524489347682, + "grad_norm": 1.9689359664916992, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7208884954452515, + "num_tokens": 227857368.0, + "step": 8928 + }, + { + "epoch": 0.980562266637382, + "grad_norm": 2.0882062911987305, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6912474632263184, + "num_tokens": 227884513.0, + "step": 8929 + }, + { + "epoch": 0.9806720843399956, + "grad_norm": 2.320307970046997, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.6992889642715454, + "num_tokens": 227906771.0, + "step": 8930 + }, + { + "epoch": 0.9807819020426093, + "grad_norm": 2.3421435356140137, + "learning_rate": 1e-06, + "loss": 1.0326, + "mean_token_accuracy": 0.6848995685577393, + "num_tokens": 227930194.0, + "step": 8931 + }, + { + "epoch": 0.9808917197452229, + "grad_norm": 2.199308156967163, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7051209807395935, + "num_tokens": 227953033.0, + "step": 8932 + }, + { + "epoch": 0.9810015374478366, + "grad_norm": 2.2603111267089844, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.7008864879608154, + "num_tokens": 227977860.0, + "step": 8933 + }, + { + "epoch": 0.9811113551504502, + "grad_norm": 2.217986822128296, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.695146918296814, + "num_tokens": 228004211.0, + "step": 8934 + }, + { + "epoch": 0.9812211728530639, + "grad_norm": 2.2106542587280273, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7090719938278198, + "num_tokens": 228029716.0, + "step": 8935 + }, + { + "epoch": 0.9813309905556776, + "grad_norm": 2.2669947147369385, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6952993273735046, + "num_tokens": 228054367.0, + "step": 8936 + }, + { + "epoch": 0.9814408082582913, + "grad_norm": 2.483103036880493, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7160153388977051, + "num_tokens": 228074713.0, + "step": 8937 + }, + { + "epoch": 0.9815506259609049, + "grad_norm": 2.505002737045288, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7298746109008789, + "num_tokens": 228094799.0, + "step": 8938 + }, + { + "epoch": 0.9816604436635186, + "grad_norm": 2.2368879318237305, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.694405198097229, + "num_tokens": 228119686.0, + "step": 8939 + }, + { + "epoch": 0.9817702613661322, + "grad_norm": 2.5257351398468018, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7281385660171509, + "num_tokens": 228139109.0, + "step": 8940 + }, + { + "epoch": 0.9818800790687459, + "grad_norm": 2.1390469074249268, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7278664112091064, + "num_tokens": 228163398.0, + "step": 8941 + }, + { + "epoch": 0.9819898967713595, + "grad_norm": 2.2715275287628174, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7320933938026428, + "num_tokens": 228186437.0, + "step": 8942 + }, + { + "epoch": 0.9820997144739732, + "grad_norm": 2.0384981632232666, + "learning_rate": 1e-06, + "loss": 1.0424, + "mean_token_accuracy": 0.688919186592102, + "num_tokens": 228215147.0, + "step": 8943 + }, + { + "epoch": 0.9822095321765869, + "grad_norm": 2.620326280593872, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.709353506565094, + "num_tokens": 228235122.0, + "step": 8944 + }, + { + "epoch": 0.9823193498792006, + "grad_norm": 2.058013916015625, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7003467082977295, + "num_tokens": 228264748.0, + "step": 8945 + }, + { + "epoch": 0.9824291675818142, + "grad_norm": 2.3914899826049805, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6943253874778748, + "num_tokens": 228288535.0, + "step": 8946 + }, + { + "epoch": 0.9825389852844278, + "grad_norm": 2.453059196472168, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7408844232559204, + "num_tokens": 228307127.0, + "step": 8947 + }, + { + "epoch": 0.9826488029870415, + "grad_norm": 2.0731472969055176, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7249078750610352, + "num_tokens": 228333776.0, + "step": 8948 + }, + { + "epoch": 0.9827586206896551, + "grad_norm": 2.160529136657715, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.6870632767677307, + "num_tokens": 228360393.0, + "step": 8949 + }, + { + "epoch": 0.9828684383922688, + "grad_norm": 2.3258023262023926, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7075905799865723, + "num_tokens": 228382681.0, + "step": 8950 + }, + { + "epoch": 0.9829782560948825, + "grad_norm": 1.9143316745758057, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.69515061378479, + "num_tokens": 228413996.0, + "step": 8951 + }, + { + "epoch": 0.9830880737974962, + "grad_norm": 2.2160372734069824, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7271194458007812, + "num_tokens": 228437682.0, + "step": 8952 + }, + { + "epoch": 0.9831978915001098, + "grad_norm": 2.396592855453491, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7112501859664917, + "num_tokens": 228456903.0, + "step": 8953 + }, + { + "epoch": 0.9833077092027235, + "grad_norm": 2.0325798988342285, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.716128408908844, + "num_tokens": 228484181.0, + "step": 8954 + }, + { + "epoch": 0.9834175269053371, + "grad_norm": 2.502423048019409, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7112035751342773, + "num_tokens": 228505244.0, + "step": 8955 + }, + { + "epoch": 0.9835273446079508, + "grad_norm": 2.1418139934539795, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7159019708633423, + "num_tokens": 228529618.0, + "step": 8956 + }, + { + "epoch": 0.9836371623105644, + "grad_norm": 2.1672706604003906, + "learning_rate": 1e-06, + "loss": 1.0228, + "mean_token_accuracy": 0.6937698721885681, + "num_tokens": 228555549.0, + "step": 8957 + }, + { + "epoch": 0.9837469800131782, + "grad_norm": 2.1971802711486816, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7162757515907288, + "num_tokens": 228581229.0, + "step": 8958 + }, + { + "epoch": 0.9838567977157918, + "grad_norm": 2.3670291900634766, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6982265710830688, + "num_tokens": 228605529.0, + "step": 8959 + }, + { + "epoch": 0.9839666154184055, + "grad_norm": 2.1184873580932617, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7345726490020752, + "num_tokens": 228630005.0, + "step": 8960 + }, + { + "epoch": 0.9840764331210191, + "grad_norm": 2.14289927482605, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7128660082817078, + "num_tokens": 228656218.0, + "step": 8961 + }, + { + "epoch": 0.9841862508236328, + "grad_norm": 2.151754379272461, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7344770431518555, + "num_tokens": 228681676.0, + "step": 8962 + }, + { + "epoch": 0.9842960685262464, + "grad_norm": 2.6487843990325928, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7150688171386719, + "num_tokens": 228702426.0, + "step": 8963 + }, + { + "epoch": 0.98440588622886, + "grad_norm": 1.8810462951660156, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.707008421421051, + "num_tokens": 228735219.0, + "step": 8964 + }, + { + "epoch": 0.9845157039314738, + "grad_norm": 2.4079995155334473, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7076388597488403, + "num_tokens": 228756987.0, + "step": 8965 + }, + { + "epoch": 0.9846255216340875, + "grad_norm": 2.233496904373169, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6959266662597656, + "num_tokens": 228781421.0, + "step": 8966 + }, + { + "epoch": 0.9847353393367011, + "grad_norm": 1.8681987524032593, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7084305882453918, + "num_tokens": 228813212.0, + "step": 8967 + }, + { + "epoch": 0.9848451570393147, + "grad_norm": 2.0003299713134766, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7010972499847412, + "num_tokens": 228842953.0, + "step": 8968 + }, + { + "epoch": 0.9849549747419284, + "grad_norm": 1.9650136232376099, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7101770043373108, + "num_tokens": 228875263.0, + "step": 8969 + }, + { + "epoch": 0.985064792444542, + "grad_norm": 2.292985677719116, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7133383750915527, + "num_tokens": 228899035.0, + "step": 8970 + }, + { + "epoch": 0.9851746101471557, + "grad_norm": 2.1194968223571777, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7287260293960571, + "num_tokens": 228926180.0, + "step": 8971 + }, + { + "epoch": 0.9852844278497693, + "grad_norm": 2.310598134994507, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.69426429271698, + "num_tokens": 228949547.0, + "step": 8972 + }, + { + "epoch": 0.9853942455523831, + "grad_norm": 2.0111286640167236, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7211191654205322, + "num_tokens": 228979496.0, + "step": 8973 + }, + { + "epoch": 0.9855040632549967, + "grad_norm": 2.0459628105163574, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.6842037439346313, + "num_tokens": 229008823.0, + "step": 8974 + }, + { + "epoch": 0.9856138809576104, + "grad_norm": 2.1690471172332764, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.6960580945014954, + "num_tokens": 229035898.0, + "step": 8975 + }, + { + "epoch": 0.985723698660224, + "grad_norm": 2.2321343421936035, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7308743596076965, + "num_tokens": 229059826.0, + "step": 8976 + }, + { + "epoch": 0.9858335163628377, + "grad_norm": 2.4620540142059326, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7161874771118164, + "num_tokens": 229079449.0, + "step": 8977 + }, + { + "epoch": 0.9859433340654513, + "grad_norm": 2.5107266902923584, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.6963952779769897, + "num_tokens": 229099542.0, + "step": 8978 + }, + { + "epoch": 0.986053151768065, + "grad_norm": 1.9744398593902588, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6961352825164795, + "num_tokens": 229131484.0, + "step": 8979 + }, + { + "epoch": 0.9861629694706787, + "grad_norm": 2.6229100227355957, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7155576944351196, + "num_tokens": 229149872.0, + "step": 8980 + }, + { + "epoch": 0.9862727871732924, + "grad_norm": 2.274827003479004, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.719829797744751, + "num_tokens": 229171832.0, + "step": 8981 + }, + { + "epoch": 0.986382604875906, + "grad_norm": 2.5567073822021484, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7159166932106018, + "num_tokens": 229191702.0, + "step": 8982 + }, + { + "epoch": 0.9864924225785197, + "grad_norm": 1.9606101512908936, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7227202653884888, + "num_tokens": 229221985.0, + "step": 8983 + }, + { + "epoch": 0.9866022402811333, + "grad_norm": 2.197861909866333, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7080661058425903, + "num_tokens": 229246833.0, + "step": 8984 + }, + { + "epoch": 0.986712057983747, + "grad_norm": 2.4006893634796143, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7150378823280334, + "num_tokens": 229269094.0, + "step": 8985 + }, + { + "epoch": 0.9868218756863606, + "grad_norm": 2.3757898807525635, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7082922458648682, + "num_tokens": 229291601.0, + "step": 8986 + }, + { + "epoch": 0.9869316933889744, + "grad_norm": 2.4092471599578857, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7160629034042358, + "num_tokens": 229314020.0, + "step": 8987 + }, + { + "epoch": 0.987041511091588, + "grad_norm": 2.23018479347229, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7264991998672485, + "num_tokens": 229338968.0, + "step": 8988 + }, + { + "epoch": 0.9871513287942016, + "grad_norm": 2.405690908432007, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6859328150749207, + "num_tokens": 229362532.0, + "step": 8989 + }, + { + "epoch": 0.9872611464968153, + "grad_norm": 2.7582101821899414, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7153849005699158, + "num_tokens": 229380989.0, + "step": 8990 + }, + { + "epoch": 0.9873709641994289, + "grad_norm": 2.610927104949951, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7235242128372192, + "num_tokens": 229400390.0, + "step": 8991 + }, + { + "epoch": 0.9874807819020426, + "grad_norm": 2.3231968879699707, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7302348613739014, + "num_tokens": 229421947.0, + "step": 8992 + }, + { + "epoch": 0.9875905996046562, + "grad_norm": 2.1136772632598877, + "learning_rate": 1e-06, + "loss": 1.0527, + "mean_token_accuracy": 0.6818814277648926, + "num_tokens": 229450483.0, + "step": 8993 + }, + { + "epoch": 0.98770041730727, + "grad_norm": 1.9455913305282593, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7189052104949951, + "num_tokens": 229479990.0, + "step": 8994 + }, + { + "epoch": 0.9878102350098836, + "grad_norm": 2.4067137241363525, + "learning_rate": 1e-06, + "loss": 0.7823, + "mean_token_accuracy": 0.7481530904769897, + "num_tokens": 229499843.0, + "step": 8995 + }, + { + "epoch": 0.9879200527124973, + "grad_norm": 2.253624439239502, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7103362083435059, + "num_tokens": 229525085.0, + "step": 8996 + }, + { + "epoch": 0.9880298704151109, + "grad_norm": 2.224207878112793, + "learning_rate": 1e-06, + "loss": 1.0638, + "mean_token_accuracy": 0.6809442043304443, + "num_tokens": 229553623.0, + "step": 8997 + }, + { + "epoch": 0.9881396881177246, + "grad_norm": 2.160311222076416, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7060909271240234, + "num_tokens": 229578910.0, + "step": 8998 + }, + { + "epoch": 0.9882495058203382, + "grad_norm": 2.5825700759887695, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7454731464385986, + "num_tokens": 229599158.0, + "step": 8999 + }, + { + "epoch": 0.9883593235229519, + "grad_norm": 2.169053792953491, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7023898959159851, + "num_tokens": 229626175.0, + "step": 9000 + }, + { + "epoch": 0.9884691412255655, + "grad_norm": 2.078007221221924, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7125456929206848, + "num_tokens": 229652112.0, + "step": 9001 + }, + { + "epoch": 0.9885789589281793, + "grad_norm": 2.2916316986083984, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.7029247283935547, + "num_tokens": 229676146.0, + "step": 9002 + }, + { + "epoch": 0.9886887766307929, + "grad_norm": 1.9663130044937134, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7029253840446472, + "num_tokens": 229707088.0, + "step": 9003 + }, + { + "epoch": 0.9887985943334066, + "grad_norm": 2.2820489406585693, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7350619435310364, + "num_tokens": 229728803.0, + "step": 9004 + }, + { + "epoch": 0.9889084120360202, + "grad_norm": 2.0029494762420654, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7384546995162964, + "num_tokens": 229756670.0, + "step": 9005 + }, + { + "epoch": 0.9890182297386338, + "grad_norm": 2.342650890350342, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.6953492760658264, + "num_tokens": 229777774.0, + "step": 9006 + }, + { + "epoch": 0.9891280474412475, + "grad_norm": 2.3772778511047363, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7003183364868164, + "num_tokens": 229801549.0, + "step": 9007 + }, + { + "epoch": 0.9892378651438611, + "grad_norm": 2.4147732257843018, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6935085654258728, + "num_tokens": 229825832.0, + "step": 9008 + }, + { + "epoch": 0.9893476828464749, + "grad_norm": 2.1589415073394775, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6988502144813538, + "num_tokens": 229851097.0, + "step": 9009 + }, + { + "epoch": 0.9894575005490885, + "grad_norm": 2.568427324295044, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7324987649917603, + "num_tokens": 229869149.0, + "step": 9010 + }, + { + "epoch": 0.9895673182517022, + "grad_norm": 2.1740362644195557, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.7013223171234131, + "num_tokens": 229895510.0, + "step": 9011 + }, + { + "epoch": 0.9896771359543158, + "grad_norm": 2.3084559440612793, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7143803238868713, + "num_tokens": 229920358.0, + "step": 9012 + }, + { + "epoch": 0.9897869536569295, + "grad_norm": 2.166376829147339, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7238001823425293, + "num_tokens": 229946884.0, + "step": 9013 + }, + { + "epoch": 0.9898967713595431, + "grad_norm": 2.2115848064422607, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7048414945602417, + "num_tokens": 229970309.0, + "step": 9014 + }, + { + "epoch": 0.9900065890621568, + "grad_norm": 2.1472508907318115, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7402839660644531, + "num_tokens": 229995139.0, + "step": 9015 + }, + { + "epoch": 0.9901164067647705, + "grad_norm": 2.254307746887207, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.704818069934845, + "num_tokens": 230019455.0, + "step": 9016 + }, + { + "epoch": 0.9902262244673842, + "grad_norm": 1.8997151851654053, + "learning_rate": 1e-06, + "loss": 1.0777, + "mean_token_accuracy": 0.6728211641311646, + "num_tokens": 230053682.0, + "step": 9017 + }, + { + "epoch": 0.9903360421699978, + "grad_norm": 2.0889456272125244, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7235890626907349, + "num_tokens": 230081086.0, + "step": 9018 + }, + { + "epoch": 0.9904458598726115, + "grad_norm": 2.540684223175049, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.714414119720459, + "num_tokens": 230100030.0, + "step": 9019 + }, + { + "epoch": 0.9905556775752251, + "grad_norm": 2.1956300735473633, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.6822916865348816, + "num_tokens": 230127122.0, + "step": 9020 + }, + { + "epoch": 0.9906654952778388, + "grad_norm": 2.230231761932373, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7127801775932312, + "num_tokens": 230151993.0, + "step": 9021 + }, + { + "epoch": 0.9907753129804524, + "grad_norm": 2.171372652053833, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7238396406173706, + "num_tokens": 230177721.0, + "step": 9022 + }, + { + "epoch": 0.9908851306830662, + "grad_norm": 2.060192823410034, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7179470658302307, + "num_tokens": 230204124.0, + "step": 9023 + }, + { + "epoch": 0.9909949483856798, + "grad_norm": 2.3379266262054443, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7120170593261719, + "num_tokens": 230227435.0, + "step": 9024 + }, + { + "epoch": 0.9911047660882935, + "grad_norm": 1.9555755853652954, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7068639993667603, + "num_tokens": 230256899.0, + "step": 9025 + }, + { + "epoch": 0.9912145837909071, + "grad_norm": 2.3768813610076904, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7117971777915955, + "num_tokens": 230278175.0, + "step": 9026 + }, + { + "epoch": 0.9913244014935207, + "grad_norm": 2.5936429500579834, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7200149893760681, + "num_tokens": 230296715.0, + "step": 9027 + }, + { + "epoch": 0.9914342191961344, + "grad_norm": 2.156446933746338, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7115129232406616, + "num_tokens": 230323676.0, + "step": 9028 + }, + { + "epoch": 0.991544036898748, + "grad_norm": 2.0491440296173096, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7067105770111084, + "num_tokens": 230351940.0, + "step": 9029 + }, + { + "epoch": 0.9916538546013618, + "grad_norm": 2.00150990486145, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6923128366470337, + "num_tokens": 230383649.0, + "step": 9030 + }, + { + "epoch": 0.9917636723039754, + "grad_norm": 2.569521903991699, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7205662727355957, + "num_tokens": 230403469.0, + "step": 9031 + }, + { + "epoch": 0.9918734900065891, + "grad_norm": 2.2664873600006104, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7321751117706299, + "num_tokens": 230424938.0, + "step": 9032 + }, + { + "epoch": 0.9919833077092027, + "grad_norm": 2.0201642513275146, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7344546318054199, + "num_tokens": 230450123.0, + "step": 9033 + }, + { + "epoch": 0.9920931254118164, + "grad_norm": 2.0992159843444824, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.729994535446167, + "num_tokens": 230479400.0, + "step": 9034 + }, + { + "epoch": 0.99220294311443, + "grad_norm": 2.2482454776763916, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7132129669189453, + "num_tokens": 230503448.0, + "step": 9035 + }, + { + "epoch": 0.9923127608170437, + "grad_norm": 2.3233683109283447, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.7090438604354858, + "num_tokens": 230526075.0, + "step": 9036 + }, + { + "epoch": 0.9924225785196573, + "grad_norm": 2.2840006351470947, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7237269878387451, + "num_tokens": 230549799.0, + "step": 9037 + }, + { + "epoch": 0.9925323962222711, + "grad_norm": 2.3278627395629883, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7263480424880981, + "num_tokens": 230573770.0, + "step": 9038 + }, + { + "epoch": 0.9926422139248847, + "grad_norm": 2.2375521659851074, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7253888249397278, + "num_tokens": 230596717.0, + "step": 9039 + }, + { + "epoch": 0.9927520316274984, + "grad_norm": 2.0419838428497314, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7166341543197632, + "num_tokens": 230624251.0, + "step": 9040 + }, + { + "epoch": 0.992861849330112, + "grad_norm": 2.129497766494751, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.7059764266014099, + "num_tokens": 230651962.0, + "step": 9041 + }, + { + "epoch": 0.9929716670327257, + "grad_norm": 2.334937334060669, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7003774642944336, + "num_tokens": 230674940.0, + "step": 9042 + }, + { + "epoch": 0.9930814847353393, + "grad_norm": 1.79407799243927, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6926069259643555, + "num_tokens": 230711387.0, + "step": 9043 + }, + { + "epoch": 0.993191302437953, + "grad_norm": 2.2735350131988525, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7372246384620667, + "num_tokens": 230734518.0, + "step": 9044 + }, + { + "epoch": 0.9933011201405667, + "grad_norm": 2.0384907722473145, + "learning_rate": 1e-06, + "loss": 1.032, + "mean_token_accuracy": 0.6874039173126221, + "num_tokens": 230762175.0, + "step": 9045 + }, + { + "epoch": 0.9934109378431804, + "grad_norm": 2.014838457107544, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7269626259803772, + "num_tokens": 230788809.0, + "step": 9046 + }, + { + "epoch": 0.993520755545794, + "grad_norm": 2.6155648231506348, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7358351945877075, + "num_tokens": 230808502.0, + "step": 9047 + }, + { + "epoch": 0.9936305732484076, + "grad_norm": 2.2259910106658936, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7135376930236816, + "num_tokens": 230834339.0, + "step": 9048 + }, + { + "epoch": 0.9937403909510213, + "grad_norm": 2.2460684776306152, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7219734787940979, + "num_tokens": 230858764.0, + "step": 9049 + }, + { + "epoch": 0.9938502086536349, + "grad_norm": 2.0466182231903076, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7299010753631592, + "num_tokens": 230887614.0, + "step": 9050 + }, + { + "epoch": 0.9939600263562486, + "grad_norm": 2.127686023712158, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7046970129013062, + "num_tokens": 230912012.0, + "step": 9051 + }, + { + "epoch": 0.9940698440588623, + "grad_norm": 2.264044761657715, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7273868322372437, + "num_tokens": 230935853.0, + "step": 9052 + }, + { + "epoch": 0.994179661761476, + "grad_norm": 2.1804590225219727, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.705646812915802, + "num_tokens": 230964652.0, + "step": 9053 + }, + { + "epoch": 0.9942894794640896, + "grad_norm": 2.316345691680908, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.7050151824951172, + "num_tokens": 230987101.0, + "step": 9054 + }, + { + "epoch": 0.9943992971667033, + "grad_norm": 2.3394763469696045, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7133337259292603, + "num_tokens": 231009560.0, + "step": 9055 + }, + { + "epoch": 0.9945091148693169, + "grad_norm": 2.307485580444336, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7067797183990479, + "num_tokens": 231033405.0, + "step": 9056 + }, + { + "epoch": 0.9946189325719306, + "grad_norm": 2.478724241256714, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.6991564631462097, + "num_tokens": 231053475.0, + "step": 9057 + }, + { + "epoch": 0.9947287502745442, + "grad_norm": 2.4273529052734375, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7012084722518921, + "num_tokens": 231080141.0, + "step": 9058 + }, + { + "epoch": 0.994838567977158, + "grad_norm": 2.2752606868743896, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7099701166152954, + "num_tokens": 231103556.0, + "step": 9059 + }, + { + "epoch": 0.9949483856797716, + "grad_norm": 2.256274461746216, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7080071568489075, + "num_tokens": 231128531.0, + "step": 9060 + }, + { + "epoch": 0.9950582033823853, + "grad_norm": 2.0638787746429443, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7350254058837891, + "num_tokens": 231153357.0, + "step": 9061 + }, + { + "epoch": 0.9951680210849989, + "grad_norm": 1.8705734014511108, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7218573093414307, + "num_tokens": 231185395.0, + "step": 9062 + }, + { + "epoch": 0.9952778387876126, + "grad_norm": 2.5553719997406006, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.7036837935447693, + "num_tokens": 231204962.0, + "step": 9063 + }, + { + "epoch": 0.9953876564902262, + "grad_norm": 2.0861024856567383, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7124361991882324, + "num_tokens": 231231616.0, + "step": 9064 + }, + { + "epoch": 0.9954974741928398, + "grad_norm": 2.271489381790161, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7158615589141846, + "num_tokens": 231254595.0, + "step": 9065 + }, + { + "epoch": 0.9956072918954535, + "grad_norm": 2.064051866531372, + "learning_rate": 1e-06, + "loss": 0.9913, + "mean_token_accuracy": 0.6951369643211365, + "num_tokens": 231281946.0, + "step": 9066 + }, + { + "epoch": 0.9957171095980673, + "grad_norm": 2.4562501907348633, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7330964207649231, + "num_tokens": 231301031.0, + "step": 9067 + }, + { + "epoch": 0.9958269273006809, + "grad_norm": 1.9223461151123047, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6927542686462402, + "num_tokens": 231332564.0, + "step": 9068 + }, + { + "epoch": 0.9959367450032945, + "grad_norm": 2.200019121170044, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.726493239402771, + "num_tokens": 231356351.0, + "step": 9069 + }, + { + "epoch": 0.9960465627059082, + "grad_norm": 2.004852056503296, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7161694765090942, + "num_tokens": 231384679.0, + "step": 9070 + }, + { + "epoch": 0.9961563804085218, + "grad_norm": 2.254638671875, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.707717776298523, + "num_tokens": 231409582.0, + "step": 9071 + }, + { + "epoch": 0.9962661981111355, + "grad_norm": 2.2635488510131836, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.717644214630127, + "num_tokens": 231432287.0, + "step": 9072 + }, + { + "epoch": 0.9963760158137491, + "grad_norm": 2.3903307914733887, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7204362154006958, + "num_tokens": 231454053.0, + "step": 9073 + }, + { + "epoch": 0.9964858335163629, + "grad_norm": 2.340378761291504, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.7005519866943359, + "num_tokens": 231478317.0, + "step": 9074 + }, + { + "epoch": 0.9965956512189765, + "grad_norm": 2.301467180252075, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7142978310585022, + "num_tokens": 231503477.0, + "step": 9075 + }, + { + "epoch": 0.9967054689215902, + "grad_norm": 2.679058790206909, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7172940373420715, + "num_tokens": 231523291.0, + "step": 9076 + }, + { + "epoch": 0.9968152866242038, + "grad_norm": 2.3428573608398438, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7180869579315186, + "num_tokens": 231544983.0, + "step": 9077 + }, + { + "epoch": 0.9969251043268175, + "grad_norm": 2.1358792781829834, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7113431692123413, + "num_tokens": 231569265.0, + "step": 9078 + }, + { + "epoch": 0.9970349220294311, + "grad_norm": 2.1348958015441895, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6997544765472412, + "num_tokens": 231596988.0, + "step": 9079 + }, + { + "epoch": 0.9971447397320448, + "grad_norm": 2.2901663780212402, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7032060623168945, + "num_tokens": 231620970.0, + "step": 9080 + }, + { + "epoch": 0.9972545574346585, + "grad_norm": 2.239633083343506, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7195366621017456, + "num_tokens": 231647746.0, + "step": 9081 + }, + { + "epoch": 0.9973643751372722, + "grad_norm": 2.1014811992645264, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7227718830108643, + "num_tokens": 231674062.0, + "step": 9082 + }, + { + "epoch": 0.9974741928398858, + "grad_norm": 2.0692460536956787, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7111406922340393, + "num_tokens": 231700377.0, + "step": 9083 + }, + { + "epoch": 0.9975840105424995, + "grad_norm": 2.214853525161743, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7278109788894653, + "num_tokens": 231723963.0, + "step": 9084 + }, + { + "epoch": 0.9976938282451131, + "grad_norm": 2.012444496154785, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7030923962593079, + "num_tokens": 231751867.0, + "step": 9085 + }, + { + "epoch": 0.9978036459477267, + "grad_norm": 2.1367135047912598, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7022002339363098, + "num_tokens": 231777384.0, + "step": 9086 + }, + { + "epoch": 0.9979134636503404, + "grad_norm": 2.3589236736297607, + "learning_rate": 1e-06, + "loss": 0.7813, + "mean_token_accuracy": 0.7503974437713623, + "num_tokens": 231796708.0, + "step": 9087 + }, + { + "epoch": 0.9980232813529542, + "grad_norm": 2.5412046909332275, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7319918870925903, + "num_tokens": 231815931.0, + "step": 9088 + }, + { + "epoch": 0.9981330990555678, + "grad_norm": 2.194058656692505, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7038323879241943, + "num_tokens": 231839762.0, + "step": 9089 + }, + { + "epoch": 0.9982429167581814, + "grad_norm": 2.289120674133301, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7237802147865295, + "num_tokens": 231861587.0, + "step": 9090 + }, + { + "epoch": 0.9983527344607951, + "grad_norm": 2.150225877761841, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7073869705200195, + "num_tokens": 231887993.0, + "step": 9091 + }, + { + "epoch": 0.9984625521634087, + "grad_norm": 2.4389803409576416, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7115667462348938, + "num_tokens": 231908928.0, + "step": 9092 + }, + { + "epoch": 0.9985723698660224, + "grad_norm": 2.4488465785980225, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7340879440307617, + "num_tokens": 231930862.0, + "step": 9093 + }, + { + "epoch": 0.998682187568636, + "grad_norm": 2.1860945224761963, + "learning_rate": 1e-06, + "loss": 1.0405, + "mean_token_accuracy": 0.6854720115661621, + "num_tokens": 231957527.0, + "step": 9094 + }, + { + "epoch": 0.9987920052712497, + "grad_norm": 2.375307321548462, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7051177620887756, + "num_tokens": 231979796.0, + "step": 9095 + }, + { + "epoch": 0.9989018229738634, + "grad_norm": 2.1824865341186523, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7074016332626343, + "num_tokens": 232002184.0, + "step": 9096 + }, + { + "epoch": 0.9990116406764771, + "grad_norm": 1.8343425989151, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.6984822750091553, + "num_tokens": 232037402.0, + "step": 9097 + }, + { + "epoch": 0.9991214583790907, + "grad_norm": 2.228581666946411, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.6956868767738342, + "num_tokens": 232063721.0, + "step": 9098 + }, + { + "epoch": 0.9992312760817044, + "grad_norm": 2.1354033946990967, + "learning_rate": 1e-06, + "loss": 1.0435, + "mean_token_accuracy": 0.6892374157905579, + "num_tokens": 232090161.0, + "step": 9099 + }, + { + "epoch": 0.999341093784318, + "grad_norm": 2.1694962978363037, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.719079852104187, + "num_tokens": 232117180.0, + "step": 9100 + }, + { + "epoch": 0.9994509114869317, + "grad_norm": 2.052783966064453, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6828700304031372, + "num_tokens": 232146754.0, + "step": 9101 + }, + { + "epoch": 0.9995607291895453, + "grad_norm": 2.3819942474365234, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7018541693687439, + "num_tokens": 232168479.0, + "step": 9102 + }, + { + "epoch": 0.9996705468921591, + "grad_norm": 2.12239670753479, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.712183952331543, + "num_tokens": 232195213.0, + "step": 9103 + }, + { + "epoch": 0.9997803645947727, + "grad_norm": 1.9264034032821655, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7074503302574158, + "num_tokens": 232223500.0, + "step": 9104 + }, + { + "epoch": 0.9998901822973864, + "grad_norm": 2.2263758182525635, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7280411720275879, + "num_tokens": 232246270.0, + "step": 9105 + }, + { + "epoch": 1.0, + "grad_norm": 2.1139488220214844, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7070310115814209, + "num_tokens": 232272548.0, + "step": 9106 + }, + { + "epoch": 1.0001098177026138, + "grad_norm": 2.1910951137542725, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7056354880332947, + "num_tokens": 232296341.0, + "step": 9107 + }, + { + "epoch": 1.0002196354052273, + "grad_norm": 2.2669126987457275, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7025541067123413, + "num_tokens": 232319198.0, + "step": 9108 + }, + { + "epoch": 1.000329453107841, + "grad_norm": 2.1554808616638184, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7279196977615356, + "num_tokens": 232345160.0, + "step": 9109 + }, + { + "epoch": 1.0004392708104546, + "grad_norm": 2.1826846599578857, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7249502539634705, + "num_tokens": 232368852.0, + "step": 9110 + }, + { + "epoch": 1.0005490885130683, + "grad_norm": 2.452033519744873, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7196730375289917, + "num_tokens": 232388617.0, + "step": 9111 + }, + { + "epoch": 1.0006589062156819, + "grad_norm": 2.1286375522613525, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7140035629272461, + "num_tokens": 232414266.0, + "step": 9112 + }, + { + "epoch": 1.0007687239182956, + "grad_norm": 2.147336959838867, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7218752503395081, + "num_tokens": 232440426.0, + "step": 9113 + }, + { + "epoch": 1.0008785416209094, + "grad_norm": 1.9937329292297363, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7147112488746643, + "num_tokens": 232468869.0, + "step": 9114 + }, + { + "epoch": 1.000988359323523, + "grad_norm": 2.104057550430298, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7258819341659546, + "num_tokens": 232495161.0, + "step": 9115 + }, + { + "epoch": 1.0010981770261367, + "grad_norm": 1.7134578227996826, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7082140445709229, + "num_tokens": 232535982.0, + "step": 9116 + }, + { + "epoch": 1.0012079947287502, + "grad_norm": 2.1853115558624268, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.6988940834999084, + "num_tokens": 232563080.0, + "step": 9117 + }, + { + "epoch": 1.001317812431364, + "grad_norm": 2.1530449390411377, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7308052778244019, + "num_tokens": 232588545.0, + "step": 9118 + }, + { + "epoch": 1.0014276301339775, + "grad_norm": 2.0301361083984375, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7221913933753967, + "num_tokens": 232618693.0, + "step": 9119 + }, + { + "epoch": 1.0015374478365913, + "grad_norm": 2.260350227355957, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7290987968444824, + "num_tokens": 232642531.0, + "step": 9120 + }, + { + "epoch": 1.001647265539205, + "grad_norm": 2.040243625640869, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7353794574737549, + "num_tokens": 232671340.0, + "step": 9121 + }, + { + "epoch": 1.0017570832418186, + "grad_norm": 2.129291296005249, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7232702970504761, + "num_tokens": 232697513.0, + "step": 9122 + }, + { + "epoch": 1.0018669009444323, + "grad_norm": 2.3877241611480713, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7024627327919006, + "num_tokens": 232721294.0, + "step": 9123 + }, + { + "epoch": 1.0019767186470458, + "grad_norm": 2.0888864994049072, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7241473197937012, + "num_tokens": 232748210.0, + "step": 9124 + }, + { + "epoch": 1.0020865363496596, + "grad_norm": 1.948437213897705, + "learning_rate": 1e-06, + "loss": 1.0834, + "mean_token_accuracy": 0.6705812215805054, + "num_tokens": 232779716.0, + "step": 9125 + }, + { + "epoch": 1.0021963540522731, + "grad_norm": 2.4881112575531006, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7303065061569214, + "num_tokens": 232800460.0, + "step": 9126 + }, + { + "epoch": 1.002306171754887, + "grad_norm": 2.23689341545105, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7179278135299683, + "num_tokens": 232824247.0, + "step": 9127 + }, + { + "epoch": 1.0024159894575007, + "grad_norm": 2.3025929927825928, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.6922709941864014, + "num_tokens": 232848090.0, + "step": 9128 + }, + { + "epoch": 1.0025258071601142, + "grad_norm": 2.4161369800567627, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7384819984436035, + "num_tokens": 232870003.0, + "step": 9129 + }, + { + "epoch": 1.002635624862728, + "grad_norm": 2.236027240753174, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7402188777923584, + "num_tokens": 232892677.0, + "step": 9130 + }, + { + "epoch": 1.0027454425653415, + "grad_norm": 2.107595443725586, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7121627330780029, + "num_tokens": 232919328.0, + "step": 9131 + }, + { + "epoch": 1.0028552602679552, + "grad_norm": 2.0813825130462646, + "learning_rate": 1e-06, + "loss": 1.05, + "mean_token_accuracy": 0.6779706478118896, + "num_tokens": 232949790.0, + "step": 9132 + }, + { + "epoch": 1.0029650779705688, + "grad_norm": 2.2008540630340576, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7307397127151489, + "num_tokens": 232974749.0, + "step": 9133 + }, + { + "epoch": 1.0030748956731825, + "grad_norm": 2.111679792404175, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.729435920715332, + "num_tokens": 232999600.0, + "step": 9134 + }, + { + "epoch": 1.0031847133757963, + "grad_norm": 2.255105495452881, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7162705659866333, + "num_tokens": 233025900.0, + "step": 9135 + }, + { + "epoch": 1.0032945310784098, + "grad_norm": 2.071063995361328, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7004572153091431, + "num_tokens": 233054439.0, + "step": 9136 + }, + { + "epoch": 1.0034043487810236, + "grad_norm": 2.2485733032226562, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7376016974449158, + "num_tokens": 233077626.0, + "step": 9137 + }, + { + "epoch": 1.0035141664836371, + "grad_norm": 2.269613742828369, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7261434197425842, + "num_tokens": 233100985.0, + "step": 9138 + }, + { + "epoch": 1.0036239841862509, + "grad_norm": 2.1955738067626953, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7136189937591553, + "num_tokens": 233127244.0, + "step": 9139 + }, + { + "epoch": 1.0037338018888644, + "grad_norm": 2.416245698928833, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7201710939407349, + "num_tokens": 233149842.0, + "step": 9140 + }, + { + "epoch": 1.0038436195914782, + "grad_norm": 2.21252703666687, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7090708017349243, + "num_tokens": 233175065.0, + "step": 9141 + }, + { + "epoch": 1.0039534372940917, + "grad_norm": 2.0348994731903076, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6892502307891846, + "num_tokens": 233207552.0, + "step": 9142 + }, + { + "epoch": 1.0040632549967055, + "grad_norm": 1.9846051931381226, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7099142074584961, + "num_tokens": 233237535.0, + "step": 9143 + }, + { + "epoch": 1.0041730726993192, + "grad_norm": 2.7773635387420654, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7353415489196777, + "num_tokens": 233256477.0, + "step": 9144 + }, + { + "epoch": 1.0042828904019327, + "grad_norm": 1.9560879468917847, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7158551216125488, + "num_tokens": 233288083.0, + "step": 9145 + }, + { + "epoch": 1.0043927081045465, + "grad_norm": 2.2086427211761475, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7133811712265015, + "num_tokens": 233314535.0, + "step": 9146 + }, + { + "epoch": 1.00450252580716, + "grad_norm": 1.8497097492218018, + "learning_rate": 1e-06, + "loss": 1.0108, + "mean_token_accuracy": 0.6905924081802368, + "num_tokens": 233349767.0, + "step": 9147 + }, + { + "epoch": 1.0046123435097738, + "grad_norm": 2.2358009815216064, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7270201444625854, + "num_tokens": 233375283.0, + "step": 9148 + }, + { + "epoch": 1.0047221612123873, + "grad_norm": 2.17989182472229, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7171928882598877, + "num_tokens": 233399951.0, + "step": 9149 + }, + { + "epoch": 1.004831978915001, + "grad_norm": 2.775707483291626, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7351824641227722, + "num_tokens": 233416912.0, + "step": 9150 + }, + { + "epoch": 1.0049417966176148, + "grad_norm": 2.26674485206604, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7089325189590454, + "num_tokens": 233443299.0, + "step": 9151 + }, + { + "epoch": 1.0050516143202284, + "grad_norm": 2.0413317680358887, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.7006462216377258, + "num_tokens": 233474074.0, + "step": 9152 + }, + { + "epoch": 1.0051614320228421, + "grad_norm": 2.396779775619507, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7284536361694336, + "num_tokens": 233495430.0, + "step": 9153 + }, + { + "epoch": 1.0052712497254557, + "grad_norm": 2.0469985008239746, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7096397876739502, + "num_tokens": 233522492.0, + "step": 9154 + }, + { + "epoch": 1.0053810674280694, + "grad_norm": 2.234302520751953, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7216134071350098, + "num_tokens": 233549278.0, + "step": 9155 + }, + { + "epoch": 1.005490885130683, + "grad_norm": 2.0612781047821045, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.725784182548523, + "num_tokens": 233578956.0, + "step": 9156 + }, + { + "epoch": 1.0056007028332967, + "grad_norm": 2.2108561992645264, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7298058271408081, + "num_tokens": 233605398.0, + "step": 9157 + }, + { + "epoch": 1.0057105205359105, + "grad_norm": 2.5145299434661865, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7030872106552124, + "num_tokens": 233626440.0, + "step": 9158 + }, + { + "epoch": 1.005820338238524, + "grad_norm": 2.0401721000671387, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.7001403570175171, + "num_tokens": 233657524.0, + "step": 9159 + }, + { + "epoch": 1.0059301559411378, + "grad_norm": 2.391267776489258, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7288421988487244, + "num_tokens": 233678956.0, + "step": 9160 + }, + { + "epoch": 1.0060399736437513, + "grad_norm": 2.121403217315674, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7208202481269836, + "num_tokens": 233707644.0, + "step": 9161 + }, + { + "epoch": 1.006149791346365, + "grad_norm": 2.637866973876953, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7195881605148315, + "num_tokens": 233726471.0, + "step": 9162 + }, + { + "epoch": 1.0062596090489786, + "grad_norm": 2.66709041595459, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7166448831558228, + "num_tokens": 233746041.0, + "step": 9163 + }, + { + "epoch": 1.0063694267515924, + "grad_norm": 2.30570387840271, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7112181186676025, + "num_tokens": 233768816.0, + "step": 9164 + }, + { + "epoch": 1.0064792444542061, + "grad_norm": 2.3613696098327637, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7090812921524048, + "num_tokens": 233793021.0, + "step": 9165 + }, + { + "epoch": 1.0065890621568196, + "grad_norm": 2.22408390045166, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7176668643951416, + "num_tokens": 233818972.0, + "step": 9166 + }, + { + "epoch": 1.0066988798594334, + "grad_norm": 2.469719171524048, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7292810678482056, + "num_tokens": 233839380.0, + "step": 9167 + }, + { + "epoch": 1.006808697562047, + "grad_norm": 2.3404436111450195, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7315322160720825, + "num_tokens": 233861779.0, + "step": 9168 + }, + { + "epoch": 1.0069185152646607, + "grad_norm": 2.0606296062469482, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7177537679672241, + "num_tokens": 233889782.0, + "step": 9169 + }, + { + "epoch": 1.0070283329672742, + "grad_norm": 2.3638501167297363, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7334687113761902, + "num_tokens": 233910394.0, + "step": 9170 + }, + { + "epoch": 1.007138150669888, + "grad_norm": 2.2442548274993896, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7245992422103882, + "num_tokens": 233936013.0, + "step": 9171 + }, + { + "epoch": 1.0072479683725017, + "grad_norm": 2.3014259338378906, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7140929102897644, + "num_tokens": 233959062.0, + "step": 9172 + }, + { + "epoch": 1.0073577860751153, + "grad_norm": 1.96196448802948, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7055312395095825, + "num_tokens": 233992631.0, + "step": 9173 + }, + { + "epoch": 1.007467603777729, + "grad_norm": 2.0469577312469482, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7122766375541687, + "num_tokens": 234022268.0, + "step": 9174 + }, + { + "epoch": 1.0075774214803426, + "grad_norm": 2.3620717525482178, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7155607342720032, + "num_tokens": 234045121.0, + "step": 9175 + }, + { + "epoch": 1.0076872391829563, + "grad_norm": 2.2165162563323975, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.721860945224762, + "num_tokens": 234068231.0, + "step": 9176 + }, + { + "epoch": 1.0077970568855699, + "grad_norm": 2.4862256050109863, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7410283088684082, + "num_tokens": 234089748.0, + "step": 9177 + }, + { + "epoch": 1.0079068745881836, + "grad_norm": 2.2614123821258545, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7279090881347656, + "num_tokens": 234112929.0, + "step": 9178 + }, + { + "epoch": 1.0080166922907974, + "grad_norm": 1.9011878967285156, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.7029953002929688, + "num_tokens": 234144230.0, + "step": 9179 + }, + { + "epoch": 1.008126509993411, + "grad_norm": 2.1846859455108643, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7361164689064026, + "num_tokens": 234169719.0, + "step": 9180 + }, + { + "epoch": 1.0082363276960247, + "grad_norm": 2.148597240447998, + "learning_rate": 1e-06, + "loss": 0.831, + "mean_token_accuracy": 0.7378333806991577, + "num_tokens": 234195027.0, + "step": 9181 + }, + { + "epoch": 1.0083461453986382, + "grad_norm": 2.115898847579956, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.736517071723938, + "num_tokens": 234219815.0, + "step": 9182 + }, + { + "epoch": 1.008455963101252, + "grad_norm": 2.3783175945281982, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7330253720283508, + "num_tokens": 234242795.0, + "step": 9183 + }, + { + "epoch": 1.0085657808038655, + "grad_norm": 2.4930121898651123, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7119182348251343, + "num_tokens": 234263824.0, + "step": 9184 + }, + { + "epoch": 1.0086755985064793, + "grad_norm": 2.2496743202209473, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7086610794067383, + "num_tokens": 234290447.0, + "step": 9185 + }, + { + "epoch": 1.008785416209093, + "grad_norm": 2.1013693809509277, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.711898922920227, + "num_tokens": 234319193.0, + "step": 9186 + }, + { + "epoch": 1.0088952339117065, + "grad_norm": 2.2535717487335205, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6984288692474365, + "num_tokens": 234346721.0, + "step": 9187 + }, + { + "epoch": 1.0090050516143203, + "grad_norm": 2.258082389831543, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6941660642623901, + "num_tokens": 234373742.0, + "step": 9188 + }, + { + "epoch": 1.0091148693169338, + "grad_norm": 2.1865909099578857, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7219324707984924, + "num_tokens": 234398668.0, + "step": 9189 + }, + { + "epoch": 1.0092246870195476, + "grad_norm": 2.740856885910034, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.7553524971008301, + "num_tokens": 234415091.0, + "step": 9190 + }, + { + "epoch": 1.0093345047221611, + "grad_norm": 2.225062608718872, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7217538952827454, + "num_tokens": 234441306.0, + "step": 9191 + }, + { + "epoch": 1.0094443224247749, + "grad_norm": 2.2706308364868164, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7104220390319824, + "num_tokens": 234465869.0, + "step": 9192 + }, + { + "epoch": 1.0095541401273886, + "grad_norm": 2.553508996963501, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7412919998168945, + "num_tokens": 234483736.0, + "step": 9193 + }, + { + "epoch": 1.0096639578300022, + "grad_norm": 2.404259443283081, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7160663604736328, + "num_tokens": 234505198.0, + "step": 9194 + }, + { + "epoch": 1.009773775532616, + "grad_norm": 1.935351014137268, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7128617763519287, + "num_tokens": 234537743.0, + "step": 9195 + }, + { + "epoch": 1.0098835932352295, + "grad_norm": 2.464240074157715, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7296674251556396, + "num_tokens": 234559248.0, + "step": 9196 + }, + { + "epoch": 1.0099934109378432, + "grad_norm": 2.3555283546447754, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7105530500411987, + "num_tokens": 234581858.0, + "step": 9197 + }, + { + "epoch": 1.0101032286404568, + "grad_norm": 2.1899447441101074, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7199720144271851, + "num_tokens": 234609289.0, + "step": 9198 + }, + { + "epoch": 1.0102130463430705, + "grad_norm": 2.042597770690918, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.6987345218658447, + "num_tokens": 234638524.0, + "step": 9199 + }, + { + "epoch": 1.010322864045684, + "grad_norm": 2.1797220706939697, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7157037258148193, + "num_tokens": 234665096.0, + "step": 9200 + }, + { + "epoch": 1.0104326817482978, + "grad_norm": 2.0970499515533447, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6896452307701111, + "num_tokens": 234692481.0, + "step": 9201 + }, + { + "epoch": 1.0105424994509116, + "grad_norm": 2.168778419494629, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7224664092063904, + "num_tokens": 234718590.0, + "step": 9202 + }, + { + "epoch": 1.010652317153525, + "grad_norm": 1.9511758089065552, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.715529203414917, + "num_tokens": 234751465.0, + "step": 9203 + }, + { + "epoch": 1.0107621348561389, + "grad_norm": 2.190727949142456, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7134035229682922, + "num_tokens": 234778256.0, + "step": 9204 + }, + { + "epoch": 1.0108719525587524, + "grad_norm": 2.0603489875793457, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.740909993648529, + "num_tokens": 234805242.0, + "step": 9205 + }, + { + "epoch": 1.0109817702613662, + "grad_norm": 2.179598569869995, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7265405654907227, + "num_tokens": 234832954.0, + "step": 9206 + }, + { + "epoch": 1.0110915879639797, + "grad_norm": 2.201605796813965, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7142757773399353, + "num_tokens": 234860819.0, + "step": 9207 + }, + { + "epoch": 1.0112014056665934, + "grad_norm": 2.340069055557251, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.6971547603607178, + "num_tokens": 234884743.0, + "step": 9208 + }, + { + "epoch": 1.0113112233692072, + "grad_norm": 1.9314439296722412, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7212719917297363, + "num_tokens": 234918403.0, + "step": 9209 + }, + { + "epoch": 1.0114210410718207, + "grad_norm": 2.0495445728302, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.6910099983215332, + "num_tokens": 234954639.0, + "step": 9210 + }, + { + "epoch": 1.0115308587744345, + "grad_norm": 2.3032522201538086, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.736370325088501, + "num_tokens": 234979080.0, + "step": 9211 + }, + { + "epoch": 1.011640676477048, + "grad_norm": 2.226613759994507, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7285901308059692, + "num_tokens": 235002615.0, + "step": 9212 + }, + { + "epoch": 1.0117504941796618, + "grad_norm": 2.4316864013671875, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7338850498199463, + "num_tokens": 235024864.0, + "step": 9213 + }, + { + "epoch": 1.0118603118822753, + "grad_norm": 1.9800399541854858, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7196857929229736, + "num_tokens": 235054325.0, + "step": 9214 + }, + { + "epoch": 1.011970129584889, + "grad_norm": 2.275012254714966, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7149076461791992, + "num_tokens": 235078956.0, + "step": 9215 + }, + { + "epoch": 1.0120799472875028, + "grad_norm": 2.293678045272827, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7201329469680786, + "num_tokens": 235104278.0, + "step": 9216 + }, + { + "epoch": 1.0121897649901164, + "grad_norm": 2.3545901775360107, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7174273729324341, + "num_tokens": 235127724.0, + "step": 9217 + }, + { + "epoch": 1.0122995826927301, + "grad_norm": 2.589547872543335, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.7269679307937622, + "num_tokens": 235146878.0, + "step": 9218 + }, + { + "epoch": 1.0124094003953437, + "grad_norm": 2.1000115871429443, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7174659967422485, + "num_tokens": 235177125.0, + "step": 9219 + }, + { + "epoch": 1.0125192180979574, + "grad_norm": 2.2760393619537354, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7697239518165588, + "num_tokens": 235199510.0, + "step": 9220 + }, + { + "epoch": 1.012629035800571, + "grad_norm": 2.7410333156585693, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7149969339370728, + "num_tokens": 235218843.0, + "step": 9221 + }, + { + "epoch": 1.0127388535031847, + "grad_norm": 2.6994338035583496, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7437794208526611, + "num_tokens": 235237463.0, + "step": 9222 + }, + { + "epoch": 1.0128486712057985, + "grad_norm": 2.5141780376434326, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7090706825256348, + "num_tokens": 235260910.0, + "step": 9223 + }, + { + "epoch": 1.012958488908412, + "grad_norm": 2.0466349124908447, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7028703689575195, + "num_tokens": 235293025.0, + "step": 9224 + }, + { + "epoch": 1.0130683066110258, + "grad_norm": 2.4781336784362793, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7373913526535034, + "num_tokens": 235315099.0, + "step": 9225 + }, + { + "epoch": 1.0131781243136393, + "grad_norm": 2.138252019882202, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7207040786743164, + "num_tokens": 235343930.0, + "step": 9226 + }, + { + "epoch": 1.013287942016253, + "grad_norm": 2.4540536403656006, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.730949878692627, + "num_tokens": 235364729.0, + "step": 9227 + }, + { + "epoch": 1.0133977597188666, + "grad_norm": 2.5506904125213623, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.723793625831604, + "num_tokens": 235385638.0, + "step": 9228 + }, + { + "epoch": 1.0135075774214803, + "grad_norm": 2.25445556640625, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7437719106674194, + "num_tokens": 235408736.0, + "step": 9229 + }, + { + "epoch": 1.013617395124094, + "grad_norm": 2.103238344192505, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7207863926887512, + "num_tokens": 235436597.0, + "step": 9230 + }, + { + "epoch": 1.0137272128267076, + "grad_norm": 2.2754085063934326, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7260195016860962, + "num_tokens": 235460238.0, + "step": 9231 + }, + { + "epoch": 1.0138370305293214, + "grad_norm": 2.5830931663513184, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7288482189178467, + "num_tokens": 235480454.0, + "step": 9232 + }, + { + "epoch": 1.013946848231935, + "grad_norm": 2.076172351837158, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7169192433357239, + "num_tokens": 235509549.0, + "step": 9233 + }, + { + "epoch": 1.0140566659345487, + "grad_norm": 2.158383369445801, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7033501863479614, + "num_tokens": 235538114.0, + "step": 9234 + }, + { + "epoch": 1.0141664836371622, + "grad_norm": 2.3261823654174805, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.721896231174469, + "num_tokens": 235562011.0, + "step": 9235 + }, + { + "epoch": 1.014276301339776, + "grad_norm": 2.401615619659424, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.7051371335983276, + "num_tokens": 235583219.0, + "step": 9236 + }, + { + "epoch": 1.0143861190423897, + "grad_norm": 2.251309871673584, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7100456953048706, + "num_tokens": 235611534.0, + "step": 9237 + }, + { + "epoch": 1.0144959367450033, + "grad_norm": 2.5308480262756348, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7152626514434814, + "num_tokens": 235632358.0, + "step": 9238 + }, + { + "epoch": 1.014605754447617, + "grad_norm": 2.3431456089019775, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7402619123458862, + "num_tokens": 235655308.0, + "step": 9239 + }, + { + "epoch": 1.0147155721502306, + "grad_norm": 2.028477430343628, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7192488312721252, + "num_tokens": 235685225.0, + "step": 9240 + }, + { + "epoch": 1.0148253898528443, + "grad_norm": 2.1792895793914795, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7138099074363708, + "num_tokens": 235711411.0, + "step": 9241 + }, + { + "epoch": 1.0149352075554579, + "grad_norm": 2.47717547416687, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7282823920249939, + "num_tokens": 235732489.0, + "step": 9242 + }, + { + "epoch": 1.0150450252580716, + "grad_norm": 2.2694268226623535, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7162141799926758, + "num_tokens": 235756527.0, + "step": 9243 + }, + { + "epoch": 1.0151548429606854, + "grad_norm": 2.426373243331909, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7342823147773743, + "num_tokens": 235777983.0, + "step": 9244 + }, + { + "epoch": 1.015264660663299, + "grad_norm": 1.9591248035430908, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.6966174840927124, + "num_tokens": 235808162.0, + "step": 9245 + }, + { + "epoch": 1.0153744783659127, + "grad_norm": 2.077432870864868, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7172964215278625, + "num_tokens": 235839550.0, + "step": 9246 + }, + { + "epoch": 1.0154842960685262, + "grad_norm": 2.245274066925049, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7222847938537598, + "num_tokens": 235865859.0, + "step": 9247 + }, + { + "epoch": 1.01559411377114, + "grad_norm": 2.029698610305786, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7148613333702087, + "num_tokens": 235893785.0, + "step": 9248 + }, + { + "epoch": 1.0157039314737535, + "grad_norm": 2.118054151535034, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7370684146881104, + "num_tokens": 235919520.0, + "step": 9249 + }, + { + "epoch": 1.0158137491763672, + "grad_norm": 2.279405117034912, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7356846332550049, + "num_tokens": 235941377.0, + "step": 9250 + }, + { + "epoch": 1.015923566878981, + "grad_norm": 2.5083487033843994, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7114288806915283, + "num_tokens": 235962876.0, + "step": 9251 + }, + { + "epoch": 1.0160333845815945, + "grad_norm": 2.3462018966674805, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7107030153274536, + "num_tokens": 235986707.0, + "step": 9252 + }, + { + "epoch": 1.0161432022842083, + "grad_norm": 2.3126938343048096, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7468258142471313, + "num_tokens": 236009440.0, + "step": 9253 + }, + { + "epoch": 1.0162530199868218, + "grad_norm": 2.2293198108673096, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7336567640304565, + "num_tokens": 236034423.0, + "step": 9254 + }, + { + "epoch": 1.0163628376894356, + "grad_norm": 2.4315223693847656, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7121021747589111, + "num_tokens": 236055943.0, + "step": 9255 + }, + { + "epoch": 1.0164726553920491, + "grad_norm": 2.3251304626464844, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7124308347702026, + "num_tokens": 236080040.0, + "step": 9256 + }, + { + "epoch": 1.0165824730946629, + "grad_norm": 2.4227609634399414, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7371000051498413, + "num_tokens": 236100273.0, + "step": 9257 + }, + { + "epoch": 1.0166922907972764, + "grad_norm": 2.294731616973877, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.703061580657959, + "num_tokens": 236123289.0, + "step": 9258 + }, + { + "epoch": 1.0168021084998902, + "grad_norm": 2.428791046142578, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7314766049385071, + "num_tokens": 236146248.0, + "step": 9259 + }, + { + "epoch": 1.016911926202504, + "grad_norm": 1.9522817134857178, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.6951687335968018, + "num_tokens": 236176241.0, + "step": 9260 + }, + { + "epoch": 1.0170217439051175, + "grad_norm": 2.317688465118408, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7086285352706909, + "num_tokens": 236200361.0, + "step": 9261 + }, + { + "epoch": 1.0171315616077312, + "grad_norm": 2.146064281463623, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7033114433288574, + "num_tokens": 236228370.0, + "step": 9262 + }, + { + "epoch": 1.0172413793103448, + "grad_norm": 2.381836175918579, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7369212508201599, + "num_tokens": 236249003.0, + "step": 9263 + }, + { + "epoch": 1.0173511970129585, + "grad_norm": 2.4802162647247314, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7510643601417542, + "num_tokens": 236268936.0, + "step": 9264 + }, + { + "epoch": 1.017461014715572, + "grad_norm": 2.3435702323913574, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7194651365280151, + "num_tokens": 236292346.0, + "step": 9265 + }, + { + "epoch": 1.0175708324181858, + "grad_norm": 2.250014543533325, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7195396423339844, + "num_tokens": 236318272.0, + "step": 9266 + }, + { + "epoch": 1.0176806501207996, + "grad_norm": 2.1551198959350586, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.728649914264679, + "num_tokens": 236346462.0, + "step": 9267 + }, + { + "epoch": 1.017790467823413, + "grad_norm": 2.3388772010803223, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.697342574596405, + "num_tokens": 236371797.0, + "step": 9268 + }, + { + "epoch": 1.0179002855260268, + "grad_norm": 2.6257402896881104, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7261237502098083, + "num_tokens": 236389779.0, + "step": 9269 + }, + { + "epoch": 1.0180101032286404, + "grad_norm": 2.2355127334594727, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7110472917556763, + "num_tokens": 236415035.0, + "step": 9270 + }, + { + "epoch": 1.0181199209312541, + "grad_norm": 2.1703319549560547, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7135614156723022, + "num_tokens": 236441876.0, + "step": 9271 + }, + { + "epoch": 1.0182297386338677, + "grad_norm": 1.9456820487976074, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7174192070960999, + "num_tokens": 236473413.0, + "step": 9272 + }, + { + "epoch": 1.0183395563364814, + "grad_norm": 1.9695274829864502, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.7444146275520325, + "num_tokens": 236500760.0, + "step": 9273 + }, + { + "epoch": 1.0184493740390952, + "grad_norm": 2.4727892875671387, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7247117757797241, + "num_tokens": 236521627.0, + "step": 9274 + }, + { + "epoch": 1.0185591917417087, + "grad_norm": 2.1539089679718018, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6927722692489624, + "num_tokens": 236550162.0, + "step": 9275 + }, + { + "epoch": 1.0186690094443225, + "grad_norm": 2.3405511379241943, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7099920511245728, + "num_tokens": 236574872.0, + "step": 9276 + }, + { + "epoch": 1.018778827146936, + "grad_norm": 2.7286651134490967, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7318053245544434, + "num_tokens": 236592220.0, + "step": 9277 + }, + { + "epoch": 1.0188886448495498, + "grad_norm": 2.2404229640960693, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7073412537574768, + "num_tokens": 236618176.0, + "step": 9278 + }, + { + "epoch": 1.0189984625521633, + "grad_norm": 2.068937301635742, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7120437622070312, + "num_tokens": 236646566.0, + "step": 9279 + }, + { + "epoch": 1.019108280254777, + "grad_norm": 2.3004982471466064, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7187344431877136, + "num_tokens": 236671452.0, + "step": 9280 + }, + { + "epoch": 1.0192180979573908, + "grad_norm": 2.5074386596679688, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7341694831848145, + "num_tokens": 236691074.0, + "step": 9281 + }, + { + "epoch": 1.0193279156600044, + "grad_norm": 2.4781196117401123, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7241998314857483, + "num_tokens": 236712527.0, + "step": 9282 + }, + { + "epoch": 1.0194377333626181, + "grad_norm": 2.5615556240081787, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7226911187171936, + "num_tokens": 236733231.0, + "step": 9283 + }, + { + "epoch": 1.0195475510652316, + "grad_norm": 2.203214168548584, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7313479781150818, + "num_tokens": 236759001.0, + "step": 9284 + }, + { + "epoch": 1.0196573687678454, + "grad_norm": 2.3227734565734863, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7114434242248535, + "num_tokens": 236783734.0, + "step": 9285 + }, + { + "epoch": 1.019767186470459, + "grad_norm": 2.0246872901916504, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7172357439994812, + "num_tokens": 236813059.0, + "step": 9286 + }, + { + "epoch": 1.0198770041730727, + "grad_norm": 2.2165729999542236, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7186768054962158, + "num_tokens": 236837655.0, + "step": 9287 + }, + { + "epoch": 1.0199868218756865, + "grad_norm": 1.866461157798767, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7005904912948608, + "num_tokens": 236874637.0, + "step": 9288 + }, + { + "epoch": 1.0200966395783, + "grad_norm": 2.3769690990448, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7217425107955933, + "num_tokens": 236897336.0, + "step": 9289 + }, + { + "epoch": 1.0202064572809137, + "grad_norm": 2.2167818546295166, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7334190607070923, + "num_tokens": 236922723.0, + "step": 9290 + }, + { + "epoch": 1.0203162749835273, + "grad_norm": 2.2859199047088623, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7140353918075562, + "num_tokens": 236946577.0, + "step": 9291 + }, + { + "epoch": 1.020426092686141, + "grad_norm": 2.358327627182007, + "learning_rate": 1e-06, + "loss": 0.7949, + "mean_token_accuracy": 0.746673047542572, + "num_tokens": 236968009.0, + "step": 9292 + }, + { + "epoch": 1.0205359103887546, + "grad_norm": 1.9787262678146362, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7450516819953918, + "num_tokens": 236998928.0, + "step": 9293 + }, + { + "epoch": 1.0206457280913683, + "grad_norm": 2.288344383239746, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7071138620376587, + "num_tokens": 237025407.0, + "step": 9294 + }, + { + "epoch": 1.020755545793982, + "grad_norm": 2.50901198387146, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7190262079238892, + "num_tokens": 237046305.0, + "step": 9295 + }, + { + "epoch": 1.0208653634965956, + "grad_norm": 2.447166919708252, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.7474203109741211, + "num_tokens": 237066156.0, + "step": 9296 + }, + { + "epoch": 1.0209751811992094, + "grad_norm": 2.304048776626587, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.711640477180481, + "num_tokens": 237091680.0, + "step": 9297 + }, + { + "epoch": 1.021084998901823, + "grad_norm": 2.145920753479004, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.6996363401412964, + "num_tokens": 237120593.0, + "step": 9298 + }, + { + "epoch": 1.0211948166044367, + "grad_norm": 2.1246955394744873, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7217175960540771, + "num_tokens": 237147591.0, + "step": 9299 + }, + { + "epoch": 1.0213046343070502, + "grad_norm": 2.462683916091919, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7324053049087524, + "num_tokens": 237170029.0, + "step": 9300 + }, + { + "epoch": 1.021414452009664, + "grad_norm": 2.5345942974090576, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7132930159568787, + "num_tokens": 237191814.0, + "step": 9301 + }, + { + "epoch": 1.0215242697122777, + "grad_norm": 2.0864310264587402, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7248513698577881, + "num_tokens": 237218181.0, + "step": 9302 + }, + { + "epoch": 1.0216340874148913, + "grad_norm": 2.2479519844055176, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7362825870513916, + "num_tokens": 237241890.0, + "step": 9303 + }, + { + "epoch": 1.021743905117505, + "grad_norm": 2.4593636989593506, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7213384509086609, + "num_tokens": 237264070.0, + "step": 9304 + }, + { + "epoch": 1.0218537228201185, + "grad_norm": 2.011784791946411, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7111660242080688, + "num_tokens": 237297295.0, + "step": 9305 + }, + { + "epoch": 1.0219635405227323, + "grad_norm": 2.299987554550171, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7280224561691284, + "num_tokens": 237324274.0, + "step": 9306 + }, + { + "epoch": 1.0220733582253458, + "grad_norm": 2.0810775756835938, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7076939940452576, + "num_tokens": 237353287.0, + "step": 9307 + }, + { + "epoch": 1.0221831759279596, + "grad_norm": 2.1595613956451416, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7191116809844971, + "num_tokens": 237380683.0, + "step": 9308 + }, + { + "epoch": 1.0222929936305734, + "grad_norm": 1.980452537536621, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7238900065422058, + "num_tokens": 237408951.0, + "step": 9309 + }, + { + "epoch": 1.022402811333187, + "grad_norm": 2.5323686599731445, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7056022882461548, + "num_tokens": 237432332.0, + "step": 9310 + }, + { + "epoch": 1.0225126290358006, + "grad_norm": 2.32460618019104, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7164547443389893, + "num_tokens": 237458955.0, + "step": 9311 + }, + { + "epoch": 1.0226224467384142, + "grad_norm": 2.282121181488037, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7073453068733215, + "num_tokens": 237488178.0, + "step": 9312 + }, + { + "epoch": 1.022732264441028, + "grad_norm": 2.111024856567383, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7177083492279053, + "num_tokens": 237513957.0, + "step": 9313 + }, + { + "epoch": 1.0228420821436415, + "grad_norm": 2.1785974502563477, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7305281162261963, + "num_tokens": 237539797.0, + "step": 9314 + }, + { + "epoch": 1.0229518998462552, + "grad_norm": 2.3847880363464355, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7231201529502869, + "num_tokens": 237563738.0, + "step": 9315 + }, + { + "epoch": 1.023061717548869, + "grad_norm": 2.3048417568206787, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7437662482261658, + "num_tokens": 237586742.0, + "step": 9316 + }, + { + "epoch": 1.0231715352514825, + "grad_norm": 2.1885669231414795, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7154305577278137, + "num_tokens": 237614219.0, + "step": 9317 + }, + { + "epoch": 1.0232813529540963, + "grad_norm": 2.1831846237182617, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7203496098518372, + "num_tokens": 237640061.0, + "step": 9318 + }, + { + "epoch": 1.0233911706567098, + "grad_norm": 2.2315480709075928, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7188829183578491, + "num_tokens": 237665997.0, + "step": 9319 + }, + { + "epoch": 1.0235009883593236, + "grad_norm": 2.071943521499634, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7055596709251404, + "num_tokens": 237699742.0, + "step": 9320 + }, + { + "epoch": 1.023610806061937, + "grad_norm": 2.0076818466186523, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7116543054580688, + "num_tokens": 237729769.0, + "step": 9321 + }, + { + "epoch": 1.0237206237645509, + "grad_norm": 2.5916755199432373, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7332661151885986, + "num_tokens": 237749688.0, + "step": 9322 + }, + { + "epoch": 1.0238304414671644, + "grad_norm": 2.2703616619110107, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7217180132865906, + "num_tokens": 237774404.0, + "step": 9323 + }, + { + "epoch": 1.0239402591697782, + "grad_norm": 2.0725417137145996, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7016111612319946, + "num_tokens": 237806329.0, + "step": 9324 + }, + { + "epoch": 1.024050076872392, + "grad_norm": 2.308415412902832, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7183532118797302, + "num_tokens": 237836001.0, + "step": 9325 + }, + { + "epoch": 1.0241598945750054, + "grad_norm": 2.247673273086548, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7249168753623962, + "num_tokens": 237860691.0, + "step": 9326 + }, + { + "epoch": 1.0242697122776192, + "grad_norm": 2.2189300060272217, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7021161913871765, + "num_tokens": 237885478.0, + "step": 9327 + }, + { + "epoch": 1.0243795299802327, + "grad_norm": 2.385371446609497, + "learning_rate": 1e-06, + "loss": 0.7992, + "mean_token_accuracy": 0.7477542161941528, + "num_tokens": 237906031.0, + "step": 9328 + }, + { + "epoch": 1.0244893476828465, + "grad_norm": 2.046398639678955, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7127727270126343, + "num_tokens": 237934055.0, + "step": 9329 + }, + { + "epoch": 1.02459916538546, + "grad_norm": 2.381269931793213, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7294642925262451, + "num_tokens": 237955187.0, + "step": 9330 + }, + { + "epoch": 1.0247089830880738, + "grad_norm": 2.3306713104248047, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.734965443611145, + "num_tokens": 237977162.0, + "step": 9331 + }, + { + "epoch": 1.0248188007906875, + "grad_norm": 1.9818814992904663, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7096363306045532, + "num_tokens": 238007337.0, + "step": 9332 + }, + { + "epoch": 1.024928618493301, + "grad_norm": 2.2490999698638916, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7461605072021484, + "num_tokens": 238030522.0, + "step": 9333 + }, + { + "epoch": 1.0250384361959148, + "grad_norm": 2.358110189437866, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7079172730445862, + "num_tokens": 238053652.0, + "step": 9334 + }, + { + "epoch": 1.0251482538985284, + "grad_norm": 1.96181058883667, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7347704768180847, + "num_tokens": 238084644.0, + "step": 9335 + }, + { + "epoch": 1.0252580716011421, + "grad_norm": 2.4684271812438965, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7204409837722778, + "num_tokens": 238105073.0, + "step": 9336 + }, + { + "epoch": 1.0253678893037557, + "grad_norm": 1.9519158601760864, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7130141258239746, + "num_tokens": 238139292.0, + "step": 9337 + }, + { + "epoch": 1.0254777070063694, + "grad_norm": 2.3395347595214844, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7357374429702759, + "num_tokens": 238162379.0, + "step": 9338 + }, + { + "epoch": 1.0255875247089832, + "grad_norm": 2.3888847827911377, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7401741743087769, + "num_tokens": 238184678.0, + "step": 9339 + }, + { + "epoch": 1.0256973424115967, + "grad_norm": 1.9658277034759521, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.6979814767837524, + "num_tokens": 238217938.0, + "step": 9340 + }, + { + "epoch": 1.0258071601142105, + "grad_norm": 2.232072114944458, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7248923182487488, + "num_tokens": 238243945.0, + "step": 9341 + }, + { + "epoch": 1.025916977816824, + "grad_norm": 2.126427412033081, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7088189721107483, + "num_tokens": 238273574.0, + "step": 9342 + }, + { + "epoch": 1.0260267955194378, + "grad_norm": 2.186697244644165, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7377109527587891, + "num_tokens": 238297876.0, + "step": 9343 + }, + { + "epoch": 1.0261366132220513, + "grad_norm": 2.095735549926758, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7288870811462402, + "num_tokens": 238325210.0, + "step": 9344 + }, + { + "epoch": 1.026246430924665, + "grad_norm": 1.9799413681030273, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.700369119644165, + "num_tokens": 238357852.0, + "step": 9345 + }, + { + "epoch": 1.0263562486272788, + "grad_norm": 2.34733510017395, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7260046601295471, + "num_tokens": 238382606.0, + "step": 9346 + }, + { + "epoch": 1.0264660663298923, + "grad_norm": 1.977770209312439, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7211956977844238, + "num_tokens": 238411013.0, + "step": 9347 + }, + { + "epoch": 1.026575884032506, + "grad_norm": 2.5474512577056885, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7259750366210938, + "num_tokens": 238431512.0, + "step": 9348 + }, + { + "epoch": 1.0266857017351196, + "grad_norm": 2.116751194000244, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7024011611938477, + "num_tokens": 238459964.0, + "step": 9349 + }, + { + "epoch": 1.0267955194377334, + "grad_norm": 2.334807872772217, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7033765912055969, + "num_tokens": 238485265.0, + "step": 9350 + }, + { + "epoch": 1.026905337140347, + "grad_norm": 2.254835844039917, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.740807294845581, + "num_tokens": 238509127.0, + "step": 9351 + }, + { + "epoch": 1.0270151548429607, + "grad_norm": 2.197711944580078, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7122282385826111, + "num_tokens": 238536840.0, + "step": 9352 + }, + { + "epoch": 1.0271249725455744, + "grad_norm": 2.244833469390869, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7404512763023376, + "num_tokens": 238560566.0, + "step": 9353 + }, + { + "epoch": 1.027234790248188, + "grad_norm": 2.037853717803955, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7243384122848511, + "num_tokens": 238588182.0, + "step": 9354 + }, + { + "epoch": 1.0273446079508017, + "grad_norm": 2.25815749168396, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7203053832054138, + "num_tokens": 238614141.0, + "step": 9355 + }, + { + "epoch": 1.0274544256534153, + "grad_norm": 2.4279873371124268, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7150365114212036, + "num_tokens": 238636810.0, + "step": 9356 + }, + { + "epoch": 1.027564243356029, + "grad_norm": 2.0189616680145264, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7080740928649902, + "num_tokens": 238666102.0, + "step": 9357 + }, + { + "epoch": 1.0276740610586426, + "grad_norm": 2.5982491970062256, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7063957452774048, + "num_tokens": 238687109.0, + "step": 9358 + }, + { + "epoch": 1.0277838787612563, + "grad_norm": 2.192112922668457, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7076606750488281, + "num_tokens": 238712196.0, + "step": 9359 + }, + { + "epoch": 1.02789369646387, + "grad_norm": 2.173560619354248, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7164230346679688, + "num_tokens": 238739877.0, + "step": 9360 + }, + { + "epoch": 1.0280035141664836, + "grad_norm": 2.050576686859131, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.7015185952186584, + "num_tokens": 238774171.0, + "step": 9361 + }, + { + "epoch": 1.0281133318690974, + "grad_norm": 2.119208812713623, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7273037433624268, + "num_tokens": 238801822.0, + "step": 9362 + }, + { + "epoch": 1.028223149571711, + "grad_norm": 2.1515238285064697, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6881616115570068, + "num_tokens": 238830120.0, + "step": 9363 + }, + { + "epoch": 1.0283329672743247, + "grad_norm": 2.315152883529663, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7398067116737366, + "num_tokens": 238853090.0, + "step": 9364 + }, + { + "epoch": 1.0284427849769382, + "grad_norm": 2.102666139602661, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.70641028881073, + "num_tokens": 238884083.0, + "step": 9365 + }, + { + "epoch": 1.028552602679552, + "grad_norm": 2.6253931522369385, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7051024436950684, + "num_tokens": 238905236.0, + "step": 9366 + }, + { + "epoch": 1.0286624203821657, + "grad_norm": 2.5192763805389404, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7367755174636841, + "num_tokens": 238925543.0, + "step": 9367 + }, + { + "epoch": 1.0287722380847792, + "grad_norm": 2.426865339279175, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7378183007240295, + "num_tokens": 238950450.0, + "step": 9368 + }, + { + "epoch": 1.028882055787393, + "grad_norm": 2.2550363540649414, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7461153268814087, + "num_tokens": 238973566.0, + "step": 9369 + }, + { + "epoch": 1.0289918734900065, + "grad_norm": 2.334730625152588, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7291689515113831, + "num_tokens": 239000480.0, + "step": 9370 + }, + { + "epoch": 1.0291016911926203, + "grad_norm": 2.1808078289031982, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.685675859451294, + "num_tokens": 239031365.0, + "step": 9371 + }, + { + "epoch": 1.0292115088952338, + "grad_norm": 2.1358602046966553, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7186067700386047, + "num_tokens": 239060838.0, + "step": 9372 + }, + { + "epoch": 1.0293213265978476, + "grad_norm": 2.5858240127563477, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7323640584945679, + "num_tokens": 239082115.0, + "step": 9373 + }, + { + "epoch": 1.0294311443004613, + "grad_norm": 2.2464888095855713, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7197747230529785, + "num_tokens": 239108123.0, + "step": 9374 + }, + { + "epoch": 1.0295409620030749, + "grad_norm": 2.103691577911377, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.7053638696670532, + "num_tokens": 239137722.0, + "step": 9375 + }, + { + "epoch": 1.0296507797056886, + "grad_norm": 2.1149559020996094, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7069869041442871, + "num_tokens": 239166013.0, + "step": 9376 + }, + { + "epoch": 1.0297605974083022, + "grad_norm": 2.5718038082122803, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7152625322341919, + "num_tokens": 239189497.0, + "step": 9377 + }, + { + "epoch": 1.029870415110916, + "grad_norm": 2.1775929927825928, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7000902891159058, + "num_tokens": 239218612.0, + "step": 9378 + }, + { + "epoch": 1.0299802328135295, + "grad_norm": 2.5193588733673096, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7313942313194275, + "num_tokens": 239239038.0, + "step": 9379 + }, + { + "epoch": 1.0300900505161432, + "grad_norm": 2.063819408416748, + "learning_rate": 1e-06, + "loss": 1.029, + "mean_token_accuracy": 0.6961705684661865, + "num_tokens": 239270704.0, + "step": 9380 + }, + { + "epoch": 1.030199868218757, + "grad_norm": 2.0154988765716553, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7033956050872803, + "num_tokens": 239298087.0, + "step": 9381 + }, + { + "epoch": 1.0303096859213705, + "grad_norm": 2.3418989181518555, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7146879434585571, + "num_tokens": 239321268.0, + "step": 9382 + }, + { + "epoch": 1.0304195036239843, + "grad_norm": 2.0782620906829834, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.6930112838745117, + "num_tokens": 239351206.0, + "step": 9383 + }, + { + "epoch": 1.0305293213265978, + "grad_norm": 2.1551339626312256, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7162142992019653, + "num_tokens": 239378317.0, + "step": 9384 + }, + { + "epoch": 1.0306391390292116, + "grad_norm": 2.2888174057006836, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7116284370422363, + "num_tokens": 239403160.0, + "step": 9385 + }, + { + "epoch": 1.030748956731825, + "grad_norm": 2.2568776607513428, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7091129422187805, + "num_tokens": 239428324.0, + "step": 9386 + }, + { + "epoch": 1.0308587744344389, + "grad_norm": 2.2840189933776855, + "learning_rate": 1e-06, + "loss": 1.0559, + "mean_token_accuracy": 0.6857742071151733, + "num_tokens": 239454687.0, + "step": 9387 + }, + { + "epoch": 1.0309685921370524, + "grad_norm": 2.3402864933013916, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7118828296661377, + "num_tokens": 239478340.0, + "step": 9388 + }, + { + "epoch": 1.0310784098396661, + "grad_norm": 2.3632137775421143, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7102558612823486, + "num_tokens": 239503553.0, + "step": 9389 + }, + { + "epoch": 1.03118822754228, + "grad_norm": 2.1735358238220215, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7110105156898499, + "num_tokens": 239532630.0, + "step": 9390 + }, + { + "epoch": 1.0312980452448934, + "grad_norm": 2.204660177230835, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7087153196334839, + "num_tokens": 239558898.0, + "step": 9391 + }, + { + "epoch": 1.0314078629475072, + "grad_norm": 1.9165456295013428, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7203863859176636, + "num_tokens": 239592561.0, + "step": 9392 + }, + { + "epoch": 1.0315176806501207, + "grad_norm": 2.3007519245147705, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7320221662521362, + "num_tokens": 239614087.0, + "step": 9393 + }, + { + "epoch": 1.0316274983527345, + "grad_norm": 2.4183945655822754, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7117304801940918, + "num_tokens": 239636310.0, + "step": 9394 + }, + { + "epoch": 1.031737316055348, + "grad_norm": 2.551823139190674, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7239494323730469, + "num_tokens": 239658508.0, + "step": 9395 + }, + { + "epoch": 1.0318471337579618, + "grad_norm": 2.6614208221435547, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7451675534248352, + "num_tokens": 239677900.0, + "step": 9396 + }, + { + "epoch": 1.0319569514605755, + "grad_norm": 2.1946980953216553, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7061492204666138, + "num_tokens": 239706028.0, + "step": 9397 + }, + { + "epoch": 1.032066769163189, + "grad_norm": 2.4149980545043945, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7196860313415527, + "num_tokens": 239728779.0, + "step": 9398 + }, + { + "epoch": 1.0321765868658028, + "grad_norm": 2.4126784801483154, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6940771341323853, + "num_tokens": 239754166.0, + "step": 9399 + }, + { + "epoch": 1.0322864045684164, + "grad_norm": 2.2563438415527344, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.728998064994812, + "num_tokens": 239780547.0, + "step": 9400 + }, + { + "epoch": 1.0323962222710301, + "grad_norm": 2.533848524093628, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7176308631896973, + "num_tokens": 239803774.0, + "step": 9401 + }, + { + "epoch": 1.0325060399736437, + "grad_norm": 2.466048002243042, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7194789052009583, + "num_tokens": 239828032.0, + "step": 9402 + }, + { + "epoch": 1.0326158576762574, + "grad_norm": 2.4361469745635986, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7241146564483643, + "num_tokens": 239853074.0, + "step": 9403 + }, + { + "epoch": 1.0327256753788712, + "grad_norm": 2.0332443714141846, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7081855535507202, + "num_tokens": 239882665.0, + "step": 9404 + }, + { + "epoch": 1.0328354930814847, + "grad_norm": 2.300447702407837, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7314127087593079, + "num_tokens": 239908221.0, + "step": 9405 + }, + { + "epoch": 1.0329453107840985, + "grad_norm": 2.3168323040008545, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.712338924407959, + "num_tokens": 239932003.0, + "step": 9406 + }, + { + "epoch": 1.033055128486712, + "grad_norm": 2.1873605251312256, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7097392082214355, + "num_tokens": 239957055.0, + "step": 9407 + }, + { + "epoch": 1.0331649461893258, + "grad_norm": 1.9464812278747559, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7177214026451111, + "num_tokens": 239992181.0, + "step": 9408 + }, + { + "epoch": 1.0332747638919393, + "grad_norm": 2.304854154586792, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7119187116622925, + "num_tokens": 240016889.0, + "step": 9409 + }, + { + "epoch": 1.033384581594553, + "grad_norm": 2.217590570449829, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7070941925048828, + "num_tokens": 240043271.0, + "step": 9410 + }, + { + "epoch": 1.0334943992971668, + "grad_norm": 2.338113784790039, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7225341796875, + "num_tokens": 240067131.0, + "step": 9411 + }, + { + "epoch": 1.0336042169997803, + "grad_norm": 2.0527729988098145, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7147246599197388, + "num_tokens": 240095533.0, + "step": 9412 + }, + { + "epoch": 1.033714034702394, + "grad_norm": 2.304473876953125, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7025028467178345, + "num_tokens": 240120640.0, + "step": 9413 + }, + { + "epoch": 1.0338238524050076, + "grad_norm": 2.3600025177001953, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.716795027256012, + "num_tokens": 240144596.0, + "step": 9414 + }, + { + "epoch": 1.0339336701076214, + "grad_norm": 2.60955810546875, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7201356887817383, + "num_tokens": 240163359.0, + "step": 9415 + }, + { + "epoch": 1.034043487810235, + "grad_norm": 2.3350930213928223, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7056845426559448, + "num_tokens": 240187000.0, + "step": 9416 + }, + { + "epoch": 1.0341533055128487, + "grad_norm": 2.5093801021575928, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7415117621421814, + "num_tokens": 240206087.0, + "step": 9417 + }, + { + "epoch": 1.0342631232154624, + "grad_norm": 2.3344922065734863, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7182466983795166, + "num_tokens": 240229469.0, + "step": 9418 + }, + { + "epoch": 1.034372940918076, + "grad_norm": 2.449012041091919, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7302564978599548, + "num_tokens": 240253383.0, + "step": 9419 + }, + { + "epoch": 1.0344827586206897, + "grad_norm": 2.0623793601989746, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.730047881603241, + "num_tokens": 240281001.0, + "step": 9420 + }, + { + "epoch": 1.0345925763233033, + "grad_norm": 2.006552219390869, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.6988000869750977, + "num_tokens": 240310601.0, + "step": 9421 + }, + { + "epoch": 1.034702394025917, + "grad_norm": 2.469470262527466, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7211836576461792, + "num_tokens": 240332044.0, + "step": 9422 + }, + { + "epoch": 1.0348122117285306, + "grad_norm": 2.4582009315490723, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7119963765144348, + "num_tokens": 240354357.0, + "step": 9423 + }, + { + "epoch": 1.0349220294311443, + "grad_norm": 2.566086769104004, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7407094240188599, + "num_tokens": 240374096.0, + "step": 9424 + }, + { + "epoch": 1.035031847133758, + "grad_norm": 2.129836082458496, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.6991350650787354, + "num_tokens": 240402834.0, + "step": 9425 + }, + { + "epoch": 1.0351416648363716, + "grad_norm": 2.099896192550659, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7267552614212036, + "num_tokens": 240428896.0, + "step": 9426 + }, + { + "epoch": 1.0352514825389854, + "grad_norm": 2.018817663192749, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6991844773292542, + "num_tokens": 240457983.0, + "step": 9427 + }, + { + "epoch": 1.035361300241599, + "grad_norm": 2.465254068374634, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7178328633308411, + "num_tokens": 240480077.0, + "step": 9428 + }, + { + "epoch": 1.0354711179442126, + "grad_norm": 1.9651554822921753, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7020771503448486, + "num_tokens": 240510485.0, + "step": 9429 + }, + { + "epoch": 1.0355809356468262, + "grad_norm": 2.288175582885742, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.725845456123352, + "num_tokens": 240534474.0, + "step": 9430 + }, + { + "epoch": 1.03569075334944, + "grad_norm": 2.1963443756103516, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7088205814361572, + "num_tokens": 240561893.0, + "step": 9431 + }, + { + "epoch": 1.0358005710520537, + "grad_norm": 2.0841052532196045, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.6997470259666443, + "num_tokens": 240589595.0, + "step": 9432 + }, + { + "epoch": 1.0359103887546672, + "grad_norm": 2.1287734508514404, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6951475739479065, + "num_tokens": 240617604.0, + "step": 9433 + }, + { + "epoch": 1.036020206457281, + "grad_norm": 2.163655996322632, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7105269432067871, + "num_tokens": 240645404.0, + "step": 9434 + }, + { + "epoch": 1.0361300241598945, + "grad_norm": 2.0527637004852295, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7093360424041748, + "num_tokens": 240679003.0, + "step": 9435 + }, + { + "epoch": 1.0362398418625083, + "grad_norm": 2.390273332595825, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7232851982116699, + "num_tokens": 240702212.0, + "step": 9436 + }, + { + "epoch": 1.0363496595651218, + "grad_norm": 2.214643716812134, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7252010107040405, + "num_tokens": 240729202.0, + "step": 9437 + }, + { + "epoch": 1.0364594772677356, + "grad_norm": 2.294156789779663, + "learning_rate": 1e-06, + "loss": 0.7862, + "mean_token_accuracy": 0.7586369514465332, + "num_tokens": 240750871.0, + "step": 9438 + }, + { + "epoch": 1.036569294970349, + "grad_norm": 2.127349853515625, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.715912938117981, + "num_tokens": 240779277.0, + "step": 9439 + }, + { + "epoch": 1.0366791126729629, + "grad_norm": 2.368921995162964, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7231559753417969, + "num_tokens": 240801178.0, + "step": 9440 + }, + { + "epoch": 1.0367889303755766, + "grad_norm": 2.1432251930236816, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7132495641708374, + "num_tokens": 240829288.0, + "step": 9441 + }, + { + "epoch": 1.0368987480781902, + "grad_norm": 2.1674158573150635, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7398031949996948, + "num_tokens": 240853478.0, + "step": 9442 + }, + { + "epoch": 1.037008565780804, + "grad_norm": 2.1027162075042725, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7033170461654663, + "num_tokens": 240884894.0, + "step": 9443 + }, + { + "epoch": 1.0371183834834174, + "grad_norm": 1.8868534564971924, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7024976015090942, + "num_tokens": 240920042.0, + "step": 9444 + }, + { + "epoch": 1.0372282011860312, + "grad_norm": 2.1816859245300293, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7227271795272827, + "num_tokens": 240947770.0, + "step": 9445 + }, + { + "epoch": 1.037338018888645, + "grad_norm": 2.44010066986084, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7149755954742432, + "num_tokens": 240970292.0, + "step": 9446 + }, + { + "epoch": 1.0374478365912585, + "grad_norm": 2.529510259628296, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7324051856994629, + "num_tokens": 240991512.0, + "step": 9447 + }, + { + "epoch": 1.0375576542938723, + "grad_norm": 2.529851198196411, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7210376262664795, + "num_tokens": 241012081.0, + "step": 9448 + }, + { + "epoch": 1.0376674719964858, + "grad_norm": 2.3036344051361084, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7032110691070557, + "num_tokens": 241036335.0, + "step": 9449 + }, + { + "epoch": 1.0377772896990995, + "grad_norm": 2.231360673904419, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7301759719848633, + "num_tokens": 241060792.0, + "step": 9450 + }, + { + "epoch": 1.037887107401713, + "grad_norm": 2.329500675201416, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7125759124755859, + "num_tokens": 241083956.0, + "step": 9451 + }, + { + "epoch": 1.0379969251043268, + "grad_norm": 2.3109190464019775, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7161228060722351, + "num_tokens": 241107666.0, + "step": 9452 + }, + { + "epoch": 1.0381067428069404, + "grad_norm": 2.196817636489868, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7323122024536133, + "num_tokens": 241133328.0, + "step": 9453 + }, + { + "epoch": 1.0382165605095541, + "grad_norm": 2.300018072128296, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7162887454032898, + "num_tokens": 241159049.0, + "step": 9454 + }, + { + "epoch": 1.0383263782121679, + "grad_norm": 2.5545122623443604, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7272660136222839, + "num_tokens": 241178223.0, + "step": 9455 + }, + { + "epoch": 1.0384361959147814, + "grad_norm": 2.394224166870117, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.740470826625824, + "num_tokens": 241199795.0, + "step": 9456 + }, + { + "epoch": 1.0385460136173952, + "grad_norm": 2.412653923034668, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.706256091594696, + "num_tokens": 241220952.0, + "step": 9457 + }, + { + "epoch": 1.0386558313200087, + "grad_norm": 2.3164830207824707, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7236166000366211, + "num_tokens": 241244497.0, + "step": 9458 + }, + { + "epoch": 1.0387656490226225, + "grad_norm": 2.0627150535583496, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7197148203849792, + "num_tokens": 241275271.0, + "step": 9459 + }, + { + "epoch": 1.038875466725236, + "grad_norm": 2.5554347038269043, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7107342481613159, + "num_tokens": 241295280.0, + "step": 9460 + }, + { + "epoch": 1.0389852844278498, + "grad_norm": 2.44474196434021, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7441346645355225, + "num_tokens": 241316171.0, + "step": 9461 + }, + { + "epoch": 1.0390951021304635, + "grad_norm": 2.213815212249756, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7140371799468994, + "num_tokens": 241343549.0, + "step": 9462 + }, + { + "epoch": 1.039204919833077, + "grad_norm": 2.1964621543884277, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.711272120475769, + "num_tokens": 241368527.0, + "step": 9463 + }, + { + "epoch": 1.0393147375356908, + "grad_norm": 2.7032809257507324, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7302476763725281, + "num_tokens": 241389191.0, + "step": 9464 + }, + { + "epoch": 1.0394245552383043, + "grad_norm": 2.3422281742095947, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7236666679382324, + "num_tokens": 241414707.0, + "step": 9465 + }, + { + "epoch": 1.039534372940918, + "grad_norm": 2.1267216205596924, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7089756727218628, + "num_tokens": 241443799.0, + "step": 9466 + }, + { + "epoch": 1.0396441906435316, + "grad_norm": 2.309911012649536, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7255710959434509, + "num_tokens": 241468428.0, + "step": 9467 + }, + { + "epoch": 1.0397540083461454, + "grad_norm": 2.2595250606536865, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7294928431510925, + "num_tokens": 241495056.0, + "step": 9468 + }, + { + "epoch": 1.0398638260487592, + "grad_norm": 2.1107490062713623, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6980434656143188, + "num_tokens": 241523829.0, + "step": 9469 + }, + { + "epoch": 1.0399736437513727, + "grad_norm": 2.023491859436035, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7103131413459778, + "num_tokens": 241554599.0, + "step": 9470 + }, + { + "epoch": 1.0400834614539864, + "grad_norm": 2.749486207962036, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7275018692016602, + "num_tokens": 241573893.0, + "step": 9471 + }, + { + "epoch": 1.0401932791566, + "grad_norm": 2.076387643814087, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.703963041305542, + "num_tokens": 241603588.0, + "step": 9472 + }, + { + "epoch": 1.0403030968592137, + "grad_norm": 2.23348331451416, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7019871473312378, + "num_tokens": 241629222.0, + "step": 9473 + }, + { + "epoch": 1.0404129145618273, + "grad_norm": 2.34037709236145, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7345930933952332, + "num_tokens": 241651919.0, + "step": 9474 + }, + { + "epoch": 1.040522732264441, + "grad_norm": 2.3768582344055176, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7269679307937622, + "num_tokens": 241674930.0, + "step": 9475 + }, + { + "epoch": 1.0406325499670548, + "grad_norm": 2.5840139389038086, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7242719531059265, + "num_tokens": 241694642.0, + "step": 9476 + }, + { + "epoch": 1.0407423676696683, + "grad_norm": 2.0880138874053955, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7142702341079712, + "num_tokens": 241722099.0, + "step": 9477 + }, + { + "epoch": 1.040852185372282, + "grad_norm": 1.9154515266418457, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7172348499298096, + "num_tokens": 241755321.0, + "step": 9478 + }, + { + "epoch": 1.0409620030748956, + "grad_norm": 2.266183614730835, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7050870656967163, + "num_tokens": 241781522.0, + "step": 9479 + }, + { + "epoch": 1.0410718207775094, + "grad_norm": 2.23166561126709, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7050647735595703, + "num_tokens": 241807750.0, + "step": 9480 + }, + { + "epoch": 1.041181638480123, + "grad_norm": 2.0092411041259766, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7267524600028992, + "num_tokens": 241838368.0, + "step": 9481 + }, + { + "epoch": 1.0412914561827367, + "grad_norm": 2.071934223175049, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7201599478721619, + "num_tokens": 241867486.0, + "step": 9482 + }, + { + "epoch": 1.0414012738853504, + "grad_norm": 2.217751979827881, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7140445709228516, + "num_tokens": 241892979.0, + "step": 9483 + }, + { + "epoch": 1.041511091587964, + "grad_norm": 2.3035178184509277, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.6997658014297485, + "num_tokens": 241917327.0, + "step": 9484 + }, + { + "epoch": 1.0416209092905777, + "grad_norm": 1.970146656036377, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.715756356716156, + "num_tokens": 241948859.0, + "step": 9485 + }, + { + "epoch": 1.0417307269931912, + "grad_norm": 2.4652788639068604, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7159218192100525, + "num_tokens": 241971817.0, + "step": 9486 + }, + { + "epoch": 1.041840544695805, + "grad_norm": 2.3214504718780518, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7420542240142822, + "num_tokens": 241993703.0, + "step": 9487 + }, + { + "epoch": 1.0419503623984185, + "grad_norm": 1.9921549558639526, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.703913152217865, + "num_tokens": 242023998.0, + "step": 9488 + }, + { + "epoch": 1.0420601801010323, + "grad_norm": 2.119715690612793, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7261760234832764, + "num_tokens": 242051369.0, + "step": 9489 + }, + { + "epoch": 1.042169997803646, + "grad_norm": 2.30961537361145, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7147594690322876, + "num_tokens": 242074618.0, + "step": 9490 + }, + { + "epoch": 1.0422798155062596, + "grad_norm": 2.3019144535064697, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7205485701560974, + "num_tokens": 242100774.0, + "step": 9491 + }, + { + "epoch": 1.0423896332088733, + "grad_norm": 2.2831122875213623, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7215025424957275, + "num_tokens": 242124466.0, + "step": 9492 + }, + { + "epoch": 1.0424994509114869, + "grad_norm": 2.6294493675231934, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7082018852233887, + "num_tokens": 242144856.0, + "step": 9493 + }, + { + "epoch": 1.0426092686141006, + "grad_norm": 2.324093818664551, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.712012529373169, + "num_tokens": 242166916.0, + "step": 9494 + }, + { + "epoch": 1.0427190863167142, + "grad_norm": 1.8995016813278198, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.719801127910614, + "num_tokens": 242198701.0, + "step": 9495 + }, + { + "epoch": 1.042828904019328, + "grad_norm": 2.1168127059936523, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.732530415058136, + "num_tokens": 242225116.0, + "step": 9496 + }, + { + "epoch": 1.0429387217219417, + "grad_norm": 2.441338300704956, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7369760274887085, + "num_tokens": 242248148.0, + "step": 9497 + }, + { + "epoch": 1.0430485394245552, + "grad_norm": 2.436608076095581, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7369895577430725, + "num_tokens": 242269375.0, + "step": 9498 + }, + { + "epoch": 1.043158357127169, + "grad_norm": 2.094266891479492, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7137844562530518, + "num_tokens": 242299032.0, + "step": 9499 + }, + { + "epoch": 1.0432681748297825, + "grad_norm": 2.238018035888672, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7282408475875854, + "num_tokens": 242324528.0, + "step": 9500 + }, + { + "epoch": 1.0433779925323963, + "grad_norm": 2.256171703338623, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7046029567718506, + "num_tokens": 242348737.0, + "step": 9501 + }, + { + "epoch": 1.0434878102350098, + "grad_norm": 2.2817673683166504, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7261652946472168, + "num_tokens": 242372017.0, + "step": 9502 + }, + { + "epoch": 1.0435976279376236, + "grad_norm": 2.3711602687835693, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7034730315208435, + "num_tokens": 242396200.0, + "step": 9503 + }, + { + "epoch": 1.043707445640237, + "grad_norm": 2.2119946479797363, + "learning_rate": 1e-06, + "loss": 0.8287, + "mean_token_accuracy": 0.7342896461486816, + "num_tokens": 242420515.0, + "step": 9504 + }, + { + "epoch": 1.0438172633428509, + "grad_norm": 2.3570480346679688, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7209131121635437, + "num_tokens": 242443487.0, + "step": 9505 + }, + { + "epoch": 1.0439270810454646, + "grad_norm": 2.486119031906128, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7230987548828125, + "num_tokens": 242466280.0, + "step": 9506 + }, + { + "epoch": 1.0440368987480781, + "grad_norm": 2.240112543106079, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7199645638465881, + "num_tokens": 242491251.0, + "step": 9507 + }, + { + "epoch": 1.044146716450692, + "grad_norm": 2.4562814235687256, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7229032516479492, + "num_tokens": 242514622.0, + "step": 9508 + }, + { + "epoch": 1.0442565341533054, + "grad_norm": 1.9877021312713623, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7175002098083496, + "num_tokens": 242544742.0, + "step": 9509 + }, + { + "epoch": 1.0443663518559192, + "grad_norm": 2.3793463706970215, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7366938591003418, + "num_tokens": 242567530.0, + "step": 9510 + }, + { + "epoch": 1.0444761695585327, + "grad_norm": 2.2220091819763184, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7319585084915161, + "num_tokens": 242591177.0, + "step": 9511 + }, + { + "epoch": 1.0445859872611465, + "grad_norm": 2.6318609714508057, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7391562461853027, + "num_tokens": 242609929.0, + "step": 9512 + }, + { + "epoch": 1.0446958049637602, + "grad_norm": 2.231262445449829, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7283362150192261, + "num_tokens": 242635112.0, + "step": 9513 + }, + { + "epoch": 1.0448056226663738, + "grad_norm": 2.231362819671631, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.727266788482666, + "num_tokens": 242658622.0, + "step": 9514 + }, + { + "epoch": 1.0449154403689875, + "grad_norm": 2.1794848442077637, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.6939194202423096, + "num_tokens": 242685881.0, + "step": 9515 + }, + { + "epoch": 1.045025258071601, + "grad_norm": 2.174816608428955, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7372061014175415, + "num_tokens": 242710740.0, + "step": 9516 + }, + { + "epoch": 1.0451350757742148, + "grad_norm": 2.403264045715332, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7250460386276245, + "num_tokens": 242733989.0, + "step": 9517 + }, + { + "epoch": 1.0452448934768284, + "grad_norm": 2.333130359649658, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.6877571940422058, + "num_tokens": 242761259.0, + "step": 9518 + }, + { + "epoch": 1.0453547111794421, + "grad_norm": 2.402388334274292, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7027779817581177, + "num_tokens": 242785354.0, + "step": 9519 + }, + { + "epoch": 1.0454645288820559, + "grad_norm": 2.0662384033203125, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.704960823059082, + "num_tokens": 242815208.0, + "step": 9520 + }, + { + "epoch": 1.0455743465846694, + "grad_norm": 2.1630892753601074, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.701892614364624, + "num_tokens": 242844890.0, + "step": 9521 + }, + { + "epoch": 1.0456841642872832, + "grad_norm": 2.1723580360412598, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7119674682617188, + "num_tokens": 242871451.0, + "step": 9522 + }, + { + "epoch": 1.0457939819898967, + "grad_norm": 2.110090970993042, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6980588436126709, + "num_tokens": 242900656.0, + "step": 9523 + }, + { + "epoch": 1.0459037996925105, + "grad_norm": 1.9309496879577637, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.737281084060669, + "num_tokens": 242931329.0, + "step": 9524 + }, + { + "epoch": 1.046013617395124, + "grad_norm": 2.488161087036133, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7223258018493652, + "num_tokens": 242952589.0, + "step": 9525 + }, + { + "epoch": 1.0461234350977378, + "grad_norm": 2.1295602321624756, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7291102409362793, + "num_tokens": 242979853.0, + "step": 9526 + }, + { + "epoch": 1.0462332528003515, + "grad_norm": 2.311424732208252, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.725980281829834, + "num_tokens": 243004980.0, + "step": 9527 + }, + { + "epoch": 1.046343070502965, + "grad_norm": 2.234759569168091, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7148735523223877, + "num_tokens": 243029836.0, + "step": 9528 + }, + { + "epoch": 1.0464528882055788, + "grad_norm": 2.036445140838623, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7299410104751587, + "num_tokens": 243058362.0, + "step": 9529 + }, + { + "epoch": 1.0465627059081923, + "grad_norm": 2.1093618869781494, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7126709818840027, + "num_tokens": 243087100.0, + "step": 9530 + }, + { + "epoch": 1.046672523610806, + "grad_norm": 2.02070689201355, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7179201245307922, + "num_tokens": 243117447.0, + "step": 9531 + }, + { + "epoch": 1.0467823413134196, + "grad_norm": 2.5846011638641357, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7386173605918884, + "num_tokens": 243136385.0, + "step": 9532 + }, + { + "epoch": 1.0468921590160334, + "grad_norm": 2.4536828994750977, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7366243004798889, + "num_tokens": 243158946.0, + "step": 9533 + }, + { + "epoch": 1.0470019767186471, + "grad_norm": 2.214991807937622, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7448314428329468, + "num_tokens": 243183727.0, + "step": 9534 + }, + { + "epoch": 1.0471117944212607, + "grad_norm": 2.3744711875915527, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7289819717407227, + "num_tokens": 243206036.0, + "step": 9535 + }, + { + "epoch": 1.0472216121238744, + "grad_norm": 2.26102876663208, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.708609402179718, + "num_tokens": 243231091.0, + "step": 9536 + }, + { + "epoch": 1.047331429826488, + "grad_norm": 2.118889570236206, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7181196808815002, + "num_tokens": 243258098.0, + "step": 9537 + }, + { + "epoch": 1.0474412475291017, + "grad_norm": 2.245231866836548, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.720829963684082, + "num_tokens": 243285843.0, + "step": 9538 + }, + { + "epoch": 1.0475510652317153, + "grad_norm": 2.182455539703369, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7110998630523682, + "num_tokens": 243313321.0, + "step": 9539 + }, + { + "epoch": 1.047660882934329, + "grad_norm": 2.2342395782470703, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7273218631744385, + "num_tokens": 243340025.0, + "step": 9540 + }, + { + "epoch": 1.0477707006369428, + "grad_norm": 2.448223352432251, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7254180908203125, + "num_tokens": 243360764.0, + "step": 9541 + }, + { + "epoch": 1.0478805183395563, + "grad_norm": 2.7876546382904053, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7200044393539429, + "num_tokens": 243378423.0, + "step": 9542 + }, + { + "epoch": 1.04799033604217, + "grad_norm": 2.0593671798706055, + "learning_rate": 1e-06, + "loss": 1.0285, + "mean_token_accuracy": 0.6838662028312683, + "num_tokens": 243410585.0, + "step": 9543 + }, + { + "epoch": 1.0481001537447836, + "grad_norm": 2.2166647911071777, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6917392015457153, + "num_tokens": 243438501.0, + "step": 9544 + }, + { + "epoch": 1.0482099714473974, + "grad_norm": 2.37488055229187, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7448049187660217, + "num_tokens": 243459618.0, + "step": 9545 + }, + { + "epoch": 1.048319789150011, + "grad_norm": 2.262413501739502, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.6986158490180969, + "num_tokens": 243485208.0, + "step": 9546 + }, + { + "epoch": 1.0484296068526247, + "grad_norm": 2.161771774291992, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7030656933784485, + "num_tokens": 243513886.0, + "step": 9547 + }, + { + "epoch": 1.0485394245552384, + "grad_norm": 2.1988627910614014, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7054500579833984, + "num_tokens": 243540414.0, + "step": 9548 + }, + { + "epoch": 1.048649242257852, + "grad_norm": 2.094193696975708, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7378803491592407, + "num_tokens": 243567448.0, + "step": 9549 + }, + { + "epoch": 1.0487590599604657, + "grad_norm": 2.2791945934295654, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.712462306022644, + "num_tokens": 243590531.0, + "step": 9550 + }, + { + "epoch": 1.0488688776630792, + "grad_norm": 2.3496854305267334, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7145828008651733, + "num_tokens": 243613042.0, + "step": 9551 + }, + { + "epoch": 1.048978695365693, + "grad_norm": 2.2852261066436768, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7094857692718506, + "num_tokens": 243636761.0, + "step": 9552 + }, + { + "epoch": 1.0490885130683065, + "grad_norm": 2.532045602798462, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.73809814453125, + "num_tokens": 243657110.0, + "step": 9553 + }, + { + "epoch": 1.0491983307709203, + "grad_norm": 2.3198580741882324, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.744699239730835, + "num_tokens": 243680586.0, + "step": 9554 + }, + { + "epoch": 1.0493081484735338, + "grad_norm": 2.185554265975952, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7013794183731079, + "num_tokens": 243708349.0, + "step": 9555 + }, + { + "epoch": 1.0494179661761476, + "grad_norm": 2.3303422927856445, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.699310302734375, + "num_tokens": 243731882.0, + "step": 9556 + }, + { + "epoch": 1.0495277838787613, + "grad_norm": 1.9419280290603638, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7148748636245728, + "num_tokens": 243764027.0, + "step": 9557 + }, + { + "epoch": 1.0496376015813749, + "grad_norm": 2.1379141807556152, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7003856301307678, + "num_tokens": 243791856.0, + "step": 9558 + }, + { + "epoch": 1.0497474192839886, + "grad_norm": 2.2390646934509277, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7065913677215576, + "num_tokens": 243818167.0, + "step": 9559 + }, + { + "epoch": 1.0498572369866022, + "grad_norm": 2.4828505516052246, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7177678346633911, + "num_tokens": 243838920.0, + "step": 9560 + }, + { + "epoch": 1.049967054689216, + "grad_norm": 2.4234023094177246, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7219461798667908, + "num_tokens": 243862088.0, + "step": 9561 + }, + { + "epoch": 1.0500768723918297, + "grad_norm": 1.8540046215057373, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7103039026260376, + "num_tokens": 243896475.0, + "step": 9562 + }, + { + "epoch": 1.0501866900944432, + "grad_norm": 2.437634229660034, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7294771671295166, + "num_tokens": 243917930.0, + "step": 9563 + }, + { + "epoch": 1.050296507797057, + "grad_norm": 2.567797899246216, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7245907783508301, + "num_tokens": 243938633.0, + "step": 9564 + }, + { + "epoch": 1.0504063254996705, + "grad_norm": 2.0193896293640137, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7185553312301636, + "num_tokens": 243971058.0, + "step": 9565 + }, + { + "epoch": 1.0505161432022843, + "grad_norm": 2.0080833435058594, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7011229395866394, + "num_tokens": 244003593.0, + "step": 9566 + }, + { + "epoch": 1.0506259609048978, + "grad_norm": 2.287810802459717, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7147502899169922, + "num_tokens": 244027459.0, + "step": 9567 + }, + { + "epoch": 1.0507357786075116, + "grad_norm": 2.477682590484619, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7355786561965942, + "num_tokens": 244049486.0, + "step": 9568 + }, + { + "epoch": 1.050845596310125, + "grad_norm": 2.3759632110595703, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7091881632804871, + "num_tokens": 244072615.0, + "step": 9569 + }, + { + "epoch": 1.0509554140127388, + "grad_norm": 2.570314645767212, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7462732195854187, + "num_tokens": 244091521.0, + "step": 9570 + }, + { + "epoch": 1.0510652317153526, + "grad_norm": 2.3878729343414307, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.6945253014564514, + "num_tokens": 244116983.0, + "step": 9571 + }, + { + "epoch": 1.0511750494179661, + "grad_norm": 2.349729299545288, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.716951847076416, + "num_tokens": 244140154.0, + "step": 9572 + }, + { + "epoch": 1.05128486712058, + "grad_norm": 2.131988763809204, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.7572078704833984, + "num_tokens": 244166834.0, + "step": 9573 + }, + { + "epoch": 1.0513946848231934, + "grad_norm": 2.136831760406494, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7149953246116638, + "num_tokens": 244192921.0, + "step": 9574 + }, + { + "epoch": 1.0515045025258072, + "grad_norm": 2.43222713470459, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7261653542518616, + "num_tokens": 244215666.0, + "step": 9575 + }, + { + "epoch": 1.0516143202284207, + "grad_norm": 2.1597371101379395, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7358399629592896, + "num_tokens": 244241718.0, + "step": 9576 + }, + { + "epoch": 1.0517241379310345, + "grad_norm": 2.2144458293914795, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7139828205108643, + "num_tokens": 244267556.0, + "step": 9577 + }, + { + "epoch": 1.0518339556336482, + "grad_norm": 2.2650554180145264, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.71955406665802, + "num_tokens": 244292752.0, + "step": 9578 + }, + { + "epoch": 1.0519437733362618, + "grad_norm": 2.535165548324585, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7389645576477051, + "num_tokens": 244311614.0, + "step": 9579 + }, + { + "epoch": 1.0520535910388755, + "grad_norm": 2.249227285385132, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7165577411651611, + "num_tokens": 244336644.0, + "step": 9580 + }, + { + "epoch": 1.052163408741489, + "grad_norm": 2.391777753829956, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7293485403060913, + "num_tokens": 244358591.0, + "step": 9581 + }, + { + "epoch": 1.0522732264441028, + "grad_norm": 2.196624279022217, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7215083241462708, + "num_tokens": 244385291.0, + "step": 9582 + }, + { + "epoch": 1.0523830441467164, + "grad_norm": 2.1760621070861816, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.6923054456710815, + "num_tokens": 244412610.0, + "step": 9583 + }, + { + "epoch": 1.05249286184933, + "grad_norm": 2.0804636478424072, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.6962069869041443, + "num_tokens": 244442709.0, + "step": 9584 + }, + { + "epoch": 1.0526026795519439, + "grad_norm": 2.155271530151367, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7225348949432373, + "num_tokens": 244469013.0, + "step": 9585 + }, + { + "epoch": 1.0527124972545574, + "grad_norm": 2.54310941696167, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7092380523681641, + "num_tokens": 244490506.0, + "step": 9586 + }, + { + "epoch": 1.0528223149571712, + "grad_norm": 2.4529669284820557, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7248778343200684, + "num_tokens": 244513414.0, + "step": 9587 + }, + { + "epoch": 1.0529321326597847, + "grad_norm": 1.9932080507278442, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7198973298072815, + "num_tokens": 244543557.0, + "step": 9588 + }, + { + "epoch": 1.0530419503623984, + "grad_norm": 2.5637853145599365, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7415510416030884, + "num_tokens": 244564374.0, + "step": 9589 + }, + { + "epoch": 1.053151768065012, + "grad_norm": 1.9908620119094849, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6919541358947754, + "num_tokens": 244597341.0, + "step": 9590 + }, + { + "epoch": 1.0532615857676257, + "grad_norm": 2.471345901489258, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.723469078540802, + "num_tokens": 244620591.0, + "step": 9591 + }, + { + "epoch": 1.0533714034702395, + "grad_norm": 2.3355212211608887, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.708011269569397, + "num_tokens": 244644550.0, + "step": 9592 + }, + { + "epoch": 1.053481221172853, + "grad_norm": 2.1188459396362305, + "learning_rate": 1e-06, + "loss": 0.7803, + "mean_token_accuracy": 0.7527103424072266, + "num_tokens": 244670587.0, + "step": 9593 + }, + { + "epoch": 1.0535910388754668, + "grad_norm": 2.149418830871582, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7011895179748535, + "num_tokens": 244698998.0, + "step": 9594 + }, + { + "epoch": 1.0537008565780803, + "grad_norm": 1.9635753631591797, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6843552589416504, + "num_tokens": 244732816.0, + "step": 9595 + }, + { + "epoch": 1.053810674280694, + "grad_norm": 2.0692214965820312, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7320632934570312, + "num_tokens": 244760968.0, + "step": 9596 + }, + { + "epoch": 1.0539204919833076, + "grad_norm": 2.206937074661255, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.729411244392395, + "num_tokens": 244785655.0, + "step": 9597 + }, + { + "epoch": 1.0540303096859214, + "grad_norm": 2.2293803691864014, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.6955059766769409, + "num_tokens": 244811409.0, + "step": 9598 + }, + { + "epoch": 1.0541401273885351, + "grad_norm": 1.9773279428482056, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7051599621772766, + "num_tokens": 244842221.0, + "step": 9599 + }, + { + "epoch": 1.0542499450911487, + "grad_norm": 1.757712960243225, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6861488223075867, + "num_tokens": 244879784.0, + "step": 9600 + }, + { + "epoch": 1.0543597627937624, + "grad_norm": 2.486077308654785, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7174938917160034, + "num_tokens": 244901325.0, + "step": 9601 + }, + { + "epoch": 1.054469580496376, + "grad_norm": 2.3895039558410645, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7239199280738831, + "num_tokens": 244924832.0, + "step": 9602 + }, + { + "epoch": 1.0545793981989897, + "grad_norm": 2.454037666320801, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7182301878929138, + "num_tokens": 244947758.0, + "step": 9603 + }, + { + "epoch": 1.0546892159016032, + "grad_norm": 2.2677621841430664, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7263392210006714, + "num_tokens": 244972012.0, + "step": 9604 + }, + { + "epoch": 1.054799033604217, + "grad_norm": 2.576251983642578, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7103375792503357, + "num_tokens": 244993831.0, + "step": 9605 + }, + { + "epoch": 1.0549088513068308, + "grad_norm": 2.4599668979644775, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7288013696670532, + "num_tokens": 245015608.0, + "step": 9606 + }, + { + "epoch": 1.0550186690094443, + "grad_norm": 2.3053505420684814, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7339372634887695, + "num_tokens": 245040681.0, + "step": 9607 + }, + { + "epoch": 1.055128486712058, + "grad_norm": 2.13116192817688, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.6990396976470947, + "num_tokens": 245070177.0, + "step": 9608 + }, + { + "epoch": 1.0552383044146716, + "grad_norm": 2.092585802078247, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7391006946563721, + "num_tokens": 245099134.0, + "step": 9609 + }, + { + "epoch": 1.0553481221172853, + "grad_norm": 2.4294753074645996, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.734256386756897, + "num_tokens": 245120137.0, + "step": 9610 + }, + { + "epoch": 1.0554579398198989, + "grad_norm": 2.2662415504455566, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7179655432701111, + "num_tokens": 245146120.0, + "step": 9611 + }, + { + "epoch": 1.0555677575225126, + "grad_norm": 2.3593950271606445, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7147129774093628, + "num_tokens": 245169100.0, + "step": 9612 + }, + { + "epoch": 1.0556775752251264, + "grad_norm": 2.1970789432525635, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7364453077316284, + "num_tokens": 245194103.0, + "step": 9613 + }, + { + "epoch": 1.05578739292774, + "grad_norm": 2.220839738845825, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.6951338052749634, + "num_tokens": 245221895.0, + "step": 9614 + }, + { + "epoch": 1.0558972106303537, + "grad_norm": 2.619377851486206, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7108601331710815, + "num_tokens": 245240217.0, + "step": 9615 + }, + { + "epoch": 1.0560070283329672, + "grad_norm": 2.1410796642303467, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7312694191932678, + "num_tokens": 245269392.0, + "step": 9616 + }, + { + "epoch": 1.056116846035581, + "grad_norm": 2.1797680854797363, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7133896350860596, + "num_tokens": 245297790.0, + "step": 9617 + }, + { + "epoch": 1.0562266637381945, + "grad_norm": 2.1510226726531982, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7072024345397949, + "num_tokens": 245327228.0, + "step": 9618 + }, + { + "epoch": 1.0563364814408083, + "grad_norm": 2.262993812561035, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7193679213523865, + "num_tokens": 245350199.0, + "step": 9619 + }, + { + "epoch": 1.0564462991434218, + "grad_norm": 1.9933476448059082, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6915702819824219, + "num_tokens": 245381374.0, + "step": 9620 + }, + { + "epoch": 1.0565561168460356, + "grad_norm": 2.273996114730835, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.71027672290802, + "num_tokens": 245406574.0, + "step": 9621 + }, + { + "epoch": 1.0566659345486493, + "grad_norm": 2.141695261001587, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7104822397232056, + "num_tokens": 245435372.0, + "step": 9622 + }, + { + "epoch": 1.0567757522512629, + "grad_norm": 2.2951979637145996, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7322289943695068, + "num_tokens": 245458261.0, + "step": 9623 + }, + { + "epoch": 1.0568855699538766, + "grad_norm": 2.4094254970550537, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7172653079032898, + "num_tokens": 245481449.0, + "step": 9624 + }, + { + "epoch": 1.0569953876564901, + "grad_norm": 1.9373137950897217, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.7001880407333374, + "num_tokens": 245515639.0, + "step": 9625 + }, + { + "epoch": 1.057105205359104, + "grad_norm": 2.3223397731781006, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.720425009727478, + "num_tokens": 245539160.0, + "step": 9626 + }, + { + "epoch": 1.0572150230617177, + "grad_norm": 2.2478485107421875, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7125916481018066, + "num_tokens": 245565424.0, + "step": 9627 + }, + { + "epoch": 1.0573248407643312, + "grad_norm": 2.4408817291259766, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7388929128646851, + "num_tokens": 245587989.0, + "step": 9628 + }, + { + "epoch": 1.057434658466945, + "grad_norm": 2.387317419052124, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7304961681365967, + "num_tokens": 245610660.0, + "step": 9629 + }, + { + "epoch": 1.0575444761695585, + "grad_norm": 2.119335889816284, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7397575378417969, + "num_tokens": 245637269.0, + "step": 9630 + }, + { + "epoch": 1.0576542938721722, + "grad_norm": 2.3657212257385254, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7430193424224854, + "num_tokens": 245657861.0, + "step": 9631 + }, + { + "epoch": 1.0577641115747858, + "grad_norm": 2.2214181423187256, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7119045257568359, + "num_tokens": 245683391.0, + "step": 9632 + }, + { + "epoch": 1.0578739292773995, + "grad_norm": 2.4248383045196533, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7082582712173462, + "num_tokens": 245707021.0, + "step": 9633 + }, + { + "epoch": 1.057983746980013, + "grad_norm": 2.163052797317505, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.70225989818573, + "num_tokens": 245735229.0, + "step": 9634 + }, + { + "epoch": 1.0580935646826268, + "grad_norm": 2.388828992843628, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.751196563243866, + "num_tokens": 245757590.0, + "step": 9635 + }, + { + "epoch": 1.0582033823852406, + "grad_norm": 2.314314126968384, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.6985701322555542, + "num_tokens": 245782031.0, + "step": 9636 + }, + { + "epoch": 1.0583132000878541, + "grad_norm": 2.1587345600128174, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7117538452148438, + "num_tokens": 245808605.0, + "step": 9637 + }, + { + "epoch": 1.0584230177904679, + "grad_norm": 2.2706494331359863, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7238981127738953, + "num_tokens": 245831878.0, + "step": 9638 + }, + { + "epoch": 1.0585328354930814, + "grad_norm": 2.2530710697174072, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7112925052642822, + "num_tokens": 245858246.0, + "step": 9639 + }, + { + "epoch": 1.0586426531956952, + "grad_norm": 2.0283634662628174, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7061035633087158, + "num_tokens": 245889466.0, + "step": 9640 + }, + { + "epoch": 1.0587524708983087, + "grad_norm": 2.412566661834717, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7386505603790283, + "num_tokens": 245911432.0, + "step": 9641 + }, + { + "epoch": 1.0588622886009225, + "grad_norm": 2.2549681663513184, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7253408432006836, + "num_tokens": 245936966.0, + "step": 9642 + }, + { + "epoch": 1.0589721063035362, + "grad_norm": 2.4599382877349854, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7443523406982422, + "num_tokens": 245959671.0, + "step": 9643 + }, + { + "epoch": 1.0590819240061498, + "grad_norm": 2.5887465476989746, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7349379658699036, + "num_tokens": 245980086.0, + "step": 9644 + }, + { + "epoch": 1.0591917417087635, + "grad_norm": 2.0962681770324707, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7353988289833069, + "num_tokens": 246008239.0, + "step": 9645 + }, + { + "epoch": 1.059301559411377, + "grad_norm": 2.2645299434661865, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7091491222381592, + "num_tokens": 246035547.0, + "step": 9646 + }, + { + "epoch": 1.0594113771139908, + "grad_norm": 2.3246517181396484, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7216664552688599, + "num_tokens": 246060026.0, + "step": 9647 + }, + { + "epoch": 1.0595211948166043, + "grad_norm": 2.215467691421509, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7221200466156006, + "num_tokens": 246085342.0, + "step": 9648 + }, + { + "epoch": 1.059631012519218, + "grad_norm": 2.1812002658843994, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6941354870796204, + "num_tokens": 246113712.0, + "step": 9649 + }, + { + "epoch": 1.0597408302218319, + "grad_norm": 2.5728962421417236, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7102833986282349, + "num_tokens": 246134108.0, + "step": 9650 + }, + { + "epoch": 1.0598506479244454, + "grad_norm": 2.0073940753936768, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7003669142723083, + "num_tokens": 246167685.0, + "step": 9651 + }, + { + "epoch": 1.0599604656270591, + "grad_norm": 2.289642810821533, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7203105688095093, + "num_tokens": 246193596.0, + "step": 9652 + }, + { + "epoch": 1.0600702833296727, + "grad_norm": 1.8397164344787598, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7045879364013672, + "num_tokens": 246228966.0, + "step": 9653 + }, + { + "epoch": 1.0601801010322864, + "grad_norm": 2.4105896949768066, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7144004106521606, + "num_tokens": 246252381.0, + "step": 9654 + }, + { + "epoch": 1.0602899187349, + "grad_norm": 2.4182662963867188, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7112730145454407, + "num_tokens": 246275464.0, + "step": 9655 + }, + { + "epoch": 1.0603997364375137, + "grad_norm": 2.4613547325134277, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7134720087051392, + "num_tokens": 246299164.0, + "step": 9656 + }, + { + "epoch": 1.0605095541401275, + "grad_norm": 2.149412155151367, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7200455069541931, + "num_tokens": 246326616.0, + "step": 9657 + }, + { + "epoch": 1.060619371842741, + "grad_norm": 2.1087570190429688, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7103872299194336, + "num_tokens": 246353619.0, + "step": 9658 + }, + { + "epoch": 1.0607291895453548, + "grad_norm": 2.795912981033325, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7239475250244141, + "num_tokens": 246372847.0, + "step": 9659 + }, + { + "epoch": 1.0608390072479683, + "grad_norm": 2.2025318145751953, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7041900157928467, + "num_tokens": 246401419.0, + "step": 9660 + }, + { + "epoch": 1.060948824950582, + "grad_norm": 2.464329719543457, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7297806143760681, + "num_tokens": 246422468.0, + "step": 9661 + }, + { + "epoch": 1.0610586426531956, + "grad_norm": 2.0541465282440186, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7002127170562744, + "num_tokens": 246452375.0, + "step": 9662 + }, + { + "epoch": 1.0611684603558094, + "grad_norm": 2.1822800636291504, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.731532871723175, + "num_tokens": 246476577.0, + "step": 9663 + }, + { + "epoch": 1.0612782780584231, + "grad_norm": 2.2513248920440674, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7291099429130554, + "num_tokens": 246500970.0, + "step": 9664 + }, + { + "epoch": 1.0613880957610367, + "grad_norm": 1.9979513883590698, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7104156017303467, + "num_tokens": 246532842.0, + "step": 9665 + }, + { + "epoch": 1.0614979134636504, + "grad_norm": 2.485163927078247, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7125265002250671, + "num_tokens": 246555968.0, + "step": 9666 + }, + { + "epoch": 1.061607731166264, + "grad_norm": 2.129490613937378, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7328747510910034, + "num_tokens": 246583246.0, + "step": 9667 + }, + { + "epoch": 1.0617175488688777, + "grad_norm": 2.2354681491851807, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7103599309921265, + "num_tokens": 246610130.0, + "step": 9668 + }, + { + "epoch": 1.0618273665714912, + "grad_norm": 2.292943239212036, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.723655104637146, + "num_tokens": 246632799.0, + "step": 9669 + }, + { + "epoch": 1.061937184274105, + "grad_norm": 2.33591628074646, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.721795916557312, + "num_tokens": 246655782.0, + "step": 9670 + }, + { + "epoch": 1.0620470019767188, + "grad_norm": 2.3060927391052246, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7400251030921936, + "num_tokens": 246678360.0, + "step": 9671 + }, + { + "epoch": 1.0621568196793323, + "grad_norm": 2.2363922595977783, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7303199172019958, + "num_tokens": 246704760.0, + "step": 9672 + }, + { + "epoch": 1.062266637381946, + "grad_norm": 2.061049699783325, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7165855169296265, + "num_tokens": 246732936.0, + "step": 9673 + }, + { + "epoch": 1.0623764550845596, + "grad_norm": 2.1857779026031494, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6926067471504211, + "num_tokens": 246763735.0, + "step": 9674 + }, + { + "epoch": 1.0624862727871733, + "grad_norm": 2.6330132484436035, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7195044755935669, + "num_tokens": 246785033.0, + "step": 9675 + }, + { + "epoch": 1.0625960904897869, + "grad_norm": 2.5443522930145264, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7340016961097717, + "num_tokens": 246806532.0, + "step": 9676 + }, + { + "epoch": 1.0627059081924006, + "grad_norm": 2.3711163997650146, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7119897603988647, + "num_tokens": 246830398.0, + "step": 9677 + }, + { + "epoch": 1.0628157258950144, + "grad_norm": 2.2342636585235596, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7095522284507751, + "num_tokens": 246855593.0, + "step": 9678 + }, + { + "epoch": 1.062925543597628, + "grad_norm": 1.876136064529419, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7147005796432495, + "num_tokens": 246890469.0, + "step": 9679 + }, + { + "epoch": 1.0630353613002417, + "grad_norm": 2.557671070098877, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7372034788131714, + "num_tokens": 246911905.0, + "step": 9680 + }, + { + "epoch": 1.0631451790028552, + "grad_norm": 2.500365972518921, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7180258631706238, + "num_tokens": 246934140.0, + "step": 9681 + }, + { + "epoch": 1.063254996705469, + "grad_norm": 2.1986169815063477, + "learning_rate": 1e-06, + "loss": 1.0543, + "mean_token_accuracy": 0.6884099245071411, + "num_tokens": 246964167.0, + "step": 9682 + }, + { + "epoch": 1.0633648144080825, + "grad_norm": 2.1347858905792236, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7084230184555054, + "num_tokens": 246993676.0, + "step": 9683 + }, + { + "epoch": 1.0634746321106963, + "grad_norm": 2.069141149520874, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7008529901504517, + "num_tokens": 247022038.0, + "step": 9684 + }, + { + "epoch": 1.0635844498133098, + "grad_norm": 2.3671460151672363, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7187116146087646, + "num_tokens": 247046226.0, + "step": 9685 + }, + { + "epoch": 1.0636942675159236, + "grad_norm": 2.702392101287842, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7264708280563354, + "num_tokens": 247066318.0, + "step": 9686 + }, + { + "epoch": 1.0638040852185373, + "grad_norm": 2.1960136890411377, + "learning_rate": 1e-06, + "loss": 1.0337, + "mean_token_accuracy": 0.6834716200828552, + "num_tokens": 247094575.0, + "step": 9687 + }, + { + "epoch": 1.0639139029211508, + "grad_norm": 2.3777308464050293, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.721731424331665, + "num_tokens": 247119211.0, + "step": 9688 + }, + { + "epoch": 1.0640237206237646, + "grad_norm": 2.0765702724456787, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7332338094711304, + "num_tokens": 247147913.0, + "step": 9689 + }, + { + "epoch": 1.0641335383263781, + "grad_norm": 2.5834197998046875, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7379210591316223, + "num_tokens": 247167500.0, + "step": 9690 + }, + { + "epoch": 1.064243356028992, + "grad_norm": 2.3452742099761963, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7319958209991455, + "num_tokens": 247189097.0, + "step": 9691 + }, + { + "epoch": 1.0643531737316057, + "grad_norm": 2.517058849334717, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7320249080657959, + "num_tokens": 247209170.0, + "step": 9692 + }, + { + "epoch": 1.0644629914342192, + "grad_norm": 2.4669625759124756, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7053107023239136, + "num_tokens": 247232232.0, + "step": 9693 + }, + { + "epoch": 1.064572809136833, + "grad_norm": 2.153026580810547, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7207307815551758, + "num_tokens": 247261387.0, + "step": 9694 + }, + { + "epoch": 1.0646826268394465, + "grad_norm": 2.6655795574188232, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7457537651062012, + "num_tokens": 247279502.0, + "step": 9695 + }, + { + "epoch": 1.0647924445420602, + "grad_norm": 2.5218589305877686, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7280860543251038, + "num_tokens": 247299470.0, + "step": 9696 + }, + { + "epoch": 1.0649022622446738, + "grad_norm": 2.2130978107452393, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7228262424468994, + "num_tokens": 247324903.0, + "step": 9697 + }, + { + "epoch": 1.0650120799472875, + "grad_norm": 2.6275205612182617, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7169944643974304, + "num_tokens": 247344541.0, + "step": 9698 + }, + { + "epoch": 1.065121897649901, + "grad_norm": 2.566673994064331, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7192947864532471, + "num_tokens": 247364463.0, + "step": 9699 + }, + { + "epoch": 1.0652317153525148, + "grad_norm": 2.184896945953369, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.731181800365448, + "num_tokens": 247390900.0, + "step": 9700 + }, + { + "epoch": 1.0653415330551286, + "grad_norm": 2.102186918258667, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7012408971786499, + "num_tokens": 247419917.0, + "step": 9701 + }, + { + "epoch": 1.065451350757742, + "grad_norm": 1.9948065280914307, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7342721223831177, + "num_tokens": 247450322.0, + "step": 9702 + }, + { + "epoch": 1.0655611684603559, + "grad_norm": 2.128391981124878, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7438328266143799, + "num_tokens": 247476077.0, + "step": 9703 + }, + { + "epoch": 1.0656709861629694, + "grad_norm": 1.9610106945037842, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7344982028007507, + "num_tokens": 247506611.0, + "step": 9704 + }, + { + "epoch": 1.0657808038655832, + "grad_norm": 2.536322593688965, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7347464561462402, + "num_tokens": 247526580.0, + "step": 9705 + }, + { + "epoch": 1.0658906215681967, + "grad_norm": 2.173769474029541, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7392650842666626, + "num_tokens": 247553530.0, + "step": 9706 + }, + { + "epoch": 1.0660004392708105, + "grad_norm": 2.739396572113037, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7358112931251526, + "num_tokens": 247571535.0, + "step": 9707 + }, + { + "epoch": 1.0661102569734242, + "grad_norm": 1.9278018474578857, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.6884937286376953, + "num_tokens": 247609223.0, + "step": 9708 + }, + { + "epoch": 1.0662200746760377, + "grad_norm": 1.9996693134307861, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7178349494934082, + "num_tokens": 247640500.0, + "step": 9709 + }, + { + "epoch": 1.0663298923786515, + "grad_norm": 2.3953471183776855, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7164960503578186, + "num_tokens": 247662621.0, + "step": 9710 + }, + { + "epoch": 1.066439710081265, + "grad_norm": 2.6249935626983643, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.714211642742157, + "num_tokens": 247684457.0, + "step": 9711 + }, + { + "epoch": 1.0665495277838788, + "grad_norm": 2.055685520172119, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7298886775970459, + "num_tokens": 247711183.0, + "step": 9712 + }, + { + "epoch": 1.0666593454864923, + "grad_norm": 2.225571632385254, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7280089855194092, + "num_tokens": 247735765.0, + "step": 9713 + }, + { + "epoch": 1.066769163189106, + "grad_norm": 2.458451747894287, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7192882299423218, + "num_tokens": 247757393.0, + "step": 9714 + }, + { + "epoch": 1.0668789808917198, + "grad_norm": 2.4982497692108154, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7023407220840454, + "num_tokens": 247781052.0, + "step": 9715 + }, + { + "epoch": 1.0669887985943334, + "grad_norm": 2.0629525184631348, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.709994912147522, + "num_tokens": 247810406.0, + "step": 9716 + }, + { + "epoch": 1.0670986162969471, + "grad_norm": 2.3253650665283203, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7143028974533081, + "num_tokens": 247835291.0, + "step": 9717 + }, + { + "epoch": 1.0672084339995607, + "grad_norm": 1.9746891260147095, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7068137526512146, + "num_tokens": 247865320.0, + "step": 9718 + }, + { + "epoch": 1.0673182517021744, + "grad_norm": 2.3906590938568115, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7263293862342834, + "num_tokens": 247887175.0, + "step": 9719 + }, + { + "epoch": 1.067428069404788, + "grad_norm": 2.214303731918335, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7047320604324341, + "num_tokens": 247912147.0, + "step": 9720 + }, + { + "epoch": 1.0675378871074017, + "grad_norm": 2.309443473815918, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7190327644348145, + "num_tokens": 247936915.0, + "step": 9721 + }, + { + "epoch": 1.0676477048100155, + "grad_norm": 2.5261788368225098, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7152124643325806, + "num_tokens": 247959546.0, + "step": 9722 + }, + { + "epoch": 1.067757522512629, + "grad_norm": 2.505481243133545, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7086104154586792, + "num_tokens": 247982390.0, + "step": 9723 + }, + { + "epoch": 1.0678673402152428, + "grad_norm": 2.3481264114379883, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7115943431854248, + "num_tokens": 248006362.0, + "step": 9724 + }, + { + "epoch": 1.0679771579178563, + "grad_norm": 1.9801018238067627, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7036958336830139, + "num_tokens": 248035282.0, + "step": 9725 + }, + { + "epoch": 1.06808697562047, + "grad_norm": 2.326596260070801, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7304876446723938, + "num_tokens": 248058639.0, + "step": 9726 + }, + { + "epoch": 1.0681967933230836, + "grad_norm": 2.2357664108276367, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7159866690635681, + "num_tokens": 248085827.0, + "step": 9727 + }, + { + "epoch": 1.0683066110256974, + "grad_norm": 2.2021801471710205, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7234730124473572, + "num_tokens": 248112579.0, + "step": 9728 + }, + { + "epoch": 1.068416428728311, + "grad_norm": 2.1157617568969727, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7040196657180786, + "num_tokens": 248140384.0, + "step": 9729 + }, + { + "epoch": 1.0685262464309246, + "grad_norm": 2.365551710128784, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7297458052635193, + "num_tokens": 248162650.0, + "step": 9730 + }, + { + "epoch": 1.0686360641335384, + "grad_norm": 2.233654260635376, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7299333810806274, + "num_tokens": 248186542.0, + "step": 9731 + }, + { + "epoch": 1.068745881836152, + "grad_norm": 2.478226661682129, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7196257710456848, + "num_tokens": 248209246.0, + "step": 9732 + }, + { + "epoch": 1.0688556995387657, + "grad_norm": 2.3529396057128906, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.7011700868606567, + "num_tokens": 248235147.0, + "step": 9733 + }, + { + "epoch": 1.0689655172413792, + "grad_norm": 2.3349928855895996, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7271194458007812, + "num_tokens": 248258788.0, + "step": 9734 + }, + { + "epoch": 1.069075334943993, + "grad_norm": 2.386334180831909, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7243747115135193, + "num_tokens": 248283271.0, + "step": 9735 + }, + { + "epoch": 1.0691851526466065, + "grad_norm": 2.3428027629852295, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7033801674842834, + "num_tokens": 248307132.0, + "step": 9736 + }, + { + "epoch": 1.0692949703492203, + "grad_norm": 2.1197781562805176, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.708595871925354, + "num_tokens": 248335133.0, + "step": 9737 + }, + { + "epoch": 1.069404788051834, + "grad_norm": 2.666050910949707, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7237362265586853, + "num_tokens": 248353990.0, + "step": 9738 + }, + { + "epoch": 1.0695146057544476, + "grad_norm": 2.4781763553619385, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7409787178039551, + "num_tokens": 248375019.0, + "step": 9739 + }, + { + "epoch": 1.0696244234570613, + "grad_norm": 2.4581966400146484, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7276634573936462, + "num_tokens": 248399462.0, + "step": 9740 + }, + { + "epoch": 1.0697342411596749, + "grad_norm": 2.3060877323150635, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7190672159194946, + "num_tokens": 248425754.0, + "step": 9741 + }, + { + "epoch": 1.0698440588622886, + "grad_norm": 2.1134986877441406, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7204951047897339, + "num_tokens": 248454211.0, + "step": 9742 + }, + { + "epoch": 1.0699538765649024, + "grad_norm": 2.071183443069458, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7238793969154358, + "num_tokens": 248482519.0, + "step": 9743 + }, + { + "epoch": 1.070063694267516, + "grad_norm": 1.98123300075531, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7079147696495056, + "num_tokens": 248512138.0, + "step": 9744 + }, + { + "epoch": 1.0701735119701297, + "grad_norm": 2.369900703430176, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7125335335731506, + "num_tokens": 248533783.0, + "step": 9745 + }, + { + "epoch": 1.0702833296727432, + "grad_norm": 2.231529951095581, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7015530467033386, + "num_tokens": 248559898.0, + "step": 9746 + }, + { + "epoch": 1.070393147375357, + "grad_norm": 2.2975056171417236, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7212345600128174, + "num_tokens": 248583458.0, + "step": 9747 + }, + { + "epoch": 1.0705029650779705, + "grad_norm": 2.182443141937256, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7332938313484192, + "num_tokens": 248608406.0, + "step": 9748 + }, + { + "epoch": 1.0706127827805842, + "grad_norm": 2.3864121437072754, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7364782094955444, + "num_tokens": 248631255.0, + "step": 9749 + }, + { + "epoch": 1.0707226004831978, + "grad_norm": 2.399597644805908, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7381346821784973, + "num_tokens": 248651676.0, + "step": 9750 + }, + { + "epoch": 1.0708324181858115, + "grad_norm": 2.371518611907959, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7227951288223267, + "num_tokens": 248678920.0, + "step": 9751 + }, + { + "epoch": 1.0709422358884253, + "grad_norm": 2.705462694168091, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7058377265930176, + "num_tokens": 248697765.0, + "step": 9752 + }, + { + "epoch": 1.0710520535910388, + "grad_norm": 1.9205485582351685, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7073575258255005, + "num_tokens": 248734230.0, + "step": 9753 + }, + { + "epoch": 1.0711618712936526, + "grad_norm": 2.350496768951416, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7099549770355225, + "num_tokens": 248757062.0, + "step": 9754 + }, + { + "epoch": 1.0712716889962661, + "grad_norm": 2.3106863498687744, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7051352262496948, + "num_tokens": 248782682.0, + "step": 9755 + }, + { + "epoch": 1.0713815066988799, + "grad_norm": 2.494849681854248, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7394623756408691, + "num_tokens": 248802534.0, + "step": 9756 + }, + { + "epoch": 1.0714913244014934, + "grad_norm": 2.09352970123291, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7058260440826416, + "num_tokens": 248832597.0, + "step": 9757 + }, + { + "epoch": 1.0716011421041072, + "grad_norm": 2.4964232444763184, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.718722403049469, + "num_tokens": 248853168.0, + "step": 9758 + }, + { + "epoch": 1.071710959806721, + "grad_norm": 2.2987899780273438, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7245322465896606, + "num_tokens": 248876007.0, + "step": 9759 + }, + { + "epoch": 1.0718207775093345, + "grad_norm": 1.9627526998519897, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7186389565467834, + "num_tokens": 248905961.0, + "step": 9760 + }, + { + "epoch": 1.0719305952119482, + "grad_norm": 2.2800474166870117, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7390737533569336, + "num_tokens": 248928554.0, + "step": 9761 + }, + { + "epoch": 1.0720404129145618, + "grad_norm": 2.4159467220306396, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7199383974075317, + "num_tokens": 248952312.0, + "step": 9762 + }, + { + "epoch": 1.0721502306171755, + "grad_norm": 2.2136247158050537, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7015396356582642, + "num_tokens": 248978269.0, + "step": 9763 + }, + { + "epoch": 1.072260048319789, + "grad_norm": 2.4361343383789062, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7401242256164551, + "num_tokens": 248998359.0, + "step": 9764 + }, + { + "epoch": 1.0723698660224028, + "grad_norm": 2.5227174758911133, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7245714664459229, + "num_tokens": 249018786.0, + "step": 9765 + }, + { + "epoch": 1.0724796837250166, + "grad_norm": 2.1062073707580566, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7154452800750732, + "num_tokens": 249048525.0, + "step": 9766 + }, + { + "epoch": 1.07258950142763, + "grad_norm": 2.2207348346710205, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7023744583129883, + "num_tokens": 249074475.0, + "step": 9767 + }, + { + "epoch": 1.0726993191302439, + "grad_norm": 2.118673801422119, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7014721632003784, + "num_tokens": 249102518.0, + "step": 9768 + }, + { + "epoch": 1.0728091368328574, + "grad_norm": 2.1271278858184814, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7078936696052551, + "num_tokens": 249132070.0, + "step": 9769 + }, + { + "epoch": 1.0729189545354711, + "grad_norm": 2.418534517288208, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7410140037536621, + "num_tokens": 249152038.0, + "step": 9770 + }, + { + "epoch": 1.0730287722380847, + "grad_norm": 2.0697202682495117, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7137448787689209, + "num_tokens": 249180919.0, + "step": 9771 + }, + { + "epoch": 1.0731385899406984, + "grad_norm": 2.5090646743774414, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7002333402633667, + "num_tokens": 249202718.0, + "step": 9772 + }, + { + "epoch": 1.0732484076433122, + "grad_norm": 2.216488838195801, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7353518009185791, + "num_tokens": 249227665.0, + "step": 9773 + }, + { + "epoch": 1.0733582253459257, + "grad_norm": 2.272836446762085, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.719434916973114, + "num_tokens": 249252536.0, + "step": 9774 + }, + { + "epoch": 1.0734680430485395, + "grad_norm": 2.1933839321136475, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.732658863067627, + "num_tokens": 249277108.0, + "step": 9775 + }, + { + "epoch": 1.073577860751153, + "grad_norm": 2.20261549949646, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7220105528831482, + "num_tokens": 249303170.0, + "step": 9776 + }, + { + "epoch": 1.0736876784537668, + "grad_norm": 2.3021726608276367, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7369709014892578, + "num_tokens": 249326585.0, + "step": 9777 + }, + { + "epoch": 1.0737974961563803, + "grad_norm": 2.184199094772339, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.744420051574707, + "num_tokens": 249351947.0, + "step": 9778 + }, + { + "epoch": 1.073907313858994, + "grad_norm": 2.441702365875244, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.7562220096588135, + "num_tokens": 249373284.0, + "step": 9779 + }, + { + "epoch": 1.0740171315616078, + "grad_norm": 1.9367733001708984, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7165929079055786, + "num_tokens": 249406292.0, + "step": 9780 + }, + { + "epoch": 1.0741269492642214, + "grad_norm": 2.044675827026367, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7199113368988037, + "num_tokens": 249435637.0, + "step": 9781 + }, + { + "epoch": 1.0742367669668351, + "grad_norm": 2.140068292617798, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7275561094284058, + "num_tokens": 249463070.0, + "step": 9782 + }, + { + "epoch": 1.0743465846694487, + "grad_norm": 2.5516960620880127, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7229271531105042, + "num_tokens": 249482660.0, + "step": 9783 + }, + { + "epoch": 1.0744564023720624, + "grad_norm": 2.1946325302124023, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6884784698486328, + "num_tokens": 249512148.0, + "step": 9784 + }, + { + "epoch": 1.074566220074676, + "grad_norm": 2.1021153926849365, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7191805839538574, + "num_tokens": 249538932.0, + "step": 9785 + }, + { + "epoch": 1.0746760377772897, + "grad_norm": 2.5928101539611816, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7080122232437134, + "num_tokens": 249559805.0, + "step": 9786 + }, + { + "epoch": 1.0747858554799035, + "grad_norm": 2.2644333839416504, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7160294651985168, + "num_tokens": 249586017.0, + "step": 9787 + }, + { + "epoch": 1.074895673182517, + "grad_norm": 2.3355801105499268, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7062630653381348, + "num_tokens": 249612526.0, + "step": 9788 + }, + { + "epoch": 1.0750054908851308, + "grad_norm": 2.16593861579895, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7086023688316345, + "num_tokens": 249640386.0, + "step": 9789 + }, + { + "epoch": 1.0751153085877443, + "grad_norm": 2.3944101333618164, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7215791940689087, + "num_tokens": 249663177.0, + "step": 9790 + }, + { + "epoch": 1.075225126290358, + "grad_norm": 2.0623035430908203, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7116050720214844, + "num_tokens": 249694690.0, + "step": 9791 + }, + { + "epoch": 1.0753349439929716, + "grad_norm": 2.133270263671875, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.70878005027771, + "num_tokens": 249721875.0, + "step": 9792 + }, + { + "epoch": 1.0754447616955853, + "grad_norm": 2.062565803527832, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7140602469444275, + "num_tokens": 249751382.0, + "step": 9793 + }, + { + "epoch": 1.075554579398199, + "grad_norm": 2.1285672187805176, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7097212076187134, + "num_tokens": 249777424.0, + "step": 9794 + }, + { + "epoch": 1.0756643971008126, + "grad_norm": 2.3768444061279297, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7105062007904053, + "num_tokens": 249801161.0, + "step": 9795 + }, + { + "epoch": 1.0757742148034264, + "grad_norm": 2.3557958602905273, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7058460116386414, + "num_tokens": 249827130.0, + "step": 9796 + }, + { + "epoch": 1.07588403250604, + "grad_norm": 2.4555211067199707, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7167708873748779, + "num_tokens": 249848813.0, + "step": 9797 + }, + { + "epoch": 1.0759938502086537, + "grad_norm": 2.147000551223755, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7321121096611023, + "num_tokens": 249876904.0, + "step": 9798 + }, + { + "epoch": 1.0761036679112672, + "grad_norm": 2.3431334495544434, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.6971635818481445, + "num_tokens": 249903809.0, + "step": 9799 + }, + { + "epoch": 1.076213485613881, + "grad_norm": 2.603184461593628, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7363660931587219, + "num_tokens": 249922303.0, + "step": 9800 + }, + { + "epoch": 1.0763233033164945, + "grad_norm": 2.3644673824310303, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7116775512695312, + "num_tokens": 249946056.0, + "step": 9801 + }, + { + "epoch": 1.0764331210191083, + "grad_norm": 2.2538461685180664, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7213747501373291, + "num_tokens": 249971605.0, + "step": 9802 + }, + { + "epoch": 1.076542938721722, + "grad_norm": 2.5154783725738525, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7373058199882507, + "num_tokens": 249992180.0, + "step": 9803 + }, + { + "epoch": 1.0766527564243356, + "grad_norm": 2.0871989727020264, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7080493569374084, + "num_tokens": 250022373.0, + "step": 9804 + }, + { + "epoch": 1.0767625741269493, + "grad_norm": 2.5914530754089355, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.7398601770401001, + "num_tokens": 250041400.0, + "step": 9805 + }, + { + "epoch": 1.0768723918295628, + "grad_norm": 2.4229280948638916, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7204024791717529, + "num_tokens": 250064959.0, + "step": 9806 + }, + { + "epoch": 1.0769822095321766, + "grad_norm": 2.5972328186035156, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7178284525871277, + "num_tokens": 250086133.0, + "step": 9807 + }, + { + "epoch": 1.0770920272347904, + "grad_norm": 2.087585926055908, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7429309487342834, + "num_tokens": 250113238.0, + "step": 9808 + }, + { + "epoch": 1.077201844937404, + "grad_norm": 1.994315505027771, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7251095771789551, + "num_tokens": 250144676.0, + "step": 9809 + }, + { + "epoch": 1.0773116626400177, + "grad_norm": 1.9659844636917114, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7380478978157043, + "num_tokens": 250173222.0, + "step": 9810 + }, + { + "epoch": 1.0774214803426312, + "grad_norm": 1.845845103263855, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7213231325149536, + "num_tokens": 250207292.0, + "step": 9811 + }, + { + "epoch": 1.077531298045245, + "grad_norm": 2.1201171875, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7288601398468018, + "num_tokens": 250233041.0, + "step": 9812 + }, + { + "epoch": 1.0776411157478585, + "grad_norm": 2.160569667816162, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7307047843933105, + "num_tokens": 250260073.0, + "step": 9813 + }, + { + "epoch": 1.0777509334504722, + "grad_norm": 2.116450786590576, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.709460973739624, + "num_tokens": 250288946.0, + "step": 9814 + }, + { + "epoch": 1.0778607511530858, + "grad_norm": 2.1352596282958984, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7289207577705383, + "num_tokens": 250314506.0, + "step": 9815 + }, + { + "epoch": 1.0779705688556995, + "grad_norm": 2.258629083633423, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.716670036315918, + "num_tokens": 250340907.0, + "step": 9816 + }, + { + "epoch": 1.0780803865583133, + "grad_norm": 2.3582563400268555, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7155259847640991, + "num_tokens": 250365038.0, + "step": 9817 + }, + { + "epoch": 1.0781902042609268, + "grad_norm": 2.1606009006500244, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7109259963035583, + "num_tokens": 250391805.0, + "step": 9818 + }, + { + "epoch": 1.0783000219635406, + "grad_norm": 2.300797939300537, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.733124315738678, + "num_tokens": 250415805.0, + "step": 9819 + }, + { + "epoch": 1.0784098396661541, + "grad_norm": 2.1329598426818848, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7157538533210754, + "num_tokens": 250442196.0, + "step": 9820 + }, + { + "epoch": 1.0785196573687679, + "grad_norm": 2.3429083824157715, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7226247787475586, + "num_tokens": 250467022.0, + "step": 9821 + }, + { + "epoch": 1.0786294750713814, + "grad_norm": 2.175297260284424, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.6986533403396606, + "num_tokens": 250494682.0, + "step": 9822 + }, + { + "epoch": 1.0787392927739952, + "grad_norm": 2.207587242126465, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6941565871238708, + "num_tokens": 250523494.0, + "step": 9823 + }, + { + "epoch": 1.078849110476609, + "grad_norm": 1.8769150972366333, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6891241669654846, + "num_tokens": 250558007.0, + "step": 9824 + }, + { + "epoch": 1.0789589281792225, + "grad_norm": 2.3047659397125244, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7170876264572144, + "num_tokens": 250581121.0, + "step": 9825 + }, + { + "epoch": 1.0790687458818362, + "grad_norm": 2.4420957565307617, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7294192314147949, + "num_tokens": 250603199.0, + "step": 9826 + }, + { + "epoch": 1.0791785635844497, + "grad_norm": 2.6095285415649414, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7308920621871948, + "num_tokens": 250623812.0, + "step": 9827 + }, + { + "epoch": 1.0792883812870635, + "grad_norm": 1.9722017049789429, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7115206122398376, + "num_tokens": 250655972.0, + "step": 9828 + }, + { + "epoch": 1.079398198989677, + "grad_norm": 2.3355085849761963, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7389600276947021, + "num_tokens": 250681718.0, + "step": 9829 + }, + { + "epoch": 1.0795080166922908, + "grad_norm": 2.0917110443115234, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.7011011838912964, + "num_tokens": 250711601.0, + "step": 9830 + }, + { + "epoch": 1.0796178343949046, + "grad_norm": 2.5888564586639404, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7231061458587646, + "num_tokens": 250732974.0, + "step": 9831 + }, + { + "epoch": 1.079727652097518, + "grad_norm": 2.476273536682129, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7271276712417603, + "num_tokens": 250754966.0, + "step": 9832 + }, + { + "epoch": 1.0798374698001318, + "grad_norm": 2.385300874710083, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7245966196060181, + "num_tokens": 250779981.0, + "step": 9833 + }, + { + "epoch": 1.0799472875027454, + "grad_norm": 2.1403086185455322, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7183573246002197, + "num_tokens": 250808095.0, + "step": 9834 + }, + { + "epoch": 1.0800571052053591, + "grad_norm": 2.1990108489990234, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7252381443977356, + "num_tokens": 250835230.0, + "step": 9835 + }, + { + "epoch": 1.0801669229079727, + "grad_norm": 2.1559181213378906, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7107862234115601, + "num_tokens": 250864102.0, + "step": 9836 + }, + { + "epoch": 1.0802767406105864, + "grad_norm": 2.4089884757995605, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7357542514801025, + "num_tokens": 250884329.0, + "step": 9837 + }, + { + "epoch": 1.0803865583132002, + "grad_norm": 2.5008580684661865, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.735043466091156, + "num_tokens": 250905061.0, + "step": 9838 + }, + { + "epoch": 1.0804963760158137, + "grad_norm": 2.184546947479248, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6990537047386169, + "num_tokens": 250935246.0, + "step": 9839 + }, + { + "epoch": 1.0806061937184275, + "grad_norm": 2.122291088104248, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7109624147415161, + "num_tokens": 250964228.0, + "step": 9840 + }, + { + "epoch": 1.080716011421041, + "grad_norm": 2.1716606616973877, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7438116073608398, + "num_tokens": 250988695.0, + "step": 9841 + }, + { + "epoch": 1.0808258291236548, + "grad_norm": 2.25754714012146, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7307486534118652, + "num_tokens": 251014889.0, + "step": 9842 + }, + { + "epoch": 1.0809356468262683, + "grad_norm": 2.070234775543213, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7302176356315613, + "num_tokens": 251043512.0, + "step": 9843 + }, + { + "epoch": 1.081045464528882, + "grad_norm": 2.372626304626465, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7139908075332642, + "num_tokens": 251066958.0, + "step": 9844 + }, + { + "epoch": 1.0811552822314958, + "grad_norm": 2.2567784786224365, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7244085669517517, + "num_tokens": 251092971.0, + "step": 9845 + }, + { + "epoch": 1.0812650999341094, + "grad_norm": 2.5466156005859375, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7397701740264893, + "num_tokens": 251113408.0, + "step": 9846 + }, + { + "epoch": 1.081374917636723, + "grad_norm": 2.3559775352478027, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7346835136413574, + "num_tokens": 251136333.0, + "step": 9847 + }, + { + "epoch": 1.0814847353393366, + "grad_norm": 2.3086998462677, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7168993949890137, + "num_tokens": 251161938.0, + "step": 9848 + }, + { + "epoch": 1.0815945530419504, + "grad_norm": 2.370574951171875, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.713333010673523, + "num_tokens": 251185436.0, + "step": 9849 + }, + { + "epoch": 1.081704370744564, + "grad_norm": 2.010801315307617, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7024230360984802, + "num_tokens": 251216287.0, + "step": 9850 + }, + { + "epoch": 1.0818141884471777, + "grad_norm": 2.4641165733337402, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7346740961074829, + "num_tokens": 251237270.0, + "step": 9851 + }, + { + "epoch": 1.0819240061497912, + "grad_norm": 2.02925181388855, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7044793367385864, + "num_tokens": 251267450.0, + "step": 9852 + }, + { + "epoch": 1.082033823852405, + "grad_norm": 2.2424395084381104, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.7451315522193909, + "num_tokens": 251296018.0, + "step": 9853 + }, + { + "epoch": 1.0821436415550187, + "grad_norm": 2.828684091567993, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7251355648040771, + "num_tokens": 251314007.0, + "step": 9854 + }, + { + "epoch": 1.0822534592576323, + "grad_norm": 2.1835851669311523, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7172449827194214, + "num_tokens": 251341186.0, + "step": 9855 + }, + { + "epoch": 1.082363276960246, + "grad_norm": 1.9379479885101318, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7114744186401367, + "num_tokens": 251373384.0, + "step": 9856 + }, + { + "epoch": 1.0824730946628596, + "grad_norm": 2.236074686050415, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7190774083137512, + "num_tokens": 251399757.0, + "step": 9857 + }, + { + "epoch": 1.0825829123654733, + "grad_norm": 2.064659595489502, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6903882026672363, + "num_tokens": 251432479.0, + "step": 9858 + }, + { + "epoch": 1.082692730068087, + "grad_norm": 2.4590816497802734, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7133480906486511, + "num_tokens": 251453952.0, + "step": 9859 + }, + { + "epoch": 1.0828025477707006, + "grad_norm": 1.9404412508010864, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7386283874511719, + "num_tokens": 251484632.0, + "step": 9860 + }, + { + "epoch": 1.0829123654733144, + "grad_norm": 2.3600406646728516, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7038918137550354, + "num_tokens": 251510524.0, + "step": 9861 + }, + { + "epoch": 1.083022183175928, + "grad_norm": 2.0869710445404053, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7193801403045654, + "num_tokens": 251537642.0, + "step": 9862 + }, + { + "epoch": 1.0831320008785417, + "grad_norm": 2.4150304794311523, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7341102361679077, + "num_tokens": 251561306.0, + "step": 9863 + }, + { + "epoch": 1.0832418185811552, + "grad_norm": 2.3456339836120605, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7365114092826843, + "num_tokens": 251585730.0, + "step": 9864 + }, + { + "epoch": 1.083351636283769, + "grad_norm": 2.3688085079193115, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7392070293426514, + "num_tokens": 251607910.0, + "step": 9865 + }, + { + "epoch": 1.0834614539863825, + "grad_norm": 2.515550374984741, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7573486566543579, + "num_tokens": 251627436.0, + "step": 9866 + }, + { + "epoch": 1.0835712716889963, + "grad_norm": 2.2405741214752197, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.735356330871582, + "num_tokens": 251651310.0, + "step": 9867 + }, + { + "epoch": 1.08368108939161, + "grad_norm": 2.5302228927612305, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.726622998714447, + "num_tokens": 251671548.0, + "step": 9868 + }, + { + "epoch": 1.0837909070942235, + "grad_norm": 2.129331111907959, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7291989326477051, + "num_tokens": 251699817.0, + "step": 9869 + }, + { + "epoch": 1.0839007247968373, + "grad_norm": 2.216825246810913, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.7063950300216675, + "num_tokens": 251725110.0, + "step": 9870 + }, + { + "epoch": 1.0840105424994508, + "grad_norm": 2.0011324882507324, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.742210328578949, + "num_tokens": 251752361.0, + "step": 9871 + }, + { + "epoch": 1.0841203602020646, + "grad_norm": 2.602311849594116, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.703773021697998, + "num_tokens": 251772430.0, + "step": 9872 + }, + { + "epoch": 1.0842301779046783, + "grad_norm": 2.3372981548309326, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6864343285560608, + "num_tokens": 251798702.0, + "step": 9873 + }, + { + "epoch": 1.0843399956072919, + "grad_norm": 2.2602083683013916, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7151026129722595, + "num_tokens": 251825666.0, + "step": 9874 + }, + { + "epoch": 1.0844498133099056, + "grad_norm": 2.223691463470459, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6998772025108337, + "num_tokens": 251851395.0, + "step": 9875 + }, + { + "epoch": 1.0845596310125192, + "grad_norm": 2.29719614982605, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.704389214515686, + "num_tokens": 251877811.0, + "step": 9876 + }, + { + "epoch": 1.084669448715133, + "grad_norm": 2.055687189102173, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6997144222259521, + "num_tokens": 251909622.0, + "step": 9877 + }, + { + "epoch": 1.0847792664177465, + "grad_norm": 2.274001121520996, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6915267109870911, + "num_tokens": 251936688.0, + "step": 9878 + }, + { + "epoch": 1.0848890841203602, + "grad_norm": 2.319646120071411, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7363062500953674, + "num_tokens": 251960929.0, + "step": 9879 + }, + { + "epoch": 1.0849989018229738, + "grad_norm": 2.078989267349243, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7292741537094116, + "num_tokens": 251988821.0, + "step": 9880 + }, + { + "epoch": 1.0851087195255875, + "grad_norm": 2.4376988410949707, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7187297344207764, + "num_tokens": 252011180.0, + "step": 9881 + }, + { + "epoch": 1.0852185372282013, + "grad_norm": 1.8872061967849731, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7041295766830444, + "num_tokens": 252048853.0, + "step": 9882 + }, + { + "epoch": 1.0853283549308148, + "grad_norm": 2.5658252239227295, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7205560207366943, + "num_tokens": 252070904.0, + "step": 9883 + }, + { + "epoch": 1.0854381726334286, + "grad_norm": 2.4132680892944336, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7194443941116333, + "num_tokens": 252094920.0, + "step": 9884 + }, + { + "epoch": 1.085547990336042, + "grad_norm": 2.0943143367767334, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.6984511613845825, + "num_tokens": 252124779.0, + "step": 9885 + }, + { + "epoch": 1.0856578080386559, + "grad_norm": 2.19600772857666, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7136399745941162, + "num_tokens": 252151459.0, + "step": 9886 + }, + { + "epoch": 1.0857676257412694, + "grad_norm": 2.3625400066375732, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.7429565787315369, + "num_tokens": 252174254.0, + "step": 9887 + }, + { + "epoch": 1.0858774434438832, + "grad_norm": 2.2912380695343018, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7114071249961853, + "num_tokens": 252199094.0, + "step": 9888 + }, + { + "epoch": 1.085987261146497, + "grad_norm": 2.4934613704681396, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7500137686729431, + "num_tokens": 252219933.0, + "step": 9889 + }, + { + "epoch": 1.0860970788491104, + "grad_norm": 2.2368760108947754, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.72751784324646, + "num_tokens": 252245032.0, + "step": 9890 + }, + { + "epoch": 1.0862068965517242, + "grad_norm": 2.3080193996429443, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7021815776824951, + "num_tokens": 252268466.0, + "step": 9891 + }, + { + "epoch": 1.0863167142543377, + "grad_norm": 2.0690250396728516, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.6986083984375, + "num_tokens": 252298897.0, + "step": 9892 + }, + { + "epoch": 1.0864265319569515, + "grad_norm": 2.1706104278564453, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7064294815063477, + "num_tokens": 252326265.0, + "step": 9893 + }, + { + "epoch": 1.086536349659565, + "grad_norm": 2.299088954925537, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7281414270401001, + "num_tokens": 252350432.0, + "step": 9894 + }, + { + "epoch": 1.0866461673621788, + "grad_norm": 3.0496926307678223, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7406389713287354, + "num_tokens": 252366728.0, + "step": 9895 + }, + { + "epoch": 1.0867559850647925, + "grad_norm": 2.1717607975006104, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.727060854434967, + "num_tokens": 252393358.0, + "step": 9896 + }, + { + "epoch": 1.086865802767406, + "grad_norm": 2.2248597145080566, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7067351341247559, + "num_tokens": 252417687.0, + "step": 9897 + }, + { + "epoch": 1.0869756204700198, + "grad_norm": 2.1169915199279785, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7274758815765381, + "num_tokens": 252443451.0, + "step": 9898 + }, + { + "epoch": 1.0870854381726334, + "grad_norm": 2.040726900100708, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.706455409526825, + "num_tokens": 252475039.0, + "step": 9899 + }, + { + "epoch": 1.0871952558752471, + "grad_norm": 2.4490301609039307, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7343684434890747, + "num_tokens": 252496886.0, + "step": 9900 + }, + { + "epoch": 1.0873050735778607, + "grad_norm": 2.187422037124634, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7118523716926575, + "num_tokens": 252526051.0, + "step": 9901 + }, + { + "epoch": 1.0874148912804744, + "grad_norm": 2.2742536067962646, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7219310402870178, + "num_tokens": 252551797.0, + "step": 9902 + }, + { + "epoch": 1.0875247089830882, + "grad_norm": 2.2340664863586426, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.727031409740448, + "num_tokens": 252577627.0, + "step": 9903 + }, + { + "epoch": 1.0876345266857017, + "grad_norm": 2.12153959274292, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7168594598770142, + "num_tokens": 252607193.0, + "step": 9904 + }, + { + "epoch": 1.0877443443883155, + "grad_norm": 2.131072521209717, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7171691060066223, + "num_tokens": 252634419.0, + "step": 9905 + }, + { + "epoch": 1.087854162090929, + "grad_norm": 2.2386841773986816, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7264623641967773, + "num_tokens": 252660260.0, + "step": 9906 + }, + { + "epoch": 1.0879639797935428, + "grad_norm": 2.29831600189209, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7253475189208984, + "num_tokens": 252684581.0, + "step": 9907 + }, + { + "epoch": 1.0880737974961563, + "grad_norm": 1.959887981414795, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7263668775558472, + "num_tokens": 252717314.0, + "step": 9908 + }, + { + "epoch": 1.08818361519877, + "grad_norm": 2.0738422870635986, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7370303273200989, + "num_tokens": 252745973.0, + "step": 9909 + }, + { + "epoch": 1.0882934329013838, + "grad_norm": 2.238909959793091, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.704056978225708, + "num_tokens": 252773023.0, + "step": 9910 + }, + { + "epoch": 1.0884032506039973, + "grad_norm": 2.182213068008423, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7258706092834473, + "num_tokens": 252799576.0, + "step": 9911 + }, + { + "epoch": 1.088513068306611, + "grad_norm": 2.435375213623047, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7307577133178711, + "num_tokens": 252821129.0, + "step": 9912 + }, + { + "epoch": 1.0886228860092246, + "grad_norm": 2.299389600753784, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7250006198883057, + "num_tokens": 252843986.0, + "step": 9913 + }, + { + "epoch": 1.0887327037118384, + "grad_norm": 2.1283152103424072, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7173181176185608, + "num_tokens": 252873253.0, + "step": 9914 + }, + { + "epoch": 1.088842521414452, + "grad_norm": 2.2326178550720215, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7156158089637756, + "num_tokens": 252899861.0, + "step": 9915 + }, + { + "epoch": 1.0889523391170657, + "grad_norm": 2.1258602142333984, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7115020155906677, + "num_tokens": 252931223.0, + "step": 9916 + }, + { + "epoch": 1.0890621568196792, + "grad_norm": 2.195300817489624, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7221229076385498, + "num_tokens": 252956992.0, + "step": 9917 + }, + { + "epoch": 1.089171974522293, + "grad_norm": 2.56103253364563, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7194918394088745, + "num_tokens": 252977484.0, + "step": 9918 + }, + { + "epoch": 1.0892817922249067, + "grad_norm": 2.304544687271118, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7039188146591187, + "num_tokens": 253004218.0, + "step": 9919 + }, + { + "epoch": 1.0893916099275203, + "grad_norm": 2.2484166622161865, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7095720767974854, + "num_tokens": 253029880.0, + "step": 9920 + }, + { + "epoch": 1.089501427630134, + "grad_norm": 2.1832473278045654, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7095577716827393, + "num_tokens": 253056734.0, + "step": 9921 + }, + { + "epoch": 1.0896112453327476, + "grad_norm": 2.0590052604675293, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7127891778945923, + "num_tokens": 253086638.0, + "step": 9922 + }, + { + "epoch": 1.0897210630353613, + "grad_norm": 2.309418201446533, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7428276538848877, + "num_tokens": 253110120.0, + "step": 9923 + }, + { + "epoch": 1.089830880737975, + "grad_norm": 2.215195417404175, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7184159755706787, + "num_tokens": 253135750.0, + "step": 9924 + }, + { + "epoch": 1.0899406984405886, + "grad_norm": 2.451347827911377, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7113596796989441, + "num_tokens": 253159515.0, + "step": 9925 + }, + { + "epoch": 1.0900505161432024, + "grad_norm": 2.086407423019409, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7109843492507935, + "num_tokens": 253187977.0, + "step": 9926 + }, + { + "epoch": 1.090160333845816, + "grad_norm": 2.135000467300415, + "learning_rate": 1e-06, + "loss": 1.0413, + "mean_token_accuracy": 0.6773360371589661, + "num_tokens": 253219436.0, + "step": 9927 + }, + { + "epoch": 1.0902701515484297, + "grad_norm": 2.184985876083374, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.6987051963806152, + "num_tokens": 253246507.0, + "step": 9928 + }, + { + "epoch": 1.0903799692510432, + "grad_norm": 2.3505520820617676, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7412304878234863, + "num_tokens": 253268626.0, + "step": 9929 + }, + { + "epoch": 1.090489786953657, + "grad_norm": 2.2119317054748535, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7154930233955383, + "num_tokens": 253296784.0, + "step": 9930 + }, + { + "epoch": 1.0905996046562705, + "grad_norm": 2.0802597999572754, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7201253175735474, + "num_tokens": 253323219.0, + "step": 9931 + }, + { + "epoch": 1.0907094223588842, + "grad_norm": 2.2991716861724854, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.7038145065307617, + "num_tokens": 253350183.0, + "step": 9932 + }, + { + "epoch": 1.090819240061498, + "grad_norm": 2.195072889328003, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7094780802726746, + "num_tokens": 253378387.0, + "step": 9933 + }, + { + "epoch": 1.0909290577641115, + "grad_norm": 2.0966145992279053, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7391190528869629, + "num_tokens": 253403933.0, + "step": 9934 + }, + { + "epoch": 1.0910388754667253, + "grad_norm": 2.0452888011932373, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.6939495801925659, + "num_tokens": 253434299.0, + "step": 9935 + }, + { + "epoch": 1.0911486931693388, + "grad_norm": 2.3613405227661133, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7052282691001892, + "num_tokens": 253457832.0, + "step": 9936 + }, + { + "epoch": 1.0912585108719526, + "grad_norm": 2.260293960571289, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7288531064987183, + "num_tokens": 253484168.0, + "step": 9937 + }, + { + "epoch": 1.0913683285745663, + "grad_norm": 2.275911331176758, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7191122174263, + "num_tokens": 253509581.0, + "step": 9938 + }, + { + "epoch": 1.0914781462771799, + "grad_norm": 2.2908616065979004, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7139749526977539, + "num_tokens": 253535138.0, + "step": 9939 + }, + { + "epoch": 1.0915879639797936, + "grad_norm": 2.3319098949432373, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.6911382675170898, + "num_tokens": 253560284.0, + "step": 9940 + }, + { + "epoch": 1.0916977816824072, + "grad_norm": 2.5461127758026123, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7156693935394287, + "num_tokens": 253582023.0, + "step": 9941 + }, + { + "epoch": 1.091807599385021, + "grad_norm": 2.068571090698242, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7046526670455933, + "num_tokens": 253610102.0, + "step": 9942 + }, + { + "epoch": 1.0919174170876345, + "grad_norm": 2.3071436882019043, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7212287783622742, + "num_tokens": 253635111.0, + "step": 9943 + }, + { + "epoch": 1.0920272347902482, + "grad_norm": 2.217184066772461, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.6949268579483032, + "num_tokens": 253662907.0, + "step": 9944 + }, + { + "epoch": 1.0921370524928617, + "grad_norm": 2.275815963745117, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7134324312210083, + "num_tokens": 253686916.0, + "step": 9945 + }, + { + "epoch": 1.0922468701954755, + "grad_norm": 2.2057011127471924, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7240042686462402, + "num_tokens": 253713465.0, + "step": 9946 + }, + { + "epoch": 1.0923566878980893, + "grad_norm": 1.966771125793457, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7054882049560547, + "num_tokens": 253745955.0, + "step": 9947 + }, + { + "epoch": 1.0924665056007028, + "grad_norm": 2.2262966632843018, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7142208814620972, + "num_tokens": 253773340.0, + "step": 9948 + }, + { + "epoch": 1.0925763233033166, + "grad_norm": 2.206594228744507, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7321759462356567, + "num_tokens": 253798901.0, + "step": 9949 + }, + { + "epoch": 1.09268614100593, + "grad_norm": 2.4942831993103027, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7240613698959351, + "num_tokens": 253819856.0, + "step": 9950 + }, + { + "epoch": 1.0927959587085438, + "grad_norm": 2.4737119674682617, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7242748141288757, + "num_tokens": 253840935.0, + "step": 9951 + }, + { + "epoch": 1.0929057764111574, + "grad_norm": 2.2849955558776855, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7256065011024475, + "num_tokens": 253864700.0, + "step": 9952 + }, + { + "epoch": 1.0930155941137711, + "grad_norm": 2.3849236965179443, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7190872430801392, + "num_tokens": 253888202.0, + "step": 9953 + }, + { + "epoch": 1.093125411816385, + "grad_norm": 2.3360936641693115, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.711780846118927, + "num_tokens": 253912101.0, + "step": 9954 + }, + { + "epoch": 1.0932352295189984, + "grad_norm": 2.725841522216797, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7425301671028137, + "num_tokens": 253929544.0, + "step": 9955 + }, + { + "epoch": 1.0933450472216122, + "grad_norm": 2.4672110080718994, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7106092572212219, + "num_tokens": 253951116.0, + "step": 9956 + }, + { + "epoch": 1.0934548649242257, + "grad_norm": 2.173600673675537, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7074731588363647, + "num_tokens": 253979751.0, + "step": 9957 + }, + { + "epoch": 1.0935646826268395, + "grad_norm": 2.182516574859619, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.717660665512085, + "num_tokens": 254006316.0, + "step": 9958 + }, + { + "epoch": 1.093674500329453, + "grad_norm": 2.033597707748413, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7203976511955261, + "num_tokens": 254037812.0, + "step": 9959 + }, + { + "epoch": 1.0937843180320668, + "grad_norm": 2.45857834815979, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7301203012466431, + "num_tokens": 254060256.0, + "step": 9960 + }, + { + "epoch": 1.0938941357346805, + "grad_norm": 2.171067237854004, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7142269611358643, + "num_tokens": 254088714.0, + "step": 9961 + }, + { + "epoch": 1.094003953437294, + "grad_norm": 2.3709285259246826, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7309662103652954, + "num_tokens": 254112410.0, + "step": 9962 + }, + { + "epoch": 1.0941137711399078, + "grad_norm": 2.2961628437042236, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7270650863647461, + "num_tokens": 254137105.0, + "step": 9963 + }, + { + "epoch": 1.0942235888425214, + "grad_norm": 2.4346113204956055, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7242364287376404, + "num_tokens": 254160219.0, + "step": 9964 + }, + { + "epoch": 1.0943334065451351, + "grad_norm": 2.0452682971954346, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7072356939315796, + "num_tokens": 254189592.0, + "step": 9965 + }, + { + "epoch": 1.0944432242477486, + "grad_norm": 2.5676262378692627, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7173510789871216, + "num_tokens": 254211253.0, + "step": 9966 + }, + { + "epoch": 1.0945530419503624, + "grad_norm": 2.1963953971862793, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7126237750053406, + "num_tokens": 254237808.0, + "step": 9967 + }, + { + "epoch": 1.0946628596529762, + "grad_norm": 2.011610507965088, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7142046093940735, + "num_tokens": 254267442.0, + "step": 9968 + }, + { + "epoch": 1.0947726773555897, + "grad_norm": 2.3180999755859375, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7253772020339966, + "num_tokens": 254291904.0, + "step": 9969 + }, + { + "epoch": 1.0948824950582035, + "grad_norm": 2.301668643951416, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7225143909454346, + "num_tokens": 254314622.0, + "step": 9970 + }, + { + "epoch": 1.094992312760817, + "grad_norm": 2.3239715099334717, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7239586114883423, + "num_tokens": 254340516.0, + "step": 9971 + }, + { + "epoch": 1.0951021304634307, + "grad_norm": 2.614314556121826, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7318025827407837, + "num_tokens": 254360536.0, + "step": 9972 + }, + { + "epoch": 1.0952119481660443, + "grad_norm": 2.370739221572876, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7231733202934265, + "num_tokens": 254384494.0, + "step": 9973 + }, + { + "epoch": 1.095321765868658, + "grad_norm": 2.2715184688568115, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7157046794891357, + "num_tokens": 254409712.0, + "step": 9974 + }, + { + "epoch": 1.0954315835712718, + "grad_norm": 2.326314926147461, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6965724229812622, + "num_tokens": 254437410.0, + "step": 9975 + }, + { + "epoch": 1.0955414012738853, + "grad_norm": 2.1867010593414307, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7148405313491821, + "num_tokens": 254466203.0, + "step": 9976 + }, + { + "epoch": 1.095651218976499, + "grad_norm": 2.264007806777954, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7156660556793213, + "num_tokens": 254493595.0, + "step": 9977 + }, + { + "epoch": 1.0957610366791126, + "grad_norm": 2.4635119438171387, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7202931642532349, + "num_tokens": 254516069.0, + "step": 9978 + }, + { + "epoch": 1.0958708543817264, + "grad_norm": 2.2951931953430176, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.7416266202926636, + "num_tokens": 254541544.0, + "step": 9979 + }, + { + "epoch": 1.09598067208434, + "grad_norm": 2.142042875289917, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.7026922702789307, + "num_tokens": 254570568.0, + "step": 9980 + }, + { + "epoch": 1.0960904897869537, + "grad_norm": 2.0164785385131836, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7078253030776978, + "num_tokens": 254601028.0, + "step": 9981 + }, + { + "epoch": 1.0962003074895672, + "grad_norm": 2.2911360263824463, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7319223880767822, + "num_tokens": 254624786.0, + "step": 9982 + }, + { + "epoch": 1.096310125192181, + "grad_norm": 2.2847819328308105, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7261250615119934, + "num_tokens": 254647815.0, + "step": 9983 + }, + { + "epoch": 1.0964199428947947, + "grad_norm": 2.019533634185791, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7103882431983948, + "num_tokens": 254679842.0, + "step": 9984 + }, + { + "epoch": 1.0965297605974083, + "grad_norm": 2.091858386993408, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7061119079589844, + "num_tokens": 254711382.0, + "step": 9985 + }, + { + "epoch": 1.096639578300022, + "grad_norm": 2.09049391746521, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7257882356643677, + "num_tokens": 254741175.0, + "step": 9986 + }, + { + "epoch": 1.0967493960026355, + "grad_norm": 2.41959810256958, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7072134017944336, + "num_tokens": 254765190.0, + "step": 9987 + }, + { + "epoch": 1.0968592137052493, + "grad_norm": 1.913692593574524, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7203341126441956, + "num_tokens": 254795910.0, + "step": 9988 + }, + { + "epoch": 1.096969031407863, + "grad_norm": 2.4191336631774902, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7176787257194519, + "num_tokens": 254818949.0, + "step": 9989 + }, + { + "epoch": 1.0970788491104766, + "grad_norm": 2.4421722888946533, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7085997462272644, + "num_tokens": 254843305.0, + "step": 9990 + }, + { + "epoch": 1.0971886668130904, + "grad_norm": 2.422071933746338, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7081507444381714, + "num_tokens": 254868818.0, + "step": 9991 + }, + { + "epoch": 1.0972984845157039, + "grad_norm": 2.125959634780884, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.7479652762413025, + "num_tokens": 254897039.0, + "step": 9992 + }, + { + "epoch": 1.0974083022183176, + "grad_norm": 2.6490819454193115, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7252209782600403, + "num_tokens": 254917744.0, + "step": 9993 + }, + { + "epoch": 1.0975181199209312, + "grad_norm": 2.0831711292266846, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7204846739768982, + "num_tokens": 254944942.0, + "step": 9994 + }, + { + "epoch": 1.097627937623545, + "grad_norm": 1.914355754852295, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7285907864570618, + "num_tokens": 254976775.0, + "step": 9995 + }, + { + "epoch": 1.0977377553261585, + "grad_norm": 2.54876971244812, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7374960780143738, + "num_tokens": 254996001.0, + "step": 9996 + }, + { + "epoch": 1.0978475730287722, + "grad_norm": 2.757448196411133, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7214294672012329, + "num_tokens": 255017073.0, + "step": 9997 + }, + { + "epoch": 1.097957390731386, + "grad_norm": 2.208019971847534, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7156364321708679, + "num_tokens": 255043411.0, + "step": 9998 + }, + { + "epoch": 1.0980672084339995, + "grad_norm": 2.081690549850464, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7182781100273132, + "num_tokens": 255073042.0, + "step": 9999 + }, + { + "epoch": 1.0981770261366133, + "grad_norm": 2.123936653137207, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7309502363204956, + "num_tokens": 255100231.0, + "step": 10000 + }, + { + "epoch": 1.0982868438392268, + "grad_norm": 2.434718370437622, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7220577001571655, + "num_tokens": 255122414.0, + "step": 10001 + }, + { + "epoch": 1.0983966615418406, + "grad_norm": 2.393629550933838, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6974060535430908, + "num_tokens": 255148596.0, + "step": 10002 + }, + { + "epoch": 1.098506479244454, + "grad_norm": 1.7405387163162231, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7231384515762329, + "num_tokens": 255187889.0, + "step": 10003 + }, + { + "epoch": 1.0986162969470679, + "grad_norm": 2.5463364124298096, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7222232222557068, + "num_tokens": 255208572.0, + "step": 10004 + }, + { + "epoch": 1.0987261146496816, + "grad_norm": 2.14782977104187, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7379366159439087, + "num_tokens": 255234035.0, + "step": 10005 + }, + { + "epoch": 1.0988359323522952, + "grad_norm": 2.5580010414123535, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.741615355014801, + "num_tokens": 255252940.0, + "step": 10006 + }, + { + "epoch": 1.098945750054909, + "grad_norm": 2.0675103664398193, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7148314714431763, + "num_tokens": 255281496.0, + "step": 10007 + }, + { + "epoch": 1.0990555677575224, + "grad_norm": 2.3365564346313477, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7361232042312622, + "num_tokens": 255304427.0, + "step": 10008 + }, + { + "epoch": 1.0991653854601362, + "grad_norm": 2.4030017852783203, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7121118307113647, + "num_tokens": 255327020.0, + "step": 10009 + }, + { + "epoch": 1.0992752031627497, + "grad_norm": 2.322932243347168, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7051913142204285, + "num_tokens": 255354424.0, + "step": 10010 + }, + { + "epoch": 1.0993850208653635, + "grad_norm": 2.1838674545288086, + "learning_rate": 1e-06, + "loss": 1.0264, + "mean_token_accuracy": 0.6849027276039124, + "num_tokens": 255386124.0, + "step": 10011 + }, + { + "epoch": 1.0994948385679773, + "grad_norm": 2.507809638977051, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7068865299224854, + "num_tokens": 255408569.0, + "step": 10012 + }, + { + "epoch": 1.0996046562705908, + "grad_norm": 1.9846415519714355, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7166849374771118, + "num_tokens": 255437916.0, + "step": 10013 + }, + { + "epoch": 1.0997144739732045, + "grad_norm": 2.258164405822754, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.713814377784729, + "num_tokens": 255463621.0, + "step": 10014 + }, + { + "epoch": 1.099824291675818, + "grad_norm": 2.6888010501861572, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7437410354614258, + "num_tokens": 255482343.0, + "step": 10015 + }, + { + "epoch": 1.0999341093784318, + "grad_norm": 2.263455629348755, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7027707695960999, + "num_tokens": 255509773.0, + "step": 10016 + }, + { + "epoch": 1.1000439270810454, + "grad_norm": 2.40084171295166, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7015018463134766, + "num_tokens": 255535260.0, + "step": 10017 + }, + { + "epoch": 1.1001537447836591, + "grad_norm": 2.198190450668335, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7224431037902832, + "num_tokens": 255561669.0, + "step": 10018 + }, + { + "epoch": 1.1002635624862729, + "grad_norm": 2.4582042694091797, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7294726967811584, + "num_tokens": 255583376.0, + "step": 10019 + }, + { + "epoch": 1.1003733801888864, + "grad_norm": 2.2653987407684326, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7212644815444946, + "num_tokens": 255609345.0, + "step": 10020 + }, + { + "epoch": 1.1004831978915002, + "grad_norm": 2.4473164081573486, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7118049263954163, + "num_tokens": 255634114.0, + "step": 10021 + }, + { + "epoch": 1.1005930155941137, + "grad_norm": 2.0223898887634277, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7241760492324829, + "num_tokens": 255663200.0, + "step": 10022 + }, + { + "epoch": 1.1007028332967275, + "grad_norm": 2.1548068523406982, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7084674835205078, + "num_tokens": 255690147.0, + "step": 10023 + }, + { + "epoch": 1.100812650999341, + "grad_norm": 2.396263360977173, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7179911136627197, + "num_tokens": 255712944.0, + "step": 10024 + }, + { + "epoch": 1.1009224687019548, + "grad_norm": 2.489738702774048, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7305947542190552, + "num_tokens": 255732815.0, + "step": 10025 + }, + { + "epoch": 1.1010322864045685, + "grad_norm": 2.5590686798095703, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7193976640701294, + "num_tokens": 255753828.0, + "step": 10026 + }, + { + "epoch": 1.101142104107182, + "grad_norm": 2.408859968185425, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7225438356399536, + "num_tokens": 255777608.0, + "step": 10027 + }, + { + "epoch": 1.1012519218097958, + "grad_norm": 2.659489154815674, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7143580317497253, + "num_tokens": 255797808.0, + "step": 10028 + }, + { + "epoch": 1.1013617395124093, + "grad_norm": 2.18481707572937, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7220882773399353, + "num_tokens": 255823558.0, + "step": 10029 + }, + { + "epoch": 1.101471557215023, + "grad_norm": 1.9151464700698853, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.7021428346633911, + "num_tokens": 255856194.0, + "step": 10030 + }, + { + "epoch": 1.1015813749176366, + "grad_norm": 1.8835458755493164, + "learning_rate": 1e-06, + "loss": 1.1149, + "mean_token_accuracy": 0.6631803512573242, + "num_tokens": 255891574.0, + "step": 10031 + }, + { + "epoch": 1.1016911926202504, + "grad_norm": 2.1779625415802, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7175543308258057, + "num_tokens": 255917730.0, + "step": 10032 + }, + { + "epoch": 1.1018010103228641, + "grad_norm": 2.4130752086639404, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7183609008789062, + "num_tokens": 255939706.0, + "step": 10033 + }, + { + "epoch": 1.1019108280254777, + "grad_norm": 2.245159149169922, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7069647908210754, + "num_tokens": 255965868.0, + "step": 10034 + }, + { + "epoch": 1.1020206457280914, + "grad_norm": 1.9801512956619263, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.708708643913269, + "num_tokens": 255997493.0, + "step": 10035 + }, + { + "epoch": 1.102130463430705, + "grad_norm": 2.751840353012085, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7201930284500122, + "num_tokens": 256016377.0, + "step": 10036 + }, + { + "epoch": 1.1022402811333187, + "grad_norm": 2.363253593444824, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7189860343933105, + "num_tokens": 256039264.0, + "step": 10037 + }, + { + "epoch": 1.1023500988359323, + "grad_norm": 2.3960418701171875, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7331628799438477, + "num_tokens": 256062014.0, + "step": 10038 + }, + { + "epoch": 1.102459916538546, + "grad_norm": 2.036068916320801, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7309832572937012, + "num_tokens": 256089319.0, + "step": 10039 + }, + { + "epoch": 1.1025697342411598, + "grad_norm": 2.276193618774414, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7175062298774719, + "num_tokens": 256115272.0, + "step": 10040 + }, + { + "epoch": 1.1026795519437733, + "grad_norm": 2.1947824954986572, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7102058529853821, + "num_tokens": 256141861.0, + "step": 10041 + }, + { + "epoch": 1.102789369646387, + "grad_norm": 2.1897289752960205, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7236055135726929, + "num_tokens": 256166145.0, + "step": 10042 + }, + { + "epoch": 1.1028991873490006, + "grad_norm": 2.431802272796631, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7191799879074097, + "num_tokens": 256189307.0, + "step": 10043 + }, + { + "epoch": 1.1030090050516144, + "grad_norm": 2.2150938510894775, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7601853609085083, + "num_tokens": 256213597.0, + "step": 10044 + }, + { + "epoch": 1.103118822754228, + "grad_norm": 2.3338558673858643, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7298679947853088, + "num_tokens": 256236362.0, + "step": 10045 + }, + { + "epoch": 1.1032286404568417, + "grad_norm": 2.39266037940979, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7300509214401245, + "num_tokens": 256260295.0, + "step": 10046 + }, + { + "epoch": 1.1033384581594552, + "grad_norm": 2.341634511947632, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7182341814041138, + "num_tokens": 256285132.0, + "step": 10047 + }, + { + "epoch": 1.103448275862069, + "grad_norm": 2.302039623260498, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7028001546859741, + "num_tokens": 256312271.0, + "step": 10048 + }, + { + "epoch": 1.1035580935646827, + "grad_norm": 2.2231874465942383, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7070982456207275, + "num_tokens": 256339696.0, + "step": 10049 + }, + { + "epoch": 1.1036679112672962, + "grad_norm": 2.5291526317596436, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7519786357879639, + "num_tokens": 256360328.0, + "step": 10050 + }, + { + "epoch": 1.10377772896991, + "grad_norm": 2.2980244159698486, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7319576740264893, + "num_tokens": 256383664.0, + "step": 10051 + }, + { + "epoch": 1.1038875466725235, + "grad_norm": 2.298585891723633, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7085036039352417, + "num_tokens": 256408006.0, + "step": 10052 + }, + { + "epoch": 1.1039973643751373, + "grad_norm": 2.4991791248321533, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7072569727897644, + "num_tokens": 256429313.0, + "step": 10053 + }, + { + "epoch": 1.104107182077751, + "grad_norm": 2.0185494422912598, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7238204479217529, + "num_tokens": 256459556.0, + "step": 10054 + }, + { + "epoch": 1.1042169997803646, + "grad_norm": 2.1565701961517334, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7136212587356567, + "num_tokens": 256488795.0, + "step": 10055 + }, + { + "epoch": 1.1043268174829783, + "grad_norm": 2.1387312412261963, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7193643450737, + "num_tokens": 256515411.0, + "step": 10056 + }, + { + "epoch": 1.1044366351855919, + "grad_norm": 2.185265064239502, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.7024142742156982, + "num_tokens": 256543485.0, + "step": 10057 + }, + { + "epoch": 1.1045464528882056, + "grad_norm": 2.2337710857391357, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7113011479377747, + "num_tokens": 256569851.0, + "step": 10058 + }, + { + "epoch": 1.1046562705908192, + "grad_norm": 2.0193634033203125, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7094953656196594, + "num_tokens": 256600941.0, + "step": 10059 + }, + { + "epoch": 1.104766088293433, + "grad_norm": 2.524505376815796, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7289174795150757, + "num_tokens": 256622491.0, + "step": 10060 + }, + { + "epoch": 1.1048759059960465, + "grad_norm": 2.2825071811676025, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.719182550907135, + "num_tokens": 256647051.0, + "step": 10061 + }, + { + "epoch": 1.1049857236986602, + "grad_norm": 2.2791099548339844, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7207361459732056, + "num_tokens": 256670489.0, + "step": 10062 + }, + { + "epoch": 1.105095541401274, + "grad_norm": 2.038738489151001, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7069609761238098, + "num_tokens": 256701116.0, + "step": 10063 + }, + { + "epoch": 1.1052053591038875, + "grad_norm": 2.2871694564819336, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7251356840133667, + "num_tokens": 256726811.0, + "step": 10064 + }, + { + "epoch": 1.1053151768065013, + "grad_norm": 2.484262704849243, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.7469226121902466, + "num_tokens": 256747727.0, + "step": 10065 + }, + { + "epoch": 1.1054249945091148, + "grad_norm": 2.4338581562042236, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.715736448764801, + "num_tokens": 256769871.0, + "step": 10066 + }, + { + "epoch": 1.1055348122117286, + "grad_norm": 2.3319950103759766, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7353521585464478, + "num_tokens": 256790692.0, + "step": 10067 + }, + { + "epoch": 1.105644629914342, + "grad_norm": 1.9666519165039062, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7254228591918945, + "num_tokens": 256823113.0, + "step": 10068 + }, + { + "epoch": 1.1057544476169558, + "grad_norm": 2.1242830753326416, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7055011987686157, + "num_tokens": 256850649.0, + "step": 10069 + }, + { + "epoch": 1.1058642653195696, + "grad_norm": 2.1626110076904297, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7212717533111572, + "num_tokens": 256877923.0, + "step": 10070 + }, + { + "epoch": 1.1059740830221831, + "grad_norm": 2.10233998298645, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7157979011535645, + "num_tokens": 256906219.0, + "step": 10071 + }, + { + "epoch": 1.106083900724797, + "grad_norm": 2.1661837100982666, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7179397940635681, + "num_tokens": 256932086.0, + "step": 10072 + }, + { + "epoch": 1.1061937184274104, + "grad_norm": 2.0030336380004883, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7296487092971802, + "num_tokens": 256963427.0, + "step": 10073 + }, + { + "epoch": 1.1063035361300242, + "grad_norm": 2.0383708477020264, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7265218496322632, + "num_tokens": 256993738.0, + "step": 10074 + }, + { + "epoch": 1.1064133538326377, + "grad_norm": 2.097640037536621, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7121259570121765, + "num_tokens": 257022600.0, + "step": 10075 + }, + { + "epoch": 1.1065231715352515, + "grad_norm": 2.458329916000366, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7220453023910522, + "num_tokens": 257043197.0, + "step": 10076 + }, + { + "epoch": 1.1066329892378652, + "grad_norm": 2.2455360889434814, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7168503999710083, + "num_tokens": 257070090.0, + "step": 10077 + }, + { + "epoch": 1.1067428069404788, + "grad_norm": 2.3827812671661377, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.733129620552063, + "num_tokens": 257091894.0, + "step": 10078 + }, + { + "epoch": 1.1068526246430925, + "grad_norm": 2.2543160915374756, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7254896759986877, + "num_tokens": 257116840.0, + "step": 10079 + }, + { + "epoch": 1.106962442345706, + "grad_norm": 2.6127848625183105, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7109191417694092, + "num_tokens": 257136699.0, + "step": 10080 + }, + { + "epoch": 1.1070722600483198, + "grad_norm": 2.461392402648926, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7310651540756226, + "num_tokens": 257157898.0, + "step": 10081 + }, + { + "epoch": 1.1071820777509334, + "grad_norm": 2.4977564811706543, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7488786578178406, + "num_tokens": 257179691.0, + "step": 10082 + }, + { + "epoch": 1.1072918954535471, + "grad_norm": 2.2344770431518555, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7226483821868896, + "num_tokens": 257203673.0, + "step": 10083 + }, + { + "epoch": 1.1074017131561609, + "grad_norm": 2.254875898361206, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7382777333259583, + "num_tokens": 257227762.0, + "step": 10084 + }, + { + "epoch": 1.1075115308587744, + "grad_norm": 2.322650194168091, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7384263277053833, + "num_tokens": 257250879.0, + "step": 10085 + }, + { + "epoch": 1.1076213485613882, + "grad_norm": 2.153355836868286, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7010968923568726, + "num_tokens": 257277957.0, + "step": 10086 + }, + { + "epoch": 1.1077311662640017, + "grad_norm": 2.2855584621429443, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7205587029457092, + "num_tokens": 257302675.0, + "step": 10087 + }, + { + "epoch": 1.1078409839666155, + "grad_norm": 2.1574883460998535, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7399854063987732, + "num_tokens": 257328224.0, + "step": 10088 + }, + { + "epoch": 1.107950801669229, + "grad_norm": 2.1655972003936768, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7183087468147278, + "num_tokens": 257352809.0, + "step": 10089 + }, + { + "epoch": 1.1080606193718427, + "grad_norm": 2.533207654953003, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.72400963306427, + "num_tokens": 257374099.0, + "step": 10090 + }, + { + "epoch": 1.1081704370744565, + "grad_norm": 2.613408088684082, + "learning_rate": 1e-06, + "loss": 0.7794, + "mean_token_accuracy": 0.7489150762557983, + "num_tokens": 257392190.0, + "step": 10091 + }, + { + "epoch": 1.10828025477707, + "grad_norm": 2.3544607162475586, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7367974519729614, + "num_tokens": 257413870.0, + "step": 10092 + }, + { + "epoch": 1.1083900724796838, + "grad_norm": 2.0087571144104004, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7258018255233765, + "num_tokens": 257441445.0, + "step": 10093 + }, + { + "epoch": 1.1084998901822973, + "grad_norm": 2.0921096801757812, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.7014831304550171, + "num_tokens": 257471039.0, + "step": 10094 + }, + { + "epoch": 1.108609707884911, + "grad_norm": 2.488042116165161, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7102193236351013, + "num_tokens": 257495190.0, + "step": 10095 + }, + { + "epoch": 1.1087195255875246, + "grad_norm": 2.2031033039093018, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7188160419464111, + "num_tokens": 257521328.0, + "step": 10096 + }, + { + "epoch": 1.1088293432901384, + "grad_norm": 2.38494873046875, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7337074279785156, + "num_tokens": 257543563.0, + "step": 10097 + }, + { + "epoch": 1.108939160992752, + "grad_norm": 2.185168981552124, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.7007085084915161, + "num_tokens": 257572492.0, + "step": 10098 + }, + { + "epoch": 1.1090489786953657, + "grad_norm": 2.216979742050171, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7265733480453491, + "num_tokens": 257598844.0, + "step": 10099 + }, + { + "epoch": 1.1091587963979794, + "grad_norm": 2.2821199893951416, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.720594584941864, + "num_tokens": 257623321.0, + "step": 10100 + }, + { + "epoch": 1.109268614100593, + "grad_norm": 2.215639114379883, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.708892285823822, + "num_tokens": 257648243.0, + "step": 10101 + }, + { + "epoch": 1.1093784318032067, + "grad_norm": 2.025294780731201, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6879333853721619, + "num_tokens": 257680094.0, + "step": 10102 + }, + { + "epoch": 1.1094882495058203, + "grad_norm": 2.2049806118011475, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.707939624786377, + "num_tokens": 257707238.0, + "step": 10103 + }, + { + "epoch": 1.109598067208434, + "grad_norm": 2.342271566390991, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7474455833435059, + "num_tokens": 257730201.0, + "step": 10104 + }, + { + "epoch": 1.1097078849110478, + "grad_norm": 2.663557529449463, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7079194784164429, + "num_tokens": 257751502.0, + "step": 10105 + }, + { + "epoch": 1.1098177026136613, + "grad_norm": 2.2034685611724854, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.696099042892456, + "num_tokens": 257777946.0, + "step": 10106 + }, + { + "epoch": 1.109927520316275, + "grad_norm": 2.262456178665161, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7271270155906677, + "num_tokens": 257801976.0, + "step": 10107 + }, + { + "epoch": 1.1100373380188886, + "grad_norm": 2.3631317615509033, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7111831903457642, + "num_tokens": 257825449.0, + "step": 10108 + }, + { + "epoch": 1.1101471557215024, + "grad_norm": 2.4729843139648438, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7071502208709717, + "num_tokens": 257853925.0, + "step": 10109 + }, + { + "epoch": 1.110256973424116, + "grad_norm": 2.2106330394744873, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7088323831558228, + "num_tokens": 257882961.0, + "step": 10110 + }, + { + "epoch": 1.1103667911267296, + "grad_norm": 2.6759982109069824, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7391670942306519, + "num_tokens": 257900718.0, + "step": 10111 + }, + { + "epoch": 1.1104766088293432, + "grad_norm": 2.2142627239227295, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7294740676879883, + "num_tokens": 257927355.0, + "step": 10112 + }, + { + "epoch": 1.110586426531957, + "grad_norm": 2.433467388153076, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7242468595504761, + "num_tokens": 257949904.0, + "step": 10113 + }, + { + "epoch": 1.1106962442345707, + "grad_norm": 2.2407703399658203, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7120692729949951, + "num_tokens": 257975861.0, + "step": 10114 + }, + { + "epoch": 1.1108060619371842, + "grad_norm": 2.412426233291626, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.727258563041687, + "num_tokens": 257999160.0, + "step": 10115 + }, + { + "epoch": 1.110915879639798, + "grad_norm": 2.459022283554077, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7214697003364563, + "num_tokens": 258021064.0, + "step": 10116 + }, + { + "epoch": 1.1110256973424115, + "grad_norm": 2.037226676940918, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7265728116035461, + "num_tokens": 258051381.0, + "step": 10117 + }, + { + "epoch": 1.1111355150450253, + "grad_norm": 2.2631778717041016, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7359085083007812, + "num_tokens": 258076563.0, + "step": 10118 + }, + { + "epoch": 1.111245332747639, + "grad_norm": 2.119588851928711, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7280287146568298, + "num_tokens": 258103450.0, + "step": 10119 + }, + { + "epoch": 1.1113551504502526, + "grad_norm": 2.088676691055298, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7355989217758179, + "num_tokens": 258133185.0, + "step": 10120 + }, + { + "epoch": 1.1114649681528663, + "grad_norm": 2.0274951457977295, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7316001653671265, + "num_tokens": 258161812.0, + "step": 10121 + }, + { + "epoch": 1.1115747858554799, + "grad_norm": 2.3044047355651855, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.6985906958580017, + "num_tokens": 258187122.0, + "step": 10122 + }, + { + "epoch": 1.1116846035580936, + "grad_norm": 2.4392588138580322, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7455527782440186, + "num_tokens": 258207683.0, + "step": 10123 + }, + { + "epoch": 1.1117944212607072, + "grad_norm": 2.1566290855407715, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.697690486907959, + "num_tokens": 258235550.0, + "step": 10124 + }, + { + "epoch": 1.111904238963321, + "grad_norm": 1.9050240516662598, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.717967689037323, + "num_tokens": 258268830.0, + "step": 10125 + }, + { + "epoch": 1.1120140566659344, + "grad_norm": 2.0288898944854736, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7235634326934814, + "num_tokens": 258297305.0, + "step": 10126 + }, + { + "epoch": 1.1121238743685482, + "grad_norm": 2.2348670959472656, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7298094034194946, + "num_tokens": 258322845.0, + "step": 10127 + }, + { + "epoch": 1.112233692071162, + "grad_norm": 2.377615213394165, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7259576320648193, + "num_tokens": 258347118.0, + "step": 10128 + }, + { + "epoch": 1.1123435097737755, + "grad_norm": 2.277523994445801, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.6940684914588928, + "num_tokens": 258376234.0, + "step": 10129 + }, + { + "epoch": 1.1124533274763893, + "grad_norm": 2.2112834453582764, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7330517172813416, + "num_tokens": 258400877.0, + "step": 10130 + }, + { + "epoch": 1.1125631451790028, + "grad_norm": 2.361114025115967, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7185846567153931, + "num_tokens": 258424961.0, + "step": 10131 + }, + { + "epoch": 1.1126729628816165, + "grad_norm": 2.46722674369812, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7238694429397583, + "num_tokens": 258445737.0, + "step": 10132 + }, + { + "epoch": 1.11278278058423, + "grad_norm": 2.455601215362549, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.714816153049469, + "num_tokens": 258466935.0, + "step": 10133 + }, + { + "epoch": 1.1128925982868438, + "grad_norm": 2.4266419410705566, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7124746441841125, + "num_tokens": 258489814.0, + "step": 10134 + }, + { + "epoch": 1.1130024159894576, + "grad_norm": 2.409999370574951, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7338349223136902, + "num_tokens": 258513292.0, + "step": 10135 + }, + { + "epoch": 1.1131122336920711, + "grad_norm": 2.2899868488311768, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.7034924030303955, + "num_tokens": 258538183.0, + "step": 10136 + }, + { + "epoch": 1.1132220513946849, + "grad_norm": 2.450807809829712, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7247143983840942, + "num_tokens": 258560316.0, + "step": 10137 + }, + { + "epoch": 1.1133318690972984, + "grad_norm": 2.0786120891571045, + "learning_rate": 1e-06, + "loss": 1.0588, + "mean_token_accuracy": 0.6880820989608765, + "num_tokens": 258588194.0, + "step": 10138 + }, + { + "epoch": 1.1134416867999122, + "grad_norm": 2.458218574523926, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7256519794464111, + "num_tokens": 258611812.0, + "step": 10139 + }, + { + "epoch": 1.1135515045025257, + "grad_norm": 2.160722255706787, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7363495826721191, + "num_tokens": 258638004.0, + "step": 10140 + }, + { + "epoch": 1.1136613222051395, + "grad_norm": 2.6276233196258545, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7308371663093567, + "num_tokens": 258659777.0, + "step": 10141 + }, + { + "epoch": 1.1137711399077532, + "grad_norm": 2.4258859157562256, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7408141493797302, + "num_tokens": 258681665.0, + "step": 10142 + }, + { + "epoch": 1.1138809576103668, + "grad_norm": 2.5743520259857178, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7195284366607666, + "num_tokens": 258703694.0, + "step": 10143 + }, + { + "epoch": 1.1139907753129805, + "grad_norm": 2.405268430709839, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7310625910758972, + "num_tokens": 258726636.0, + "step": 10144 + }, + { + "epoch": 1.114100593015594, + "grad_norm": 2.747950315475464, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.7493029236793518, + "num_tokens": 258743721.0, + "step": 10145 + }, + { + "epoch": 1.1142104107182078, + "grad_norm": 2.186018466949463, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7096570730209351, + "num_tokens": 258772374.0, + "step": 10146 + }, + { + "epoch": 1.1143202284208213, + "grad_norm": 2.1804749965667725, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7143914699554443, + "num_tokens": 258796402.0, + "step": 10147 + }, + { + "epoch": 1.114430046123435, + "grad_norm": 2.1321089267730713, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.723223865032196, + "num_tokens": 258825243.0, + "step": 10148 + }, + { + "epoch": 1.1145398638260489, + "grad_norm": 2.244891881942749, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7176304459571838, + "num_tokens": 258850404.0, + "step": 10149 + }, + { + "epoch": 1.1146496815286624, + "grad_norm": 2.2730424404144287, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7369235754013062, + "num_tokens": 258873578.0, + "step": 10150 + }, + { + "epoch": 1.1147594992312762, + "grad_norm": 2.252105951309204, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.6895595788955688, + "num_tokens": 258899765.0, + "step": 10151 + }, + { + "epoch": 1.1148693169338897, + "grad_norm": 2.251905679702759, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7207288146018982, + "num_tokens": 258927666.0, + "step": 10152 + }, + { + "epoch": 1.1149791346365034, + "grad_norm": 2.477658271789551, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7232864499092102, + "num_tokens": 258950672.0, + "step": 10153 + }, + { + "epoch": 1.115088952339117, + "grad_norm": 2.184868574142456, + "learning_rate": 1e-06, + "loss": 1.0527, + "mean_token_accuracy": 0.6893004179000854, + "num_tokens": 258979610.0, + "step": 10154 + }, + { + "epoch": 1.1151987700417307, + "grad_norm": 2.2498316764831543, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7131147384643555, + "num_tokens": 259005311.0, + "step": 10155 + }, + { + "epoch": 1.1153085877443445, + "grad_norm": 2.2536230087280273, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7050613760948181, + "num_tokens": 259033081.0, + "step": 10156 + }, + { + "epoch": 1.115418405446958, + "grad_norm": 2.3660781383514404, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7276503443717957, + "num_tokens": 259055505.0, + "step": 10157 + }, + { + "epoch": 1.1155282231495718, + "grad_norm": 2.376556396484375, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.732642412185669, + "num_tokens": 259077832.0, + "step": 10158 + }, + { + "epoch": 1.1156380408521853, + "grad_norm": 2.333242416381836, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7246706485748291, + "num_tokens": 259101215.0, + "step": 10159 + }, + { + "epoch": 1.115747858554799, + "grad_norm": 2.0404767990112305, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.710543155670166, + "num_tokens": 259129293.0, + "step": 10160 + }, + { + "epoch": 1.1158576762574126, + "grad_norm": 2.637467622756958, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.715329647064209, + "num_tokens": 259150334.0, + "step": 10161 + }, + { + "epoch": 1.1159674939600264, + "grad_norm": 2.514300584793091, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7217869758605957, + "num_tokens": 259172149.0, + "step": 10162 + }, + { + "epoch": 1.11607731166264, + "grad_norm": 2.254631996154785, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7021118402481079, + "num_tokens": 259199502.0, + "step": 10163 + }, + { + "epoch": 1.1161871293652537, + "grad_norm": 2.0394129753112793, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7075965404510498, + "num_tokens": 259229730.0, + "step": 10164 + }, + { + "epoch": 1.1162969470678674, + "grad_norm": 2.2075932025909424, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7143411636352539, + "num_tokens": 259256063.0, + "step": 10165 + }, + { + "epoch": 1.116406764770481, + "grad_norm": 2.1890475749969482, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7123428583145142, + "num_tokens": 259283548.0, + "step": 10166 + }, + { + "epoch": 1.1165165824730947, + "grad_norm": 2.3615031242370605, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7040256857872009, + "num_tokens": 259304998.0, + "step": 10167 + }, + { + "epoch": 1.1166264001757082, + "grad_norm": 2.416616201400757, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7184646129608154, + "num_tokens": 259328829.0, + "step": 10168 + }, + { + "epoch": 1.116736217878322, + "grad_norm": 2.289156913757324, + "learning_rate": 1e-06, + "loss": 1.0216, + "mean_token_accuracy": 0.6858452558517456, + "num_tokens": 259356359.0, + "step": 10169 + }, + { + "epoch": 1.1168460355809358, + "grad_norm": 2.3175294399261475, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.728723406791687, + "num_tokens": 259383979.0, + "step": 10170 + }, + { + "epoch": 1.1169558532835493, + "grad_norm": 2.1914525032043457, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7109274864196777, + "num_tokens": 259412001.0, + "step": 10171 + }, + { + "epoch": 1.117065670986163, + "grad_norm": 2.6100103855133057, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7427814602851868, + "num_tokens": 259432283.0, + "step": 10172 + }, + { + "epoch": 1.1171754886887766, + "grad_norm": 2.5655226707458496, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7176269292831421, + "num_tokens": 259454755.0, + "step": 10173 + }, + { + "epoch": 1.1172853063913903, + "grad_norm": 2.323315143585205, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7029752731323242, + "num_tokens": 259480538.0, + "step": 10174 + }, + { + "epoch": 1.1173951240940039, + "grad_norm": 2.326523542404175, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7143980264663696, + "num_tokens": 259507324.0, + "step": 10175 + }, + { + "epoch": 1.1175049417966176, + "grad_norm": 2.1535112857818604, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7325295209884644, + "num_tokens": 259534367.0, + "step": 10176 + }, + { + "epoch": 1.1176147594992312, + "grad_norm": 2.518023729324341, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7449512481689453, + "num_tokens": 259555932.0, + "step": 10177 + }, + { + "epoch": 1.117724577201845, + "grad_norm": 2.478307008743286, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7326513528823853, + "num_tokens": 259578328.0, + "step": 10178 + }, + { + "epoch": 1.1178343949044587, + "grad_norm": 2.387054204940796, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7404346466064453, + "num_tokens": 259602705.0, + "step": 10179 + }, + { + "epoch": 1.1179442126070722, + "grad_norm": 2.1983108520507812, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7024530172348022, + "num_tokens": 259631599.0, + "step": 10180 + }, + { + "epoch": 1.118054030309686, + "grad_norm": 2.3674654960632324, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7098631858825684, + "num_tokens": 259658027.0, + "step": 10181 + }, + { + "epoch": 1.1181638480122995, + "grad_norm": 2.4525084495544434, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7135410904884338, + "num_tokens": 259680043.0, + "step": 10182 + }, + { + "epoch": 1.1182736657149133, + "grad_norm": 2.0646796226501465, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7297649383544922, + "num_tokens": 259708922.0, + "step": 10183 + }, + { + "epoch": 1.1183834834175268, + "grad_norm": 2.448216438293457, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7061322927474976, + "num_tokens": 259731321.0, + "step": 10184 + }, + { + "epoch": 1.1184933011201406, + "grad_norm": 2.2933151721954346, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7228142023086548, + "num_tokens": 259754886.0, + "step": 10185 + }, + { + "epoch": 1.1186031188227543, + "grad_norm": 2.40759015083313, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7153545618057251, + "num_tokens": 259779200.0, + "step": 10186 + }, + { + "epoch": 1.1187129365253679, + "grad_norm": 2.3650312423706055, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.6927871704101562, + "num_tokens": 259804952.0, + "step": 10187 + }, + { + "epoch": 1.1188227542279816, + "grad_norm": 2.6060521602630615, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7320019006729126, + "num_tokens": 259825351.0, + "step": 10188 + }, + { + "epoch": 1.1189325719305951, + "grad_norm": 2.098419427871704, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7056558132171631, + "num_tokens": 259853661.0, + "step": 10189 + }, + { + "epoch": 1.119042389633209, + "grad_norm": 2.289055824279785, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7164994478225708, + "num_tokens": 259877536.0, + "step": 10190 + }, + { + "epoch": 1.1191522073358224, + "grad_norm": 2.056715250015259, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7261296510696411, + "num_tokens": 259905402.0, + "step": 10191 + }, + { + "epoch": 1.1192620250384362, + "grad_norm": 2.412565231323242, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7205637097358704, + "num_tokens": 259926203.0, + "step": 10192 + }, + { + "epoch": 1.11937184274105, + "grad_norm": 2.484527826309204, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7401643991470337, + "num_tokens": 259948077.0, + "step": 10193 + }, + { + "epoch": 1.1194816604436635, + "grad_norm": 2.335343360900879, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7240347266197205, + "num_tokens": 259971185.0, + "step": 10194 + }, + { + "epoch": 1.1195914781462772, + "grad_norm": 2.1383533477783203, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.6988821625709534, + "num_tokens": 259999880.0, + "step": 10195 + }, + { + "epoch": 1.1197012958488908, + "grad_norm": 2.412667751312256, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7173406481742859, + "num_tokens": 260022175.0, + "step": 10196 + }, + { + "epoch": 1.1198111135515045, + "grad_norm": 2.4620630741119385, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7328224778175354, + "num_tokens": 260045919.0, + "step": 10197 + }, + { + "epoch": 1.119920931254118, + "grad_norm": 2.206251859664917, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7270017862319946, + "num_tokens": 260072295.0, + "step": 10198 + }, + { + "epoch": 1.1200307489567318, + "grad_norm": 2.217778444290161, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7042416334152222, + "num_tokens": 260100917.0, + "step": 10199 + }, + { + "epoch": 1.1201405666593456, + "grad_norm": 2.4646618366241455, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7404419779777527, + "num_tokens": 260121675.0, + "step": 10200 + }, + { + "epoch": 1.1202503843619591, + "grad_norm": 1.9691221714019775, + "learning_rate": 1e-06, + "loss": 1.0222, + "mean_token_accuracy": 0.6834285259246826, + "num_tokens": 260155486.0, + "step": 10201 + }, + { + "epoch": 1.1203602020645729, + "grad_norm": 2.1786131858825684, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.705796480178833, + "num_tokens": 260183990.0, + "step": 10202 + }, + { + "epoch": 1.1204700197671864, + "grad_norm": 2.363079309463501, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7085848450660706, + "num_tokens": 260211410.0, + "step": 10203 + }, + { + "epoch": 1.1205798374698002, + "grad_norm": 2.3561244010925293, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7244510650634766, + "num_tokens": 260237254.0, + "step": 10204 + }, + { + "epoch": 1.1206896551724137, + "grad_norm": 2.1269123554229736, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6994193196296692, + "num_tokens": 260266506.0, + "step": 10205 + }, + { + "epoch": 1.1207994728750275, + "grad_norm": 2.123081684112549, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7094835042953491, + "num_tokens": 260296205.0, + "step": 10206 + }, + { + "epoch": 1.1209092905776412, + "grad_norm": 2.554678440093994, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7436727285385132, + "num_tokens": 260316434.0, + "step": 10207 + }, + { + "epoch": 1.1210191082802548, + "grad_norm": 2.396501302719116, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.7017689943313599, + "num_tokens": 260340376.0, + "step": 10208 + }, + { + "epoch": 1.1211289259828685, + "grad_norm": 2.2208194732666016, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7106953859329224, + "num_tokens": 260368238.0, + "step": 10209 + }, + { + "epoch": 1.121238743685482, + "grad_norm": 2.313020944595337, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7173205018043518, + "num_tokens": 260392338.0, + "step": 10210 + }, + { + "epoch": 1.1213485613880958, + "grad_norm": 2.2253711223602295, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7207192182540894, + "num_tokens": 260418244.0, + "step": 10211 + }, + { + "epoch": 1.1214583790907093, + "grad_norm": 2.1723625659942627, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7126149535179138, + "num_tokens": 260444936.0, + "step": 10212 + }, + { + "epoch": 1.121568196793323, + "grad_norm": 2.326909065246582, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7305800914764404, + "num_tokens": 260467211.0, + "step": 10213 + }, + { + "epoch": 1.1216780144959368, + "grad_norm": 2.0229873657226562, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7112053036689758, + "num_tokens": 260496878.0, + "step": 10214 + }, + { + "epoch": 1.1217878321985504, + "grad_norm": 2.4048216342926025, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7210314869880676, + "num_tokens": 260519776.0, + "step": 10215 + }, + { + "epoch": 1.1218976499011641, + "grad_norm": 2.2194409370422363, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7005748152732849, + "num_tokens": 260544936.0, + "step": 10216 + }, + { + "epoch": 1.1220074676037777, + "grad_norm": 2.415349006652832, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7364543676376343, + "num_tokens": 260565999.0, + "step": 10217 + }, + { + "epoch": 1.1221172853063914, + "grad_norm": 2.164393901824951, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6910386681556702, + "num_tokens": 260596449.0, + "step": 10218 + }, + { + "epoch": 1.122227103009005, + "grad_norm": 2.350572109222412, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.738945722579956, + "num_tokens": 260618919.0, + "step": 10219 + }, + { + "epoch": 1.1223369207116187, + "grad_norm": 2.160814046859741, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7050358057022095, + "num_tokens": 260644894.0, + "step": 10220 + }, + { + "epoch": 1.1224467384142325, + "grad_norm": 2.147583484649658, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7140910029411316, + "num_tokens": 260671117.0, + "step": 10221 + }, + { + "epoch": 1.122556556116846, + "grad_norm": 2.11797833442688, + "learning_rate": 1e-06, + "loss": 1.0468, + "mean_token_accuracy": 0.6800402402877808, + "num_tokens": 260702353.0, + "step": 10222 + }, + { + "epoch": 1.1226663738194598, + "grad_norm": 1.8081578016281128, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7204392552375793, + "num_tokens": 260736805.0, + "step": 10223 + }, + { + "epoch": 1.1227761915220733, + "grad_norm": 2.217787981033325, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7018105983734131, + "num_tokens": 260761553.0, + "step": 10224 + }, + { + "epoch": 1.122886009224687, + "grad_norm": 2.131544351577759, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7089380025863647, + "num_tokens": 260791524.0, + "step": 10225 + }, + { + "epoch": 1.1229958269273006, + "grad_norm": 2.1735897064208984, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.6985865831375122, + "num_tokens": 260822389.0, + "step": 10226 + }, + { + "epoch": 1.1231056446299144, + "grad_norm": 2.5698750019073486, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7240878939628601, + "num_tokens": 260841882.0, + "step": 10227 + }, + { + "epoch": 1.123215462332528, + "grad_norm": 2.1559956073760986, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7244849801063538, + "num_tokens": 260869469.0, + "step": 10228 + }, + { + "epoch": 1.1233252800351416, + "grad_norm": 2.148555040359497, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7181326150894165, + "num_tokens": 260896803.0, + "step": 10229 + }, + { + "epoch": 1.1234350977377554, + "grad_norm": 2.175593852996826, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7243969440460205, + "num_tokens": 260922081.0, + "step": 10230 + }, + { + "epoch": 1.123544915440369, + "grad_norm": 2.151510715484619, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.6929903030395508, + "num_tokens": 260951778.0, + "step": 10231 + }, + { + "epoch": 1.1236547331429827, + "grad_norm": 2.703345775604248, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7220335602760315, + "num_tokens": 260972067.0, + "step": 10232 + }, + { + "epoch": 1.1237645508455962, + "grad_norm": 2.1578547954559326, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7171497344970703, + "num_tokens": 260997704.0, + "step": 10233 + }, + { + "epoch": 1.12387436854821, + "grad_norm": 2.2974376678466797, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.716452419757843, + "num_tokens": 261021779.0, + "step": 10234 + }, + { + "epoch": 1.1239841862508237, + "grad_norm": 2.536266326904297, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7256648540496826, + "num_tokens": 261042964.0, + "step": 10235 + }, + { + "epoch": 1.1240940039534373, + "grad_norm": 2.453963279724121, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7368829250335693, + "num_tokens": 261063596.0, + "step": 10236 + }, + { + "epoch": 1.124203821656051, + "grad_norm": 2.3457624912261963, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7311131954193115, + "num_tokens": 261087015.0, + "step": 10237 + }, + { + "epoch": 1.1243136393586646, + "grad_norm": 2.3605897426605225, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7158375978469849, + "num_tokens": 261110065.0, + "step": 10238 + }, + { + "epoch": 1.1244234570612783, + "grad_norm": 2.259340763092041, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6979601383209229, + "num_tokens": 261135567.0, + "step": 10239 + }, + { + "epoch": 1.1245332747638919, + "grad_norm": 2.1500537395477295, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7041257619857788, + "num_tokens": 261165412.0, + "step": 10240 + }, + { + "epoch": 1.1246430924665056, + "grad_norm": 2.0769755840301514, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7163582444190979, + "num_tokens": 261193936.0, + "step": 10241 + }, + { + "epoch": 1.1247529101691192, + "grad_norm": 2.169051170349121, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7106984853744507, + "num_tokens": 261223935.0, + "step": 10242 + }, + { + "epoch": 1.124862727871733, + "grad_norm": 2.169386386871338, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7303985357284546, + "num_tokens": 261251263.0, + "step": 10243 + }, + { + "epoch": 1.1249725455743467, + "grad_norm": 2.256542682647705, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7238978147506714, + "num_tokens": 261276144.0, + "step": 10244 + }, + { + "epoch": 1.1250823632769602, + "grad_norm": 2.4240357875823975, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7106219530105591, + "num_tokens": 261299470.0, + "step": 10245 + }, + { + "epoch": 1.125192180979574, + "grad_norm": 2.2265241146087646, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6867260932922363, + "num_tokens": 261329509.0, + "step": 10246 + }, + { + "epoch": 1.1253019986821875, + "grad_norm": 2.484922170639038, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7267838716506958, + "num_tokens": 261351372.0, + "step": 10247 + }, + { + "epoch": 1.1254118163848013, + "grad_norm": 2.200028657913208, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.716643214225769, + "num_tokens": 261377555.0, + "step": 10248 + }, + { + "epoch": 1.125521634087415, + "grad_norm": 2.2697196006774902, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7207345962524414, + "num_tokens": 261404323.0, + "step": 10249 + }, + { + "epoch": 1.1256314517900285, + "grad_norm": 2.4021871089935303, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7327491044998169, + "num_tokens": 261426047.0, + "step": 10250 + }, + { + "epoch": 1.1257412694926423, + "grad_norm": 2.2317144870758057, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.7007778882980347, + "num_tokens": 261452396.0, + "step": 10251 + }, + { + "epoch": 1.1258510871952558, + "grad_norm": 2.1834940910339355, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7178457975387573, + "num_tokens": 261479973.0, + "step": 10252 + }, + { + "epoch": 1.1259609048978696, + "grad_norm": 2.067497491836548, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7296563386917114, + "num_tokens": 261507691.0, + "step": 10253 + }, + { + "epoch": 1.1260707226004831, + "grad_norm": 2.2841837406158447, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7199798822402954, + "num_tokens": 261532467.0, + "step": 10254 + }, + { + "epoch": 1.1261805403030969, + "grad_norm": 2.129054069519043, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7101027965545654, + "num_tokens": 261561790.0, + "step": 10255 + }, + { + "epoch": 1.1262903580057104, + "grad_norm": 2.501976490020752, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.7429112792015076, + "num_tokens": 261583278.0, + "step": 10256 + }, + { + "epoch": 1.1264001757083242, + "grad_norm": 2.2584471702575684, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7127106785774231, + "num_tokens": 261609807.0, + "step": 10257 + }, + { + "epoch": 1.126509993410938, + "grad_norm": 2.2357280254364014, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.714409351348877, + "num_tokens": 261636122.0, + "step": 10258 + }, + { + "epoch": 1.1266198111135515, + "grad_norm": 2.153184652328491, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.690447211265564, + "num_tokens": 261666181.0, + "step": 10259 + }, + { + "epoch": 1.1267296288161652, + "grad_norm": 2.244353771209717, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7071830630302429, + "num_tokens": 261692570.0, + "step": 10260 + }, + { + "epoch": 1.1268394465187788, + "grad_norm": 2.2159171104431152, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7138347625732422, + "num_tokens": 261719966.0, + "step": 10261 + }, + { + "epoch": 1.1269492642213925, + "grad_norm": 2.311659336090088, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7305254340171814, + "num_tokens": 261743931.0, + "step": 10262 + }, + { + "epoch": 1.127059081924006, + "grad_norm": 2.5911648273468018, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7276257276535034, + "num_tokens": 261763620.0, + "step": 10263 + }, + { + "epoch": 1.1271688996266198, + "grad_norm": 2.501225233078003, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7226804494857788, + "num_tokens": 261785783.0, + "step": 10264 + }, + { + "epoch": 1.1272787173292333, + "grad_norm": 2.2283976078033447, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7054529190063477, + "num_tokens": 261812391.0, + "step": 10265 + }, + { + "epoch": 1.127388535031847, + "grad_norm": 2.1353986263275146, + "learning_rate": 1e-06, + "loss": 0.7755, + "mean_token_accuracy": 0.7508044242858887, + "num_tokens": 261837164.0, + "step": 10266 + }, + { + "epoch": 1.1274983527344609, + "grad_norm": 2.2148070335388184, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7096001505851746, + "num_tokens": 261864409.0, + "step": 10267 + }, + { + "epoch": 1.1276081704370744, + "grad_norm": 2.5402562618255615, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7351117134094238, + "num_tokens": 261884774.0, + "step": 10268 + }, + { + "epoch": 1.1277179881396882, + "grad_norm": 2.633009195327759, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7143682837486267, + "num_tokens": 261903772.0, + "step": 10269 + }, + { + "epoch": 1.1278278058423017, + "grad_norm": 2.41737699508667, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7224030494689941, + "num_tokens": 261926951.0, + "step": 10270 + }, + { + "epoch": 1.1279376235449154, + "grad_norm": 2.2298946380615234, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7020630240440369, + "num_tokens": 261952269.0, + "step": 10271 + }, + { + "epoch": 1.1280474412475292, + "grad_norm": 2.1429295539855957, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7158514261245728, + "num_tokens": 261982539.0, + "step": 10272 + }, + { + "epoch": 1.1281572589501427, + "grad_norm": 2.2656450271606445, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7155802249908447, + "num_tokens": 262006885.0, + "step": 10273 + }, + { + "epoch": 1.1282670766527565, + "grad_norm": 2.407926082611084, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7085742354393005, + "num_tokens": 262030826.0, + "step": 10274 + }, + { + "epoch": 1.12837689435537, + "grad_norm": 2.2502822875976562, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7248634099960327, + "num_tokens": 262055755.0, + "step": 10275 + }, + { + "epoch": 1.1284867120579838, + "grad_norm": 2.0034985542297363, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7253813743591309, + "num_tokens": 262084353.0, + "step": 10276 + }, + { + "epoch": 1.1285965297605973, + "grad_norm": 2.696131706237793, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7210978269577026, + "num_tokens": 262103165.0, + "step": 10277 + }, + { + "epoch": 1.128706347463211, + "grad_norm": 2.5664963722229004, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7092677354812622, + "num_tokens": 262124797.0, + "step": 10278 + }, + { + "epoch": 1.1288161651658246, + "grad_norm": 2.171910524368286, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6961266994476318, + "num_tokens": 262153369.0, + "step": 10279 + }, + { + "epoch": 1.1289259828684384, + "grad_norm": 2.359912633895874, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7177180051803589, + "num_tokens": 262178199.0, + "step": 10280 + }, + { + "epoch": 1.1290358005710521, + "grad_norm": 2.0706803798675537, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7170782089233398, + "num_tokens": 262206335.0, + "step": 10281 + }, + { + "epoch": 1.1291456182736657, + "grad_norm": 2.531454563140869, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7215481400489807, + "num_tokens": 262228170.0, + "step": 10282 + }, + { + "epoch": 1.1292554359762794, + "grad_norm": 2.0990772247314453, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.701808512210846, + "num_tokens": 262256785.0, + "step": 10283 + }, + { + "epoch": 1.129365253678893, + "grad_norm": 2.1030917167663574, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6942366361618042, + "num_tokens": 262287055.0, + "step": 10284 + }, + { + "epoch": 1.1294750713815067, + "grad_norm": 2.0381734371185303, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7264564037322998, + "num_tokens": 262318892.0, + "step": 10285 + }, + { + "epoch": 1.1295848890841205, + "grad_norm": 2.3826262950897217, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7155377268791199, + "num_tokens": 262343210.0, + "step": 10286 + }, + { + "epoch": 1.129694706786734, + "grad_norm": 2.5000710487365723, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7369123697280884, + "num_tokens": 262363971.0, + "step": 10287 + }, + { + "epoch": 1.1298045244893478, + "grad_norm": 2.2374775409698486, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7315088510513306, + "num_tokens": 262388824.0, + "step": 10288 + }, + { + "epoch": 1.1299143421919613, + "grad_norm": 2.3126916885375977, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7296910285949707, + "num_tokens": 262412977.0, + "step": 10289 + }, + { + "epoch": 1.130024159894575, + "grad_norm": 2.044736623764038, + "learning_rate": 1e-06, + "loss": 1.037, + "mean_token_accuracy": 0.6847419142723083, + "num_tokens": 262442501.0, + "step": 10290 + }, + { + "epoch": 1.1301339775971886, + "grad_norm": 2.36938214302063, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7195390462875366, + "num_tokens": 262464928.0, + "step": 10291 + }, + { + "epoch": 1.1302437952998023, + "grad_norm": 2.087228775024414, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7195099592208862, + "num_tokens": 262493819.0, + "step": 10292 + }, + { + "epoch": 1.1303536130024159, + "grad_norm": 2.2221944332122803, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7176088094711304, + "num_tokens": 262521204.0, + "step": 10293 + }, + { + "epoch": 1.1304634307050296, + "grad_norm": 2.631091833114624, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7069642543792725, + "num_tokens": 262541948.0, + "step": 10294 + }, + { + "epoch": 1.1305732484076434, + "grad_norm": 2.464282751083374, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7234342098236084, + "num_tokens": 262563630.0, + "step": 10295 + }, + { + "epoch": 1.130683066110257, + "grad_norm": 2.0055229663848877, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7099440097808838, + "num_tokens": 262595495.0, + "step": 10296 + }, + { + "epoch": 1.1307928838128707, + "grad_norm": 2.288144588470459, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7377413511276245, + "num_tokens": 262618398.0, + "step": 10297 + }, + { + "epoch": 1.1309027015154842, + "grad_norm": 2.3740234375, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7067283987998962, + "num_tokens": 262640489.0, + "step": 10298 + }, + { + "epoch": 1.131012519218098, + "grad_norm": 2.3550736904144287, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7126631736755371, + "num_tokens": 262665582.0, + "step": 10299 + }, + { + "epoch": 1.1311223369207117, + "grad_norm": 2.3396036624908447, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7031323909759521, + "num_tokens": 262689111.0, + "step": 10300 + }, + { + "epoch": 1.1312321546233253, + "grad_norm": 2.2906253337860107, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7043379545211792, + "num_tokens": 262714428.0, + "step": 10301 + }, + { + "epoch": 1.131341972325939, + "grad_norm": 2.2378625869750977, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7200122475624084, + "num_tokens": 262739050.0, + "step": 10302 + }, + { + "epoch": 1.1314517900285526, + "grad_norm": 2.306605100631714, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7057275772094727, + "num_tokens": 262766015.0, + "step": 10303 + }, + { + "epoch": 1.1315616077311663, + "grad_norm": 2.550615072250366, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7211482524871826, + "num_tokens": 262787733.0, + "step": 10304 + }, + { + "epoch": 1.1316714254337799, + "grad_norm": 2.333085060119629, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.730265736579895, + "num_tokens": 262810638.0, + "step": 10305 + }, + { + "epoch": 1.1317812431363936, + "grad_norm": 1.9721910953521729, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7214105129241943, + "num_tokens": 262844043.0, + "step": 10306 + }, + { + "epoch": 1.1318910608390071, + "grad_norm": 2.1509974002838135, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.708672285079956, + "num_tokens": 262872123.0, + "step": 10307 + }, + { + "epoch": 1.132000878541621, + "grad_norm": 2.631868362426758, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.743918240070343, + "num_tokens": 262891014.0, + "step": 10308 + }, + { + "epoch": 1.1321106962442347, + "grad_norm": 2.124342441558838, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7085354924201965, + "num_tokens": 262918970.0, + "step": 10309 + }, + { + "epoch": 1.1322205139468482, + "grad_norm": 2.0634052753448486, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7372384667396545, + "num_tokens": 262945488.0, + "step": 10310 + }, + { + "epoch": 1.132330331649462, + "grad_norm": 2.1787750720977783, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7085293531417847, + "num_tokens": 262973589.0, + "step": 10311 + }, + { + "epoch": 1.1324401493520755, + "grad_norm": 2.485910415649414, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7255969047546387, + "num_tokens": 262996022.0, + "step": 10312 + }, + { + "epoch": 1.1325499670546892, + "grad_norm": 2.1174137592315674, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.7045209407806396, + "num_tokens": 263025217.0, + "step": 10313 + }, + { + "epoch": 1.132659784757303, + "grad_norm": 2.8307266235351562, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7206635475158691, + "num_tokens": 263043876.0, + "step": 10314 + }, + { + "epoch": 1.1327696024599165, + "grad_norm": 2.7642312049865723, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7100602388381958, + "num_tokens": 263063495.0, + "step": 10315 + }, + { + "epoch": 1.1328794201625303, + "grad_norm": 2.1697914600372314, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7215577363967896, + "num_tokens": 263090451.0, + "step": 10316 + }, + { + "epoch": 1.1329892378651438, + "grad_norm": 2.306760311126709, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7131376266479492, + "num_tokens": 263114857.0, + "step": 10317 + }, + { + "epoch": 1.1330990555677576, + "grad_norm": 2.028168201446533, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.7088022828102112, + "num_tokens": 263144246.0, + "step": 10318 + }, + { + "epoch": 1.1332088732703711, + "grad_norm": 2.2441885471343994, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7123239636421204, + "num_tokens": 263170722.0, + "step": 10319 + }, + { + "epoch": 1.1333186909729849, + "grad_norm": 2.1907010078430176, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7365169525146484, + "num_tokens": 263196229.0, + "step": 10320 + }, + { + "epoch": 1.1334285086755984, + "grad_norm": 2.3765742778778076, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.708154559135437, + "num_tokens": 263223219.0, + "step": 10321 + }, + { + "epoch": 1.1335383263782122, + "grad_norm": 2.736706018447876, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7253255844116211, + "num_tokens": 263240714.0, + "step": 10322 + }, + { + "epoch": 1.133648144080826, + "grad_norm": 2.021940231323242, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7118970155715942, + "num_tokens": 263270940.0, + "step": 10323 + }, + { + "epoch": 1.1337579617834395, + "grad_norm": 2.537081003189087, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7110700011253357, + "num_tokens": 263293407.0, + "step": 10324 + }, + { + "epoch": 1.1338677794860532, + "grad_norm": 2.147400379180908, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6997663974761963, + "num_tokens": 263325567.0, + "step": 10325 + }, + { + "epoch": 1.1339775971886668, + "grad_norm": 2.3438282012939453, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7012500166893005, + "num_tokens": 263351806.0, + "step": 10326 + }, + { + "epoch": 1.1340874148912805, + "grad_norm": 2.4090046882629395, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7108557820320129, + "num_tokens": 263376156.0, + "step": 10327 + }, + { + "epoch": 1.134197232593894, + "grad_norm": 2.408565044403076, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7235016226768494, + "num_tokens": 263398613.0, + "step": 10328 + }, + { + "epoch": 1.1343070502965078, + "grad_norm": 2.1947519779205322, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.7002118825912476, + "num_tokens": 263427554.0, + "step": 10329 + }, + { + "epoch": 1.1344168679991213, + "grad_norm": 2.4603891372680664, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7276973128318787, + "num_tokens": 263449329.0, + "step": 10330 + }, + { + "epoch": 1.134526685701735, + "grad_norm": 2.212956666946411, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7237112522125244, + "num_tokens": 263474194.0, + "step": 10331 + }, + { + "epoch": 1.1346365034043489, + "grad_norm": 2.208743095397949, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.712855339050293, + "num_tokens": 263501200.0, + "step": 10332 + }, + { + "epoch": 1.1347463211069624, + "grad_norm": 2.315225601196289, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7238630056381226, + "num_tokens": 263524883.0, + "step": 10333 + }, + { + "epoch": 1.1348561388095761, + "grad_norm": 2.2684738636016846, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7038365006446838, + "num_tokens": 263551581.0, + "step": 10334 + }, + { + "epoch": 1.1349659565121897, + "grad_norm": 2.15205979347229, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7004508376121521, + "num_tokens": 263582557.0, + "step": 10335 + }, + { + "epoch": 1.1350757742148034, + "grad_norm": 2.219244956970215, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7131608724594116, + "num_tokens": 263608869.0, + "step": 10336 + }, + { + "epoch": 1.1351855919174172, + "grad_norm": 2.1147353649139404, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7372751832008362, + "num_tokens": 263636706.0, + "step": 10337 + }, + { + "epoch": 1.1352954096200307, + "grad_norm": 2.1871273517608643, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7170907855033875, + "num_tokens": 263664205.0, + "step": 10338 + }, + { + "epoch": 1.1354052273226445, + "grad_norm": 2.573064088821411, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7378247380256653, + "num_tokens": 263683666.0, + "step": 10339 + }, + { + "epoch": 1.135515045025258, + "grad_norm": 2.2730839252471924, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7027281522750854, + "num_tokens": 263709328.0, + "step": 10340 + }, + { + "epoch": 1.1356248627278718, + "grad_norm": 2.0874407291412354, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6947058439254761, + "num_tokens": 263738849.0, + "step": 10341 + }, + { + "epoch": 1.1357346804304853, + "grad_norm": 2.713632345199585, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7323474287986755, + "num_tokens": 263758993.0, + "step": 10342 + }, + { + "epoch": 1.135844498133099, + "grad_norm": 2.0182101726531982, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7171951532363892, + "num_tokens": 263790082.0, + "step": 10343 + }, + { + "epoch": 1.1359543158357126, + "grad_norm": 2.652776002883911, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7041513919830322, + "num_tokens": 263814752.0, + "step": 10344 + }, + { + "epoch": 1.1360641335383264, + "grad_norm": 2.219169855117798, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7192145586013794, + "num_tokens": 263840749.0, + "step": 10345 + }, + { + "epoch": 1.1361739512409401, + "grad_norm": 2.169564723968506, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7202683687210083, + "num_tokens": 263866127.0, + "step": 10346 + }, + { + "epoch": 1.1362837689435537, + "grad_norm": 2.341498851776123, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.730377733707428, + "num_tokens": 263888740.0, + "step": 10347 + }, + { + "epoch": 1.1363935866461674, + "grad_norm": 2.129122495651245, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6951074004173279, + "num_tokens": 263919530.0, + "step": 10348 + }, + { + "epoch": 1.136503404348781, + "grad_norm": 2.3757176399230957, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.750452995300293, + "num_tokens": 263941099.0, + "step": 10349 + }, + { + "epoch": 1.1366132220513947, + "grad_norm": 2.4285571575164795, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7172829508781433, + "num_tokens": 263964172.0, + "step": 10350 + }, + { + "epoch": 1.1367230397540085, + "grad_norm": 2.487475633621216, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7249642610549927, + "num_tokens": 263984592.0, + "step": 10351 + }, + { + "epoch": 1.136832857456622, + "grad_norm": 2.3505890369415283, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7212094664573669, + "num_tokens": 264008780.0, + "step": 10352 + }, + { + "epoch": 1.1369426751592357, + "grad_norm": 2.2357230186462402, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7147926092147827, + "num_tokens": 264033764.0, + "step": 10353 + }, + { + "epoch": 1.1370524928618493, + "grad_norm": 2.4935834407806396, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.6983926296234131, + "num_tokens": 264057028.0, + "step": 10354 + }, + { + "epoch": 1.137162310564463, + "grad_norm": 2.2000205516815186, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7375999689102173, + "num_tokens": 264080583.0, + "step": 10355 + }, + { + "epoch": 1.1372721282670766, + "grad_norm": 2.3115832805633545, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7350598573684692, + "num_tokens": 264103580.0, + "step": 10356 + }, + { + "epoch": 1.1373819459696903, + "grad_norm": 1.9569413661956787, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.731110692024231, + "num_tokens": 264133802.0, + "step": 10357 + }, + { + "epoch": 1.1374917636723039, + "grad_norm": 2.2902910709381104, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.715864360332489, + "num_tokens": 264157658.0, + "step": 10358 + }, + { + "epoch": 1.1376015813749176, + "grad_norm": 2.036508798599243, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7102088928222656, + "num_tokens": 264187200.0, + "step": 10359 + }, + { + "epoch": 1.1377113990775314, + "grad_norm": 2.110269069671631, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7102659940719604, + "num_tokens": 264216529.0, + "step": 10360 + }, + { + "epoch": 1.137821216780145, + "grad_norm": 2.2224128246307373, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.6995425820350647, + "num_tokens": 264243289.0, + "step": 10361 + }, + { + "epoch": 1.1379310344827587, + "grad_norm": 2.394695520401001, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7373225688934326, + "num_tokens": 264266566.0, + "step": 10362 + }, + { + "epoch": 1.1380408521853722, + "grad_norm": 2.6430721282958984, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7269237041473389, + "num_tokens": 264285064.0, + "step": 10363 + }, + { + "epoch": 1.138150669887986, + "grad_norm": 2.200148105621338, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7201958894729614, + "num_tokens": 264309955.0, + "step": 10364 + }, + { + "epoch": 1.1382604875905997, + "grad_norm": 2.107801914215088, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7091010808944702, + "num_tokens": 264339702.0, + "step": 10365 + }, + { + "epoch": 1.1383703052932133, + "grad_norm": 2.3162779808044434, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7176090478897095, + "num_tokens": 264363531.0, + "step": 10366 + }, + { + "epoch": 1.138480122995827, + "grad_norm": 1.9395966529846191, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7173950672149658, + "num_tokens": 264396395.0, + "step": 10367 + }, + { + "epoch": 1.1385899406984406, + "grad_norm": 2.1900923252105713, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7109066247940063, + "num_tokens": 264423236.0, + "step": 10368 + }, + { + "epoch": 1.1386997584010543, + "grad_norm": 2.291292667388916, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7095682621002197, + "num_tokens": 264447550.0, + "step": 10369 + }, + { + "epoch": 1.1388095761036678, + "grad_norm": 2.214674711227417, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7259116768836975, + "num_tokens": 264472574.0, + "step": 10370 + }, + { + "epoch": 1.1389193938062816, + "grad_norm": 2.324394702911377, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7113420963287354, + "num_tokens": 264497069.0, + "step": 10371 + }, + { + "epoch": 1.1390292115088951, + "grad_norm": 2.189467430114746, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7338438034057617, + "num_tokens": 264522762.0, + "step": 10372 + }, + { + "epoch": 1.139139029211509, + "grad_norm": 2.067549228668213, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.697015106678009, + "num_tokens": 264552043.0, + "step": 10373 + }, + { + "epoch": 1.1392488469141226, + "grad_norm": 2.131484031677246, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.6916719675064087, + "num_tokens": 264581048.0, + "step": 10374 + }, + { + "epoch": 1.1393586646167362, + "grad_norm": 2.1421992778778076, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7154204845428467, + "num_tokens": 264608931.0, + "step": 10375 + }, + { + "epoch": 1.13946848231935, + "grad_norm": 2.3423306941986084, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7230361700057983, + "num_tokens": 264632869.0, + "step": 10376 + }, + { + "epoch": 1.1395783000219635, + "grad_norm": 2.3454253673553467, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7244175672531128, + "num_tokens": 264655339.0, + "step": 10377 + }, + { + "epoch": 1.1396881177245772, + "grad_norm": 2.0515758991241455, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7043707966804504, + "num_tokens": 264686198.0, + "step": 10378 + }, + { + "epoch": 1.1397979354271908, + "grad_norm": 2.227445363998413, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7619274258613586, + "num_tokens": 264710671.0, + "step": 10379 + }, + { + "epoch": 1.1399077531298045, + "grad_norm": 2.6641900539398193, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7377129793167114, + "num_tokens": 264730041.0, + "step": 10380 + }, + { + "epoch": 1.1400175708324183, + "grad_norm": 2.155170440673828, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7136197090148926, + "num_tokens": 264757470.0, + "step": 10381 + }, + { + "epoch": 1.1401273885350318, + "grad_norm": 2.3027374744415283, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7145359516143799, + "num_tokens": 264783220.0, + "step": 10382 + }, + { + "epoch": 1.1402372062376456, + "grad_norm": 2.3940279483795166, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7130178213119507, + "num_tokens": 264808437.0, + "step": 10383 + }, + { + "epoch": 1.140347023940259, + "grad_norm": 2.026794672012329, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7049069404602051, + "num_tokens": 264840273.0, + "step": 10384 + }, + { + "epoch": 1.1404568416428729, + "grad_norm": 2.514705181121826, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7254447937011719, + "num_tokens": 264861562.0, + "step": 10385 + }, + { + "epoch": 1.1405666593454864, + "grad_norm": 2.6597046852111816, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7425698041915894, + "num_tokens": 264880520.0, + "step": 10386 + }, + { + "epoch": 1.1406764770481002, + "grad_norm": 2.213393449783325, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7162336707115173, + "num_tokens": 264906732.0, + "step": 10387 + }, + { + "epoch": 1.140786294750714, + "grad_norm": 2.183661699295044, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7125890254974365, + "num_tokens": 264932827.0, + "step": 10388 + }, + { + "epoch": 1.1408961124533274, + "grad_norm": 2.3144896030426025, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7358274459838867, + "num_tokens": 264954404.0, + "step": 10389 + }, + { + "epoch": 1.1410059301559412, + "grad_norm": 2.23052716255188, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7369928359985352, + "num_tokens": 264978986.0, + "step": 10390 + }, + { + "epoch": 1.1411157478585547, + "grad_norm": 2.01039981842041, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7060354948043823, + "num_tokens": 265012346.0, + "step": 10391 + }, + { + "epoch": 1.1412255655611685, + "grad_norm": 2.1167843341827393, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7022359371185303, + "num_tokens": 265041317.0, + "step": 10392 + }, + { + "epoch": 1.141335383263782, + "grad_norm": 2.2163591384887695, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.6959604024887085, + "num_tokens": 265071839.0, + "step": 10393 + }, + { + "epoch": 1.1414452009663958, + "grad_norm": 2.4114468097686768, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7333993315696716, + "num_tokens": 265094218.0, + "step": 10394 + }, + { + "epoch": 1.1415550186690093, + "grad_norm": 2.459979295730591, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7182108759880066, + "num_tokens": 265116739.0, + "step": 10395 + }, + { + "epoch": 1.141664836371623, + "grad_norm": 2.064239263534546, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.716923713684082, + "num_tokens": 265146718.0, + "step": 10396 + }, + { + "epoch": 1.1417746540742368, + "grad_norm": 2.436000347137451, + "learning_rate": 1e-06, + "loss": 0.7686, + "mean_token_accuracy": 0.7618299722671509, + "num_tokens": 265167811.0, + "step": 10397 + }, + { + "epoch": 1.1418844717768504, + "grad_norm": 2.184917688369751, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7264449596405029, + "num_tokens": 265194599.0, + "step": 10398 + }, + { + "epoch": 1.1419942894794641, + "grad_norm": 2.222243070602417, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7191499471664429, + "num_tokens": 265219894.0, + "step": 10399 + }, + { + "epoch": 1.1421041071820777, + "grad_norm": 2.2683801651000977, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7117854356765747, + "num_tokens": 265249162.0, + "step": 10400 + }, + { + "epoch": 1.1422139248846914, + "grad_norm": 2.5042216777801514, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7138901948928833, + "num_tokens": 265275487.0, + "step": 10401 + }, + { + "epoch": 1.1423237425873052, + "grad_norm": 2.2155463695526123, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.6914671659469604, + "num_tokens": 265305045.0, + "step": 10402 + }, + { + "epoch": 1.1424335602899187, + "grad_norm": 2.410698890686035, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7001672983169556, + "num_tokens": 265328439.0, + "step": 10403 + }, + { + "epoch": 1.1425433779925325, + "grad_norm": 2.2816925048828125, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7064096331596375, + "num_tokens": 265354222.0, + "step": 10404 + }, + { + "epoch": 1.142653195695146, + "grad_norm": 2.3896031379699707, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7173622846603394, + "num_tokens": 265376452.0, + "step": 10405 + }, + { + "epoch": 1.1427630133977598, + "grad_norm": 2.13498592376709, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7269558906555176, + "num_tokens": 265403267.0, + "step": 10406 + }, + { + "epoch": 1.1428728311003733, + "grad_norm": 2.2223360538482666, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.711616039276123, + "num_tokens": 265431160.0, + "step": 10407 + }, + { + "epoch": 1.142982648802987, + "grad_norm": 2.612283945083618, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7136972546577454, + "num_tokens": 265453731.0, + "step": 10408 + }, + { + "epoch": 1.1430924665056006, + "grad_norm": 2.175684928894043, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7127354145050049, + "num_tokens": 265480675.0, + "step": 10409 + }, + { + "epoch": 1.1432022842082143, + "grad_norm": 2.28609561920166, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7215167284011841, + "num_tokens": 265505939.0, + "step": 10410 + }, + { + "epoch": 1.143312101910828, + "grad_norm": 2.53336501121521, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7181788086891174, + "num_tokens": 265527944.0, + "step": 10411 + }, + { + "epoch": 1.1434219196134416, + "grad_norm": 2.6285736560821533, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7218933701515198, + "num_tokens": 265548670.0, + "step": 10412 + }, + { + "epoch": 1.1435317373160554, + "grad_norm": 2.586717367172241, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7428391575813293, + "num_tokens": 265569034.0, + "step": 10413 + }, + { + "epoch": 1.143641555018669, + "grad_norm": 2.285550355911255, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.734755277633667, + "num_tokens": 265595632.0, + "step": 10414 + }, + { + "epoch": 1.1437513727212827, + "grad_norm": 2.2972981929779053, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7217684388160706, + "num_tokens": 265620015.0, + "step": 10415 + }, + { + "epoch": 1.1438611904238964, + "grad_norm": 2.2754054069519043, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7019596099853516, + "num_tokens": 265648099.0, + "step": 10416 + }, + { + "epoch": 1.14397100812651, + "grad_norm": 2.093235969543457, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7284286618232727, + "num_tokens": 265675741.0, + "step": 10417 + }, + { + "epoch": 1.1440808258291237, + "grad_norm": 2.2258799076080322, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.6971675157546997, + "num_tokens": 265701451.0, + "step": 10418 + }, + { + "epoch": 1.1441906435317373, + "grad_norm": 2.37858247756958, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7376344203948975, + "num_tokens": 265724000.0, + "step": 10419 + }, + { + "epoch": 1.144300461234351, + "grad_norm": 2.0456202030181885, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7010836601257324, + "num_tokens": 265755320.0, + "step": 10420 + }, + { + "epoch": 1.1444102789369646, + "grad_norm": 2.377056360244751, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7223442792892456, + "num_tokens": 265778556.0, + "step": 10421 + }, + { + "epoch": 1.1445200966395783, + "grad_norm": 2.055250406265259, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7159823179244995, + "num_tokens": 265809572.0, + "step": 10422 + }, + { + "epoch": 1.1446299143421919, + "grad_norm": 2.28110408782959, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7203200459480286, + "num_tokens": 265834530.0, + "step": 10423 + }, + { + "epoch": 1.1447397320448056, + "grad_norm": 2.2466557025909424, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7263495922088623, + "num_tokens": 265861158.0, + "step": 10424 + }, + { + "epoch": 1.1448495497474194, + "grad_norm": 2.502265214920044, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7214059233665466, + "num_tokens": 265883447.0, + "step": 10425 + }, + { + "epoch": 1.144959367450033, + "grad_norm": 2.5476348400115967, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7306586503982544, + "num_tokens": 265905842.0, + "step": 10426 + }, + { + "epoch": 1.1450691851526467, + "grad_norm": 2.364725351333618, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7126388549804688, + "num_tokens": 265930703.0, + "step": 10427 + }, + { + "epoch": 1.1451790028552602, + "grad_norm": 2.5926804542541504, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7047343850135803, + "num_tokens": 265950103.0, + "step": 10428 + }, + { + "epoch": 1.145288820557874, + "grad_norm": 2.225687026977539, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7417888641357422, + "num_tokens": 265976162.0, + "step": 10429 + }, + { + "epoch": 1.1453986382604877, + "grad_norm": 2.281691551208496, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7121022939682007, + "num_tokens": 266002236.0, + "step": 10430 + }, + { + "epoch": 1.1455084559631012, + "grad_norm": 2.202181577682495, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7182828187942505, + "num_tokens": 266028596.0, + "step": 10431 + }, + { + "epoch": 1.145618273665715, + "grad_norm": 2.2765448093414307, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7125075459480286, + "num_tokens": 266054467.0, + "step": 10432 + }, + { + "epoch": 1.1457280913683285, + "grad_norm": 2.4776711463928223, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7269935607910156, + "num_tokens": 266075264.0, + "step": 10433 + }, + { + "epoch": 1.1458379090709423, + "grad_norm": 2.4927632808685303, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7366013526916504, + "num_tokens": 266095901.0, + "step": 10434 + }, + { + "epoch": 1.1459477267735558, + "grad_norm": 2.2240686416625977, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7276805639266968, + "num_tokens": 266122308.0, + "step": 10435 + }, + { + "epoch": 1.1460575444761696, + "grad_norm": 2.180697441101074, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7243965864181519, + "num_tokens": 266149022.0, + "step": 10436 + }, + { + "epoch": 1.1461673621787831, + "grad_norm": 2.359856367111206, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7097750902175903, + "num_tokens": 266173160.0, + "step": 10437 + }, + { + "epoch": 1.1462771798813969, + "grad_norm": 2.400513172149658, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7221108675003052, + "num_tokens": 266196619.0, + "step": 10438 + }, + { + "epoch": 1.1463869975840106, + "grad_norm": 2.4030280113220215, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7128891944885254, + "num_tokens": 266219610.0, + "step": 10439 + }, + { + "epoch": 1.1464968152866242, + "grad_norm": 2.392270088195801, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7283159494400024, + "num_tokens": 266242101.0, + "step": 10440 + }, + { + "epoch": 1.146606632989238, + "grad_norm": 2.3769962787628174, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7113777995109558, + "num_tokens": 266267294.0, + "step": 10441 + }, + { + "epoch": 1.1467164506918515, + "grad_norm": 2.465761661529541, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7147787809371948, + "num_tokens": 266291955.0, + "step": 10442 + }, + { + "epoch": 1.1468262683944652, + "grad_norm": 2.369257688522339, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7061678171157837, + "num_tokens": 266315342.0, + "step": 10443 + }, + { + "epoch": 1.1469360860970788, + "grad_norm": 2.162339687347412, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.7539099454879761, + "num_tokens": 266341103.0, + "step": 10444 + }, + { + "epoch": 1.1470459037996925, + "grad_norm": 2.156794309616089, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7388020753860474, + "num_tokens": 266365229.0, + "step": 10445 + }, + { + "epoch": 1.147155721502306, + "grad_norm": 2.1777496337890625, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7138494849205017, + "num_tokens": 266392374.0, + "step": 10446 + }, + { + "epoch": 1.1472655392049198, + "grad_norm": 2.036041498184204, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7264374494552612, + "num_tokens": 266421796.0, + "step": 10447 + }, + { + "epoch": 1.1473753569075336, + "grad_norm": 2.4231414794921875, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7043040990829468, + "num_tokens": 266445212.0, + "step": 10448 + }, + { + "epoch": 1.147485174610147, + "grad_norm": 2.340165376663208, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7289700508117676, + "num_tokens": 266468297.0, + "step": 10449 + }, + { + "epoch": 1.1475949923127609, + "grad_norm": 2.3074543476104736, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7116106748580933, + "num_tokens": 266492936.0, + "step": 10450 + }, + { + "epoch": 1.1477048100153744, + "grad_norm": 2.2247073650360107, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.721028208732605, + "num_tokens": 266518997.0, + "step": 10451 + }, + { + "epoch": 1.1478146277179881, + "grad_norm": 2.1033260822296143, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7039588689804077, + "num_tokens": 266547744.0, + "step": 10452 + }, + { + "epoch": 1.147924445420602, + "grad_norm": 2.660689353942871, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7185527682304382, + "num_tokens": 266567862.0, + "step": 10453 + }, + { + "epoch": 1.1480342631232154, + "grad_norm": 2.392054319381714, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.731855034828186, + "num_tokens": 266592352.0, + "step": 10454 + }, + { + "epoch": 1.1481440808258292, + "grad_norm": 2.41265869140625, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7361522912979126, + "num_tokens": 266615145.0, + "step": 10455 + }, + { + "epoch": 1.1482538985284427, + "grad_norm": 2.221815586090088, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7256304025650024, + "num_tokens": 266640869.0, + "step": 10456 + }, + { + "epoch": 1.1483637162310565, + "grad_norm": 2.140622615814209, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.706013023853302, + "num_tokens": 266671220.0, + "step": 10457 + }, + { + "epoch": 1.14847353393367, + "grad_norm": 2.4015753269195557, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6950671672821045, + "num_tokens": 266695901.0, + "step": 10458 + }, + { + "epoch": 1.1485833516362838, + "grad_norm": 2.1491987705230713, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.6960043907165527, + "num_tokens": 266724854.0, + "step": 10459 + }, + { + "epoch": 1.1486931693388973, + "grad_norm": 2.4148426055908203, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7285844087600708, + "num_tokens": 266748095.0, + "step": 10460 + }, + { + "epoch": 1.148802987041511, + "grad_norm": 2.809908390045166, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7252536416053772, + "num_tokens": 266766830.0, + "step": 10461 + }, + { + "epoch": 1.1489128047441248, + "grad_norm": 2.3338568210601807, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7312766313552856, + "num_tokens": 266788309.0, + "step": 10462 + }, + { + "epoch": 1.1490226224467384, + "grad_norm": 2.275259256362915, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7416194677352905, + "num_tokens": 266812249.0, + "step": 10463 + }, + { + "epoch": 1.1491324401493521, + "grad_norm": 2.30766224861145, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7272608876228333, + "num_tokens": 266836167.0, + "step": 10464 + }, + { + "epoch": 1.1492422578519657, + "grad_norm": 2.4601433277130127, + "learning_rate": 1e-06, + "loss": 0.8269, + "mean_token_accuracy": 0.7412238121032715, + "num_tokens": 266858152.0, + "step": 10465 + }, + { + "epoch": 1.1493520755545794, + "grad_norm": 2.0974106788635254, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7157881855964661, + "num_tokens": 266886147.0, + "step": 10466 + }, + { + "epoch": 1.1494618932571932, + "grad_norm": 2.320915699005127, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.732201874256134, + "num_tokens": 266912268.0, + "step": 10467 + }, + { + "epoch": 1.1495717109598067, + "grad_norm": 2.2889201641082764, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.7376066446304321, + "num_tokens": 266935268.0, + "step": 10468 + }, + { + "epoch": 1.1496815286624205, + "grad_norm": 2.2923922538757324, + "learning_rate": 1e-06, + "loss": 0.7673, + "mean_token_accuracy": 0.7574514150619507, + "num_tokens": 266958756.0, + "step": 10469 + }, + { + "epoch": 1.149791346365034, + "grad_norm": 2.2766952514648438, + "learning_rate": 1e-06, + "loss": 0.831, + "mean_token_accuracy": 0.7358907461166382, + "num_tokens": 266982481.0, + "step": 10470 + }, + { + "epoch": 1.1499011640676478, + "grad_norm": 2.31118106842041, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.720974326133728, + "num_tokens": 267006588.0, + "step": 10471 + }, + { + "epoch": 1.1500109817702613, + "grad_norm": 2.3139288425445557, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7268191576004028, + "num_tokens": 267029853.0, + "step": 10472 + }, + { + "epoch": 1.150120799472875, + "grad_norm": 2.4121146202087402, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7385239601135254, + "num_tokens": 267052210.0, + "step": 10473 + }, + { + "epoch": 1.1502306171754886, + "grad_norm": 2.0476362705230713, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7173947691917419, + "num_tokens": 267082961.0, + "step": 10474 + }, + { + "epoch": 1.1503404348781023, + "grad_norm": 2.320042610168457, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7218193411827087, + "num_tokens": 267108246.0, + "step": 10475 + }, + { + "epoch": 1.150450252580716, + "grad_norm": 2.1499829292297363, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.718788206577301, + "num_tokens": 267134740.0, + "step": 10476 + }, + { + "epoch": 1.1505600702833296, + "grad_norm": 2.18782377243042, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.6879328489303589, + "num_tokens": 267162257.0, + "step": 10477 + }, + { + "epoch": 1.1506698879859434, + "grad_norm": 2.136914014816284, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7368577122688293, + "num_tokens": 267189468.0, + "step": 10478 + }, + { + "epoch": 1.150779705688557, + "grad_norm": 2.193967580795288, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7137152552604675, + "num_tokens": 267216117.0, + "step": 10479 + }, + { + "epoch": 1.1508895233911707, + "grad_norm": 2.194880247116089, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.709613561630249, + "num_tokens": 267242560.0, + "step": 10480 + }, + { + "epoch": 1.1509993410937844, + "grad_norm": 2.110168218612671, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7103176116943359, + "num_tokens": 267273134.0, + "step": 10481 + }, + { + "epoch": 1.151109158796398, + "grad_norm": 2.2848474979400635, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7182019352912903, + "num_tokens": 267297066.0, + "step": 10482 + }, + { + "epoch": 1.1512189764990117, + "grad_norm": 2.3396551609039307, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7336063385009766, + "num_tokens": 267320532.0, + "step": 10483 + }, + { + "epoch": 1.1513287942016253, + "grad_norm": 2.322993755340576, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7195679545402527, + "num_tokens": 267344521.0, + "step": 10484 + }, + { + "epoch": 1.151438611904239, + "grad_norm": 2.1124956607818604, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7434810996055603, + "num_tokens": 267370497.0, + "step": 10485 + }, + { + "epoch": 1.1515484296068526, + "grad_norm": 2.012697458267212, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.7059437036514282, + "num_tokens": 267402817.0, + "step": 10486 + }, + { + "epoch": 1.1516582473094663, + "grad_norm": 2.699394702911377, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7253537774085999, + "num_tokens": 267422137.0, + "step": 10487 + }, + { + "epoch": 1.1517680650120798, + "grad_norm": 2.3513731956481934, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.7007800340652466, + "num_tokens": 267447108.0, + "step": 10488 + }, + { + "epoch": 1.1518778827146936, + "grad_norm": 2.093942403793335, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7249434590339661, + "num_tokens": 267474095.0, + "step": 10489 + }, + { + "epoch": 1.1519877004173074, + "grad_norm": 2.422771453857422, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.69186931848526, + "num_tokens": 267500723.0, + "step": 10490 + }, + { + "epoch": 1.152097518119921, + "grad_norm": 2.535261631011963, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7362738251686096, + "num_tokens": 267520647.0, + "step": 10491 + }, + { + "epoch": 1.1522073358225347, + "grad_norm": 2.1596572399139404, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7098751068115234, + "num_tokens": 267548363.0, + "step": 10492 + }, + { + "epoch": 1.1523171535251482, + "grad_norm": 2.385145664215088, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7152646780014038, + "num_tokens": 267571609.0, + "step": 10493 + }, + { + "epoch": 1.152426971227762, + "grad_norm": 2.277552366256714, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7012265920639038, + "num_tokens": 267596737.0, + "step": 10494 + }, + { + "epoch": 1.1525367889303757, + "grad_norm": 2.2237513065338135, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.719395101070404, + "num_tokens": 267621606.0, + "step": 10495 + }, + { + "epoch": 1.1526466066329892, + "grad_norm": 2.1360466480255127, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7086807489395142, + "num_tokens": 267652555.0, + "step": 10496 + }, + { + "epoch": 1.152756424335603, + "grad_norm": 2.264821767807007, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7129770517349243, + "num_tokens": 267676339.0, + "step": 10497 + }, + { + "epoch": 1.1528662420382165, + "grad_norm": 2.2815446853637695, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7109249830245972, + "num_tokens": 267701085.0, + "step": 10498 + }, + { + "epoch": 1.1529760597408303, + "grad_norm": 2.4715564250946045, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7235264778137207, + "num_tokens": 267723202.0, + "step": 10499 + }, + { + "epoch": 1.1530858774434438, + "grad_norm": 2.3141536712646484, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7285508513450623, + "num_tokens": 267746177.0, + "step": 10500 + }, + { + "epoch": 1.1531956951460576, + "grad_norm": 2.347266435623169, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.710145115852356, + "num_tokens": 267770890.0, + "step": 10501 + }, + { + "epoch": 1.153305512848671, + "grad_norm": 2.1590449810028076, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7144865989685059, + "num_tokens": 267798596.0, + "step": 10502 + }, + { + "epoch": 1.1534153305512849, + "grad_norm": 2.104339122772217, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7156728506088257, + "num_tokens": 267826367.0, + "step": 10503 + }, + { + "epoch": 1.1535251482538986, + "grad_norm": 2.404040575027466, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7182179689407349, + "num_tokens": 267852491.0, + "step": 10504 + }, + { + "epoch": 1.1536349659565122, + "grad_norm": 2.201117992401123, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7259157299995422, + "num_tokens": 267877876.0, + "step": 10505 + }, + { + "epoch": 1.153744783659126, + "grad_norm": 2.2995810508728027, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7442192435264587, + "num_tokens": 267900535.0, + "step": 10506 + }, + { + "epoch": 1.1538546013617395, + "grad_norm": 2.3764708042144775, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7119970917701721, + "num_tokens": 267923338.0, + "step": 10507 + }, + { + "epoch": 1.1539644190643532, + "grad_norm": 2.272170066833496, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7187955379486084, + "num_tokens": 267947683.0, + "step": 10508 + }, + { + "epoch": 1.1540742367669667, + "grad_norm": 2.1224660873413086, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7192153930664062, + "num_tokens": 267974201.0, + "step": 10509 + }, + { + "epoch": 1.1541840544695805, + "grad_norm": 2.558598756790161, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7216875553131104, + "num_tokens": 267994516.0, + "step": 10510 + }, + { + "epoch": 1.154293872172194, + "grad_norm": 2.126702308654785, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7320006489753723, + "num_tokens": 268022457.0, + "step": 10511 + }, + { + "epoch": 1.1544036898748078, + "grad_norm": 2.284860849380493, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7142764925956726, + "num_tokens": 268047281.0, + "step": 10512 + }, + { + "epoch": 1.1545135075774215, + "grad_norm": 2.009249448776245, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6908565759658813, + "num_tokens": 268079282.0, + "step": 10513 + }, + { + "epoch": 1.154623325280035, + "grad_norm": 2.5811667442321777, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7349014282226562, + "num_tokens": 268100619.0, + "step": 10514 + }, + { + "epoch": 1.1547331429826488, + "grad_norm": 2.168395519256592, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7154749631881714, + "num_tokens": 268127233.0, + "step": 10515 + }, + { + "epoch": 1.1548429606852624, + "grad_norm": 2.0697712898254395, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7120001316070557, + "num_tokens": 268158091.0, + "step": 10516 + }, + { + "epoch": 1.1549527783878761, + "grad_norm": 2.2096166610717773, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.726689338684082, + "num_tokens": 268184864.0, + "step": 10517 + }, + { + "epoch": 1.15506259609049, + "grad_norm": 2.2105422019958496, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7260965704917908, + "num_tokens": 268209127.0, + "step": 10518 + }, + { + "epoch": 1.1551724137931034, + "grad_norm": 2.1361823081970215, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.7015605568885803, + "num_tokens": 268241361.0, + "step": 10519 + }, + { + "epoch": 1.1552822314957172, + "grad_norm": 2.2685611248016357, + "learning_rate": 1e-06, + "loss": 0.7824, + "mean_token_accuracy": 0.7534757256507874, + "num_tokens": 268265644.0, + "step": 10520 + }, + { + "epoch": 1.1553920491983307, + "grad_norm": 2.5523340702056885, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7273604869842529, + "num_tokens": 268285434.0, + "step": 10521 + }, + { + "epoch": 1.1555018669009445, + "grad_norm": 2.139042854309082, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7342873215675354, + "num_tokens": 268312183.0, + "step": 10522 + }, + { + "epoch": 1.155611684603558, + "grad_norm": 2.072575092315674, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7119312286376953, + "num_tokens": 268342096.0, + "step": 10523 + }, + { + "epoch": 1.1557215023061718, + "grad_norm": 2.2492637634277344, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7109502553939819, + "num_tokens": 268367358.0, + "step": 10524 + }, + { + "epoch": 1.1558313200087853, + "grad_norm": 2.3039157390594482, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7162710428237915, + "num_tokens": 268391544.0, + "step": 10525 + }, + { + "epoch": 1.155941137711399, + "grad_norm": 2.074031352996826, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7152050137519836, + "num_tokens": 268421162.0, + "step": 10526 + }, + { + "epoch": 1.1560509554140128, + "grad_norm": 2.3694396018981934, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.6986035108566284, + "num_tokens": 268443986.0, + "step": 10527 + }, + { + "epoch": 1.1561607731166264, + "grad_norm": 2.730358600616455, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7341318726539612, + "num_tokens": 268463297.0, + "step": 10528 + }, + { + "epoch": 1.15627059081924, + "grad_norm": 2.330296277999878, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.705348789691925, + "num_tokens": 268487113.0, + "step": 10529 + }, + { + "epoch": 1.1563804085218536, + "grad_norm": 2.368454694747925, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7263280153274536, + "num_tokens": 268509865.0, + "step": 10530 + }, + { + "epoch": 1.1564902262244674, + "grad_norm": 2.3970112800598145, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7203192710876465, + "num_tokens": 268532971.0, + "step": 10531 + }, + { + "epoch": 1.1566000439270812, + "grad_norm": 2.330307722091675, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7029262781143188, + "num_tokens": 268557563.0, + "step": 10532 + }, + { + "epoch": 1.1567098616296947, + "grad_norm": 2.3289217948913574, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.738714337348938, + "num_tokens": 268581053.0, + "step": 10533 + }, + { + "epoch": 1.1568196793323084, + "grad_norm": 2.116020917892456, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7047815322875977, + "num_tokens": 268609551.0, + "step": 10534 + }, + { + "epoch": 1.156929497034922, + "grad_norm": 2.261521100997925, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.73919677734375, + "num_tokens": 268632954.0, + "step": 10535 + }, + { + "epoch": 1.1570393147375357, + "grad_norm": 2.189026117324829, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.706845760345459, + "num_tokens": 268661612.0, + "step": 10536 + }, + { + "epoch": 1.1571491324401493, + "grad_norm": 2.3172457218170166, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7153115272521973, + "num_tokens": 268686689.0, + "step": 10537 + }, + { + "epoch": 1.157258950142763, + "grad_norm": 2.3665621280670166, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7566203474998474, + "num_tokens": 268708477.0, + "step": 10538 + }, + { + "epoch": 1.1573687678453766, + "grad_norm": 2.4150869846343994, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7284926772117615, + "num_tokens": 268732262.0, + "step": 10539 + }, + { + "epoch": 1.1574785855479903, + "grad_norm": 2.4687960147857666, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7389736175537109, + "num_tokens": 268753804.0, + "step": 10540 + }, + { + "epoch": 1.157588403250604, + "grad_norm": 2.5156431198120117, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7406734228134155, + "num_tokens": 268774519.0, + "step": 10541 + }, + { + "epoch": 1.1576982209532176, + "grad_norm": 2.396460771560669, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.715254008769989, + "num_tokens": 268797498.0, + "step": 10542 + }, + { + "epoch": 1.1578080386558314, + "grad_norm": 2.458432674407959, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7323517203330994, + "num_tokens": 268821988.0, + "step": 10543 + }, + { + "epoch": 1.157917856358445, + "grad_norm": 2.0880165100097656, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7085093259811401, + "num_tokens": 268852162.0, + "step": 10544 + }, + { + "epoch": 1.1580276740610587, + "grad_norm": 2.092193603515625, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7026463747024536, + "num_tokens": 268880174.0, + "step": 10545 + }, + { + "epoch": 1.1581374917636724, + "grad_norm": 2.1036715507507324, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7475101351737976, + "num_tokens": 268906794.0, + "step": 10546 + }, + { + "epoch": 1.158247309466286, + "grad_norm": 2.1543869972229004, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7202788591384888, + "num_tokens": 268934894.0, + "step": 10547 + }, + { + "epoch": 1.1583571271688997, + "grad_norm": 2.3082640171051025, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7087389230728149, + "num_tokens": 268959631.0, + "step": 10548 + }, + { + "epoch": 1.1584669448715132, + "grad_norm": 2.198714256286621, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7217150330543518, + "num_tokens": 268986298.0, + "step": 10549 + }, + { + "epoch": 1.158576762574127, + "grad_norm": 2.2561280727386475, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7177867889404297, + "num_tokens": 269012165.0, + "step": 10550 + }, + { + "epoch": 1.1586865802767405, + "grad_norm": 2.348214864730835, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7278728485107422, + "num_tokens": 269036992.0, + "step": 10551 + }, + { + "epoch": 1.1587963979793543, + "grad_norm": 2.317202568054199, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7411990761756897, + "num_tokens": 269060449.0, + "step": 10552 + }, + { + "epoch": 1.1589062156819678, + "grad_norm": 2.170358657836914, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7151844501495361, + "num_tokens": 269087764.0, + "step": 10553 + }, + { + "epoch": 1.1590160333845816, + "grad_norm": 2.1966662406921387, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7214844226837158, + "num_tokens": 269113045.0, + "step": 10554 + }, + { + "epoch": 1.1591258510871953, + "grad_norm": 2.5151755809783936, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7425155639648438, + "num_tokens": 269133051.0, + "step": 10555 + }, + { + "epoch": 1.1592356687898089, + "grad_norm": 2.131943702697754, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7294683456420898, + "num_tokens": 269161061.0, + "step": 10556 + }, + { + "epoch": 1.1593454864924226, + "grad_norm": 2.2899632453918457, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7117077112197876, + "num_tokens": 269185656.0, + "step": 10557 + }, + { + "epoch": 1.1594553041950362, + "grad_norm": 2.1631782054901123, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.6964409351348877, + "num_tokens": 269212600.0, + "step": 10558 + }, + { + "epoch": 1.15956512189765, + "grad_norm": 2.399010181427002, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7467602491378784, + "num_tokens": 269235530.0, + "step": 10559 + }, + { + "epoch": 1.1596749396002635, + "grad_norm": 2.4456465244293213, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7360089421272278, + "num_tokens": 269258779.0, + "step": 10560 + }, + { + "epoch": 1.1597847573028772, + "grad_norm": 2.486039400100708, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7434107661247253, + "num_tokens": 269281053.0, + "step": 10561 + }, + { + "epoch": 1.159894575005491, + "grad_norm": 2.402675151824951, + "learning_rate": 1e-06, + "loss": 0.8029, + "mean_token_accuracy": 0.7442991733551025, + "num_tokens": 269304246.0, + "step": 10562 + }, + { + "epoch": 1.1600043927081045, + "grad_norm": 2.373852252960205, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7237171530723572, + "num_tokens": 269327697.0, + "step": 10563 + }, + { + "epoch": 1.1601142104107183, + "grad_norm": 2.23480486869812, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7071298360824585, + "num_tokens": 269353422.0, + "step": 10564 + }, + { + "epoch": 1.1602240281133318, + "grad_norm": 2.190274715423584, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7199374437332153, + "num_tokens": 269380651.0, + "step": 10565 + }, + { + "epoch": 1.1603338458159456, + "grad_norm": 2.198315382003784, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.693356990814209, + "num_tokens": 269408566.0, + "step": 10566 + }, + { + "epoch": 1.160443663518559, + "grad_norm": 2.2014319896698, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7208207845687866, + "num_tokens": 269437438.0, + "step": 10567 + }, + { + "epoch": 1.1605534812211729, + "grad_norm": 2.6400861740112305, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.72136390209198, + "num_tokens": 269457971.0, + "step": 10568 + }, + { + "epoch": 1.1606632989237866, + "grad_norm": 2.1862471103668213, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7195439338684082, + "num_tokens": 269484981.0, + "step": 10569 + }, + { + "epoch": 1.1607731166264001, + "grad_norm": 2.48252534866333, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7238134741783142, + "num_tokens": 269504839.0, + "step": 10570 + }, + { + "epoch": 1.160882934329014, + "grad_norm": 2.1116325855255127, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7465043663978577, + "num_tokens": 269531759.0, + "step": 10571 + }, + { + "epoch": 1.1609927520316274, + "grad_norm": 2.2577157020568848, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.6998586654663086, + "num_tokens": 269556768.0, + "step": 10572 + }, + { + "epoch": 1.1611025697342412, + "grad_norm": 2.2139828205108643, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7203213572502136, + "num_tokens": 269583052.0, + "step": 10573 + }, + { + "epoch": 1.1612123874368547, + "grad_norm": 2.342700481414795, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7419425249099731, + "num_tokens": 269605250.0, + "step": 10574 + }, + { + "epoch": 1.1613222051394685, + "grad_norm": 2.1304256916046143, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7297110557556152, + "num_tokens": 269632305.0, + "step": 10575 + }, + { + "epoch": 1.161432022842082, + "grad_norm": 1.9475914239883423, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7118407487869263, + "num_tokens": 269666000.0, + "step": 10576 + }, + { + "epoch": 1.1615418405446958, + "grad_norm": 2.3569698333740234, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7415961623191833, + "num_tokens": 269688990.0, + "step": 10577 + }, + { + "epoch": 1.1616516582473095, + "grad_norm": 2.440692186355591, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7458572387695312, + "num_tokens": 269711031.0, + "step": 10578 + }, + { + "epoch": 1.161761475949923, + "grad_norm": 2.0757477283477783, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7046751379966736, + "num_tokens": 269741446.0, + "step": 10579 + }, + { + "epoch": 1.1618712936525368, + "grad_norm": 2.2461156845092773, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.714601993560791, + "num_tokens": 269768744.0, + "step": 10580 + }, + { + "epoch": 1.1619811113551504, + "grad_norm": 2.0454657077789307, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.708579421043396, + "num_tokens": 269799607.0, + "step": 10581 + }, + { + "epoch": 1.1620909290577641, + "grad_norm": 2.1413426399230957, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7249582409858704, + "num_tokens": 269825055.0, + "step": 10582 + }, + { + "epoch": 1.1622007467603779, + "grad_norm": 2.7420566082000732, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7607564926147461, + "num_tokens": 269842129.0, + "step": 10583 + }, + { + "epoch": 1.1623105644629914, + "grad_norm": 2.494473457336426, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7271090745925903, + "num_tokens": 269864163.0, + "step": 10584 + }, + { + "epoch": 1.1624203821656052, + "grad_norm": 2.4910941123962402, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7397241592407227, + "num_tokens": 269885523.0, + "step": 10585 + }, + { + "epoch": 1.1625301998682187, + "grad_norm": 2.275609016418457, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6898033618927002, + "num_tokens": 269913531.0, + "step": 10586 + }, + { + "epoch": 1.1626400175708325, + "grad_norm": 2.2658989429473877, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7358721494674683, + "num_tokens": 269936154.0, + "step": 10587 + }, + { + "epoch": 1.162749835273446, + "grad_norm": 2.267596960067749, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7122442722320557, + "num_tokens": 269963236.0, + "step": 10588 + }, + { + "epoch": 1.1628596529760598, + "grad_norm": 2.122756004333496, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.735439658164978, + "num_tokens": 269991136.0, + "step": 10589 + }, + { + "epoch": 1.1629694706786733, + "grad_norm": 2.3238465785980225, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7012223601341248, + "num_tokens": 270015802.0, + "step": 10590 + }, + { + "epoch": 1.163079288381287, + "grad_norm": 2.266277313232422, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7085449695587158, + "num_tokens": 270040746.0, + "step": 10591 + }, + { + "epoch": 1.1631891060839008, + "grad_norm": 2.2862370014190674, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7214475870132446, + "num_tokens": 270065436.0, + "step": 10592 + }, + { + "epoch": 1.1632989237865143, + "grad_norm": 2.141108989715576, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7212690114974976, + "num_tokens": 270092477.0, + "step": 10593 + }, + { + "epoch": 1.163408741489128, + "grad_norm": 2.2532176971435547, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7384872436523438, + "num_tokens": 270116717.0, + "step": 10594 + }, + { + "epoch": 1.1635185591917416, + "grad_norm": 2.3987982273101807, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7345018982887268, + "num_tokens": 270138424.0, + "step": 10595 + }, + { + "epoch": 1.1636283768943554, + "grad_norm": 2.269164562225342, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7096753120422363, + "num_tokens": 270163314.0, + "step": 10596 + }, + { + "epoch": 1.1637381945969691, + "grad_norm": 2.6142916679382324, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7467684745788574, + "num_tokens": 270182495.0, + "step": 10597 + }, + { + "epoch": 1.1638480122995827, + "grad_norm": 2.4296669960021973, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7121812105178833, + "num_tokens": 270203935.0, + "step": 10598 + }, + { + "epoch": 1.1639578300021964, + "grad_norm": 2.3489553928375244, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.721288800239563, + "num_tokens": 270228172.0, + "step": 10599 + }, + { + "epoch": 1.16406764770481, + "grad_norm": 2.4887492656707764, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.734285831451416, + "num_tokens": 270250136.0, + "step": 10600 + }, + { + "epoch": 1.1641774654074237, + "grad_norm": 2.4774177074432373, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7061306834220886, + "num_tokens": 270272417.0, + "step": 10601 + }, + { + "epoch": 1.1642872831100373, + "grad_norm": 2.224548816680908, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7447102665901184, + "num_tokens": 270298091.0, + "step": 10602 + }, + { + "epoch": 1.164397100812651, + "grad_norm": 2.123255968093872, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7051855325698853, + "num_tokens": 270327370.0, + "step": 10603 + }, + { + "epoch": 1.1645069185152646, + "grad_norm": 2.134378433227539, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7206458449363708, + "num_tokens": 270355556.0, + "step": 10604 + }, + { + "epoch": 1.1646167362178783, + "grad_norm": 2.1823902130126953, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.726405918598175, + "num_tokens": 270381144.0, + "step": 10605 + }, + { + "epoch": 1.164726553920492, + "grad_norm": 2.0020909309387207, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.751115083694458, + "num_tokens": 270407919.0, + "step": 10606 + }, + { + "epoch": 1.1648363716231056, + "grad_norm": 2.7416794300079346, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7412922978401184, + "num_tokens": 270428455.0, + "step": 10607 + }, + { + "epoch": 1.1649461893257194, + "grad_norm": 2.1705641746520996, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.6877981424331665, + "num_tokens": 270456381.0, + "step": 10608 + }, + { + "epoch": 1.165056007028333, + "grad_norm": 2.229306697845459, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7038455009460449, + "num_tokens": 270481388.0, + "step": 10609 + }, + { + "epoch": 1.1651658247309467, + "grad_norm": 2.403451681137085, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7256104350090027, + "num_tokens": 270503427.0, + "step": 10610 + }, + { + "epoch": 1.1652756424335604, + "grad_norm": 2.6157991886138916, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7474128603935242, + "num_tokens": 270523520.0, + "step": 10611 + }, + { + "epoch": 1.165385460136174, + "grad_norm": 2.3559019565582275, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7183049917221069, + "num_tokens": 270548345.0, + "step": 10612 + }, + { + "epoch": 1.1654952778387877, + "grad_norm": 2.5792086124420166, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7395649552345276, + "num_tokens": 270568059.0, + "step": 10613 + }, + { + "epoch": 1.1656050955414012, + "grad_norm": 2.1600966453552246, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7351998090744019, + "num_tokens": 270594410.0, + "step": 10614 + }, + { + "epoch": 1.165714913244015, + "grad_norm": 2.382918357849121, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7270714640617371, + "num_tokens": 270617253.0, + "step": 10615 + }, + { + "epoch": 1.1658247309466285, + "grad_norm": 2.2872447967529297, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7397061586380005, + "num_tokens": 270640722.0, + "step": 10616 + }, + { + "epoch": 1.1659345486492423, + "grad_norm": 2.426057815551758, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7348188161849976, + "num_tokens": 270661662.0, + "step": 10617 + }, + { + "epoch": 1.1660443663518558, + "grad_norm": 2.5031609535217285, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.732103705406189, + "num_tokens": 270682593.0, + "step": 10618 + }, + { + "epoch": 1.1661541840544696, + "grad_norm": 2.220083475112915, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7405562400817871, + "num_tokens": 270709656.0, + "step": 10619 + }, + { + "epoch": 1.1662640017570833, + "grad_norm": 2.5780906677246094, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7078404426574707, + "num_tokens": 270732978.0, + "step": 10620 + }, + { + "epoch": 1.1663738194596969, + "grad_norm": 2.225883722305298, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7198666334152222, + "num_tokens": 270758615.0, + "step": 10621 + }, + { + "epoch": 1.1664836371623106, + "grad_norm": 2.06124210357666, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7307376265525818, + "num_tokens": 270786971.0, + "step": 10622 + }, + { + "epoch": 1.1665934548649242, + "grad_norm": 2.309722423553467, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7372174263000488, + "num_tokens": 270810419.0, + "step": 10623 + }, + { + "epoch": 1.166703272567538, + "grad_norm": 2.1761832237243652, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7198554277420044, + "num_tokens": 270835765.0, + "step": 10624 + }, + { + "epoch": 1.1668130902701515, + "grad_norm": 2.3169844150543213, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7311853170394897, + "num_tokens": 270859320.0, + "step": 10625 + }, + { + "epoch": 1.1669229079727652, + "grad_norm": 2.271422863006592, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6961908340454102, + "num_tokens": 270885268.0, + "step": 10626 + }, + { + "epoch": 1.167032725675379, + "grad_norm": 2.181844472885132, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7091580629348755, + "num_tokens": 270913555.0, + "step": 10627 + }, + { + "epoch": 1.1671425433779925, + "grad_norm": 2.0741689205169678, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7253473401069641, + "num_tokens": 270942085.0, + "step": 10628 + }, + { + "epoch": 1.1672523610806063, + "grad_norm": 2.098675489425659, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7124452590942383, + "num_tokens": 270971463.0, + "step": 10629 + }, + { + "epoch": 1.1673621787832198, + "grad_norm": 2.676684856414795, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7179765701293945, + "num_tokens": 270991470.0, + "step": 10630 + }, + { + "epoch": 1.1674719964858336, + "grad_norm": 2.2354800701141357, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7080378532409668, + "num_tokens": 271018187.0, + "step": 10631 + }, + { + "epoch": 1.167581814188447, + "grad_norm": 2.1982052326202393, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7258021831512451, + "num_tokens": 271044467.0, + "step": 10632 + }, + { + "epoch": 1.1676916318910608, + "grad_norm": 2.2220473289489746, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7275251150131226, + "num_tokens": 271069470.0, + "step": 10633 + }, + { + "epoch": 1.1678014495936746, + "grad_norm": 2.479747772216797, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7451475858688354, + "num_tokens": 271090535.0, + "step": 10634 + }, + { + "epoch": 1.1679112672962881, + "grad_norm": 2.4251797199249268, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.722656786441803, + "num_tokens": 271112397.0, + "step": 10635 + }, + { + "epoch": 1.168021084998902, + "grad_norm": 2.2235422134399414, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.7030662894248962, + "num_tokens": 271139629.0, + "step": 10636 + }, + { + "epoch": 1.1681309027015154, + "grad_norm": 2.0824472904205322, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7215907573699951, + "num_tokens": 271167859.0, + "step": 10637 + }, + { + "epoch": 1.1682407204041292, + "grad_norm": 2.0629007816314697, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7144145965576172, + "num_tokens": 271195742.0, + "step": 10638 + }, + { + "epoch": 1.1683505381067427, + "grad_norm": 2.5940232276916504, + "learning_rate": 1e-06, + "loss": 0.7999, + "mean_token_accuracy": 0.7544212341308594, + "num_tokens": 271215033.0, + "step": 10639 + }, + { + "epoch": 1.1684603558093565, + "grad_norm": 2.338186502456665, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.702920138835907, + "num_tokens": 271239887.0, + "step": 10640 + }, + { + "epoch": 1.16857017351197, + "grad_norm": 2.2839195728302, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7114263772964478, + "num_tokens": 271266486.0, + "step": 10641 + }, + { + "epoch": 1.1686799912145838, + "grad_norm": 2.660479784011841, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7173036336898804, + "num_tokens": 271285462.0, + "step": 10642 + }, + { + "epoch": 1.1687898089171975, + "grad_norm": 2.065218925476074, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7099190354347229, + "num_tokens": 271316321.0, + "step": 10643 + }, + { + "epoch": 1.168899626619811, + "grad_norm": 2.1970174312591553, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7254400253295898, + "num_tokens": 271340398.0, + "step": 10644 + }, + { + "epoch": 1.1690094443224248, + "grad_norm": 2.1875038146972656, + "learning_rate": 1e-06, + "loss": 1.0572, + "mean_token_accuracy": 0.6837031841278076, + "num_tokens": 271370868.0, + "step": 10645 + }, + { + "epoch": 1.1691192620250384, + "grad_norm": 2.4782660007476807, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7274045348167419, + "num_tokens": 271392525.0, + "step": 10646 + }, + { + "epoch": 1.169229079727652, + "grad_norm": 1.9917160272598267, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.6998234987258911, + "num_tokens": 271423903.0, + "step": 10647 + }, + { + "epoch": 1.1693388974302659, + "grad_norm": 2.2845957279205322, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7240967154502869, + "num_tokens": 271449076.0, + "step": 10648 + }, + { + "epoch": 1.1694487151328794, + "grad_norm": 2.3992726802825928, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7157698273658752, + "num_tokens": 271472804.0, + "step": 10649 + }, + { + "epoch": 1.1695585328354932, + "grad_norm": 2.643728017807007, + "learning_rate": 1e-06, + "loss": 0.7643, + "mean_token_accuracy": 0.7561472654342651, + "num_tokens": 271492078.0, + "step": 10650 + }, + { + "epoch": 1.1696683505381067, + "grad_norm": 2.099003791809082, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.712553083896637, + "num_tokens": 271522387.0, + "step": 10651 + }, + { + "epoch": 1.1697781682407205, + "grad_norm": 2.0339860916137695, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7294739484786987, + "num_tokens": 271552801.0, + "step": 10652 + }, + { + "epoch": 1.169887985943334, + "grad_norm": 1.8654009103775024, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7316212058067322, + "num_tokens": 271586699.0, + "step": 10653 + }, + { + "epoch": 1.1699978036459477, + "grad_norm": 2.274260997772217, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7180661559104919, + "num_tokens": 271613692.0, + "step": 10654 + }, + { + "epoch": 1.1701076213485613, + "grad_norm": 2.4393527507781982, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7403721213340759, + "num_tokens": 271635629.0, + "step": 10655 + }, + { + "epoch": 1.170217439051175, + "grad_norm": 2.186723470687866, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7168416976928711, + "num_tokens": 271662921.0, + "step": 10656 + }, + { + "epoch": 1.1703272567537888, + "grad_norm": 2.228651523590088, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7290889024734497, + "num_tokens": 271687013.0, + "step": 10657 + }, + { + "epoch": 1.1704370744564023, + "grad_norm": 2.332909345626831, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7385839223861694, + "num_tokens": 271709376.0, + "step": 10658 + }, + { + "epoch": 1.170546892159016, + "grad_norm": 2.733126640319824, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7003804445266724, + "num_tokens": 271731152.0, + "step": 10659 + }, + { + "epoch": 1.1706567098616296, + "grad_norm": 2.532348394393921, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7309873104095459, + "num_tokens": 271751361.0, + "step": 10660 + }, + { + "epoch": 1.1707665275642434, + "grad_norm": 2.215182065963745, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7266421914100647, + "num_tokens": 271776638.0, + "step": 10661 + }, + { + "epoch": 1.1708763452668571, + "grad_norm": 2.1888926029205322, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7260820865631104, + "num_tokens": 271802005.0, + "step": 10662 + }, + { + "epoch": 1.1709861629694707, + "grad_norm": 2.2367618083953857, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7124546766281128, + "num_tokens": 271827235.0, + "step": 10663 + }, + { + "epoch": 1.1710959806720844, + "grad_norm": 2.725506067276001, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.724435567855835, + "num_tokens": 271846927.0, + "step": 10664 + }, + { + "epoch": 1.171205798374698, + "grad_norm": 2.4375760555267334, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7294288873672485, + "num_tokens": 271868101.0, + "step": 10665 + }, + { + "epoch": 1.1713156160773117, + "grad_norm": 2.144646167755127, + "learning_rate": 1e-06, + "loss": 1.0409, + "mean_token_accuracy": 0.686642050743103, + "num_tokens": 271896935.0, + "step": 10666 + }, + { + "epoch": 1.1714254337799253, + "grad_norm": 2.121678590774536, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7009451389312744, + "num_tokens": 271928647.0, + "step": 10667 + }, + { + "epoch": 1.171535251482539, + "grad_norm": 2.19724702835083, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7064996361732483, + "num_tokens": 271955964.0, + "step": 10668 + }, + { + "epoch": 1.1716450691851525, + "grad_norm": 2.178393840789795, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7327193021774292, + "num_tokens": 271982427.0, + "step": 10669 + }, + { + "epoch": 1.1717548868877663, + "grad_norm": 2.22965145111084, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7334533929824829, + "num_tokens": 272006157.0, + "step": 10670 + }, + { + "epoch": 1.17186470459038, + "grad_norm": 2.532790184020996, + "learning_rate": 1e-06, + "loss": 0.8234, + "mean_token_accuracy": 0.741752564907074, + "num_tokens": 272026728.0, + "step": 10671 + }, + { + "epoch": 1.1719745222929936, + "grad_norm": 2.3427298069000244, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.6982545852661133, + "num_tokens": 272053224.0, + "step": 10672 + }, + { + "epoch": 1.1720843399956073, + "grad_norm": 2.0785861015319824, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7118741273880005, + "num_tokens": 272081644.0, + "step": 10673 + }, + { + "epoch": 1.1721941576982209, + "grad_norm": 1.963681936264038, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7274290919303894, + "num_tokens": 272112180.0, + "step": 10674 + }, + { + "epoch": 1.1723039754008346, + "grad_norm": 2.2678630352020264, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7102591395378113, + "num_tokens": 272136547.0, + "step": 10675 + }, + { + "epoch": 1.1724137931034484, + "grad_norm": 2.5646748542785645, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7500328421592712, + "num_tokens": 272156799.0, + "step": 10676 + }, + { + "epoch": 1.172523610806062, + "grad_norm": 2.049971342086792, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7135425209999084, + "num_tokens": 272188839.0, + "step": 10677 + }, + { + "epoch": 1.1726334285086757, + "grad_norm": 2.4332218170166016, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7109625935554504, + "num_tokens": 272212411.0, + "step": 10678 + }, + { + "epoch": 1.1727432462112892, + "grad_norm": 2.3881168365478516, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7196534872055054, + "num_tokens": 272236275.0, + "step": 10679 + }, + { + "epoch": 1.172853063913903, + "grad_norm": 2.15218186378479, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7039234638214111, + "num_tokens": 272265473.0, + "step": 10680 + }, + { + "epoch": 1.1729628816165165, + "grad_norm": 2.4151909351348877, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7314506769180298, + "num_tokens": 272290322.0, + "step": 10681 + }, + { + "epoch": 1.1730726993191303, + "grad_norm": 2.5020432472229004, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7352508306503296, + "num_tokens": 272309541.0, + "step": 10682 + }, + { + "epoch": 1.1731825170217438, + "grad_norm": 2.1407556533813477, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7188297510147095, + "num_tokens": 272336086.0, + "step": 10683 + }, + { + "epoch": 1.1732923347243576, + "grad_norm": 2.2651562690734863, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7118675112724304, + "num_tokens": 272361485.0, + "step": 10684 + }, + { + "epoch": 1.1734021524269713, + "grad_norm": 2.298494577407837, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7210729122161865, + "num_tokens": 272388014.0, + "step": 10685 + }, + { + "epoch": 1.1735119701295849, + "grad_norm": 2.399614095687866, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7053049802780151, + "num_tokens": 272412198.0, + "step": 10686 + }, + { + "epoch": 1.1736217878321986, + "grad_norm": 2.308906316757202, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7292225360870361, + "num_tokens": 272434364.0, + "step": 10687 + }, + { + "epoch": 1.1737316055348122, + "grad_norm": 2.1759705543518066, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.6951834559440613, + "num_tokens": 272461166.0, + "step": 10688 + }, + { + "epoch": 1.173841423237426, + "grad_norm": 2.1113736629486084, + "learning_rate": 1e-06, + "loss": 0.7772, + "mean_token_accuracy": 0.7518993616104126, + "num_tokens": 272485749.0, + "step": 10689 + }, + { + "epoch": 1.1739512409400394, + "grad_norm": 2.3718223571777344, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7306686639785767, + "num_tokens": 272508898.0, + "step": 10690 + }, + { + "epoch": 1.1740610586426532, + "grad_norm": 2.067499876022339, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7056475281715393, + "num_tokens": 272536740.0, + "step": 10691 + }, + { + "epoch": 1.1741708763452667, + "grad_norm": 2.3651270866394043, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7354573607444763, + "num_tokens": 272560802.0, + "step": 10692 + }, + { + "epoch": 1.1742806940478805, + "grad_norm": 2.5110044479370117, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7413854598999023, + "num_tokens": 272583153.0, + "step": 10693 + }, + { + "epoch": 1.1743905117504942, + "grad_norm": 2.2412121295928955, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7123610973358154, + "num_tokens": 272609281.0, + "step": 10694 + }, + { + "epoch": 1.1745003294531078, + "grad_norm": 2.266733407974243, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7031400799751282, + "num_tokens": 272636751.0, + "step": 10695 + }, + { + "epoch": 1.1746101471557215, + "grad_norm": 2.4143786430358887, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7070842385292053, + "num_tokens": 272660800.0, + "step": 10696 + }, + { + "epoch": 1.174719964858335, + "grad_norm": 1.9800617694854736, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6928011178970337, + "num_tokens": 272692391.0, + "step": 10697 + }, + { + "epoch": 1.1748297825609488, + "grad_norm": 1.9020066261291504, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7171463966369629, + "num_tokens": 272727578.0, + "step": 10698 + }, + { + "epoch": 1.1749396002635626, + "grad_norm": 2.063476085662842, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7153885960578918, + "num_tokens": 272759256.0, + "step": 10699 + }, + { + "epoch": 1.1750494179661761, + "grad_norm": 1.996396780014038, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7030493021011353, + "num_tokens": 272789117.0, + "step": 10700 + }, + { + "epoch": 1.1751592356687899, + "grad_norm": 2.3489975929260254, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7361328601837158, + "num_tokens": 272812447.0, + "step": 10701 + }, + { + "epoch": 1.1752690533714034, + "grad_norm": 2.3798253536224365, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.7460142970085144, + "num_tokens": 272833840.0, + "step": 10702 + }, + { + "epoch": 1.1753788710740172, + "grad_norm": 2.1191048622131348, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7286157011985779, + "num_tokens": 272860496.0, + "step": 10703 + }, + { + "epoch": 1.1754886887766307, + "grad_norm": 2.177374839782715, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.6972509622573853, + "num_tokens": 272887577.0, + "step": 10704 + }, + { + "epoch": 1.1755985064792445, + "grad_norm": 2.451752185821533, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7265599966049194, + "num_tokens": 272910729.0, + "step": 10705 + }, + { + "epoch": 1.175708324181858, + "grad_norm": 2.1449320316314697, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.738819420337677, + "num_tokens": 272938091.0, + "step": 10706 + }, + { + "epoch": 1.1758181418844718, + "grad_norm": 2.1376335620880127, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7209985852241516, + "num_tokens": 272965734.0, + "step": 10707 + }, + { + "epoch": 1.1759279595870855, + "grad_norm": 2.339379072189331, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7436248064041138, + "num_tokens": 272987432.0, + "step": 10708 + }, + { + "epoch": 1.176037777289699, + "grad_norm": 2.627767324447632, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7224471569061279, + "num_tokens": 273007406.0, + "step": 10709 + }, + { + "epoch": 1.1761475949923128, + "grad_norm": 2.137045383453369, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7029719948768616, + "num_tokens": 273035796.0, + "step": 10710 + }, + { + "epoch": 1.1762574126949263, + "grad_norm": 1.9379743337631226, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7232407331466675, + "num_tokens": 273066369.0, + "step": 10711 + }, + { + "epoch": 1.17636723039754, + "grad_norm": 2.538425922393799, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7070350646972656, + "num_tokens": 273087424.0, + "step": 10712 + }, + { + "epoch": 1.1764770481001539, + "grad_norm": 2.1233036518096924, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7291473746299744, + "num_tokens": 273116011.0, + "step": 10713 + }, + { + "epoch": 1.1765868658027674, + "grad_norm": 2.808802366256714, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7256572842597961, + "num_tokens": 273133840.0, + "step": 10714 + }, + { + "epoch": 1.1766966835053811, + "grad_norm": 2.0378167629241943, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7091240286827087, + "num_tokens": 273163031.0, + "step": 10715 + }, + { + "epoch": 1.1768065012079947, + "grad_norm": 2.334047555923462, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7189513444900513, + "num_tokens": 273187920.0, + "step": 10716 + }, + { + "epoch": 1.1769163189106084, + "grad_norm": 2.19846248626709, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7186282277107239, + "num_tokens": 273213947.0, + "step": 10717 + }, + { + "epoch": 1.177026136613222, + "grad_norm": 2.619581937789917, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7263897657394409, + "num_tokens": 273234524.0, + "step": 10718 + }, + { + "epoch": 1.1771359543158357, + "grad_norm": 2.383502960205078, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7175465822219849, + "num_tokens": 273257886.0, + "step": 10719 + }, + { + "epoch": 1.1772457720184493, + "grad_norm": 2.269087076187134, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7070325016975403, + "num_tokens": 273283516.0, + "step": 10720 + }, + { + "epoch": 1.177355589721063, + "grad_norm": 2.3962435722351074, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7223716974258423, + "num_tokens": 273305513.0, + "step": 10721 + }, + { + "epoch": 1.1774654074236768, + "grad_norm": 2.321115016937256, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.6975291967391968, + "num_tokens": 273331589.0, + "step": 10722 + }, + { + "epoch": 1.1775752251262903, + "grad_norm": 2.2732856273651123, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.741584837436676, + "num_tokens": 273358367.0, + "step": 10723 + }, + { + "epoch": 1.177685042828904, + "grad_norm": 2.1188108921051025, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7305285930633545, + "num_tokens": 273386526.0, + "step": 10724 + }, + { + "epoch": 1.1777948605315176, + "grad_norm": 2.432497978210449, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7263442873954773, + "num_tokens": 273409444.0, + "step": 10725 + }, + { + "epoch": 1.1779046782341314, + "grad_norm": 2.351322889328003, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7193384170532227, + "num_tokens": 273432462.0, + "step": 10726 + }, + { + "epoch": 1.1780144959367451, + "grad_norm": 2.334489107131958, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7042115926742554, + "num_tokens": 273459277.0, + "step": 10727 + }, + { + "epoch": 1.1781243136393587, + "grad_norm": 2.0795412063598633, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7126250267028809, + "num_tokens": 273488029.0, + "step": 10728 + }, + { + "epoch": 1.1782341313419724, + "grad_norm": 2.2140536308288574, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.7619584798812866, + "num_tokens": 273513398.0, + "step": 10729 + }, + { + "epoch": 1.178343949044586, + "grad_norm": 2.1277596950531006, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.753723680973053, + "num_tokens": 273537818.0, + "step": 10730 + }, + { + "epoch": 1.1784537667471997, + "grad_norm": 2.1932084560394287, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7272967100143433, + "num_tokens": 273564700.0, + "step": 10731 + }, + { + "epoch": 1.1785635844498132, + "grad_norm": 2.2947254180908203, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7117606401443481, + "num_tokens": 273589887.0, + "step": 10732 + }, + { + "epoch": 1.178673402152427, + "grad_norm": 2.2005512714385986, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.704524576663971, + "num_tokens": 273618259.0, + "step": 10733 + }, + { + "epoch": 1.1787832198550405, + "grad_norm": 2.2761847972869873, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7155452966690063, + "num_tokens": 273643538.0, + "step": 10734 + }, + { + "epoch": 1.1788930375576543, + "grad_norm": 2.343825578689575, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7071548700332642, + "num_tokens": 273668252.0, + "step": 10735 + }, + { + "epoch": 1.179002855260268, + "grad_norm": 2.4643588066101074, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7313801050186157, + "num_tokens": 273691832.0, + "step": 10736 + }, + { + "epoch": 1.1791126729628816, + "grad_norm": 2.263781785964966, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7237844467163086, + "num_tokens": 273715906.0, + "step": 10737 + }, + { + "epoch": 1.1792224906654953, + "grad_norm": 2.78275465965271, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7202656269073486, + "num_tokens": 273734859.0, + "step": 10738 + }, + { + "epoch": 1.1793323083681089, + "grad_norm": 2.0750648975372314, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7146222591400146, + "num_tokens": 273762962.0, + "step": 10739 + }, + { + "epoch": 1.1794421260707226, + "grad_norm": 1.9824222326278687, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7181034088134766, + "num_tokens": 273793103.0, + "step": 10740 + }, + { + "epoch": 1.1795519437733364, + "grad_norm": 1.8918704986572266, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6963750123977661, + "num_tokens": 273828856.0, + "step": 10741 + }, + { + "epoch": 1.17966176147595, + "grad_norm": 2.396111488342285, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7051216959953308, + "num_tokens": 273854620.0, + "step": 10742 + }, + { + "epoch": 1.1797715791785637, + "grad_norm": 2.3152213096618652, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.749555230140686, + "num_tokens": 273877315.0, + "step": 10743 + }, + { + "epoch": 1.1798813968811772, + "grad_norm": 2.394871950149536, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7127280235290527, + "num_tokens": 273901048.0, + "step": 10744 + }, + { + "epoch": 1.179991214583791, + "grad_norm": 2.570948839187622, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7274507880210876, + "num_tokens": 273921063.0, + "step": 10745 + }, + { + "epoch": 1.1801010322864045, + "grad_norm": 2.436126470565796, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7353321313858032, + "num_tokens": 273942957.0, + "step": 10746 + }, + { + "epoch": 1.1802108499890183, + "grad_norm": 1.9881633520126343, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7014397382736206, + "num_tokens": 273976329.0, + "step": 10747 + }, + { + "epoch": 1.1803206676916318, + "grad_norm": 1.9487929344177246, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7231467366218567, + "num_tokens": 274007881.0, + "step": 10748 + }, + { + "epoch": 1.1804304853942456, + "grad_norm": 2.481358766555786, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.7546350359916687, + "num_tokens": 274026659.0, + "step": 10749 + }, + { + "epoch": 1.1805403030968593, + "grad_norm": 2.2564938068389893, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7168627381324768, + "num_tokens": 274051683.0, + "step": 10750 + }, + { + "epoch": 1.1806501207994728, + "grad_norm": 2.0272912979125977, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.6988896131515503, + "num_tokens": 274084259.0, + "step": 10751 + }, + { + "epoch": 1.1807599385020866, + "grad_norm": 2.4376416206359863, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.7369976043701172, + "num_tokens": 274106507.0, + "step": 10752 + }, + { + "epoch": 1.1808697562047001, + "grad_norm": 2.295426368713379, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7093643546104431, + "num_tokens": 274130555.0, + "step": 10753 + }, + { + "epoch": 1.180979573907314, + "grad_norm": 2.237422466278076, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.702177882194519, + "num_tokens": 274158067.0, + "step": 10754 + }, + { + "epoch": 1.1810893916099274, + "grad_norm": 2.0961334705352783, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7078048586845398, + "num_tokens": 274187019.0, + "step": 10755 + }, + { + "epoch": 1.1811992093125412, + "grad_norm": 2.3665544986724854, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7261593341827393, + "num_tokens": 274212750.0, + "step": 10756 + }, + { + "epoch": 1.1813090270151547, + "grad_norm": 2.141310691833496, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7240875959396362, + "num_tokens": 274240642.0, + "step": 10757 + }, + { + "epoch": 1.1814188447177685, + "grad_norm": 2.288539171218872, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7377198338508606, + "num_tokens": 274265537.0, + "step": 10758 + }, + { + "epoch": 1.1815286624203822, + "grad_norm": 1.9729434251785278, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7342462539672852, + "num_tokens": 274295163.0, + "step": 10759 + }, + { + "epoch": 1.1816384801229958, + "grad_norm": 2.163914918899536, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.6944557428359985, + "num_tokens": 274322887.0, + "step": 10760 + }, + { + "epoch": 1.1817482978256095, + "grad_norm": 2.217299461364746, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7045671939849854, + "num_tokens": 274350241.0, + "step": 10761 + }, + { + "epoch": 1.181858115528223, + "grad_norm": 1.9472342729568481, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7171273231506348, + "num_tokens": 274382199.0, + "step": 10762 + }, + { + "epoch": 1.1819679332308368, + "grad_norm": 2.508382558822632, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.720931887626648, + "num_tokens": 274405118.0, + "step": 10763 + }, + { + "epoch": 1.1820777509334506, + "grad_norm": 2.2390763759613037, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7264324426651001, + "num_tokens": 274428993.0, + "step": 10764 + }, + { + "epoch": 1.1821875686360641, + "grad_norm": 2.2793517112731934, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7077773809432983, + "num_tokens": 274455031.0, + "step": 10765 + }, + { + "epoch": 1.1822973863386779, + "grad_norm": 2.30442476272583, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7352585792541504, + "num_tokens": 274478287.0, + "step": 10766 + }, + { + "epoch": 1.1824072040412914, + "grad_norm": 2.5158588886260986, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7320954203605652, + "num_tokens": 274498726.0, + "step": 10767 + }, + { + "epoch": 1.1825170217439052, + "grad_norm": 2.2547554969787598, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7127941250801086, + "num_tokens": 274523308.0, + "step": 10768 + }, + { + "epoch": 1.1826268394465187, + "grad_norm": 2.245093822479248, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7305596470832825, + "num_tokens": 274546469.0, + "step": 10769 + }, + { + "epoch": 1.1827366571491325, + "grad_norm": 2.1450655460357666, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7230844497680664, + "num_tokens": 274573988.0, + "step": 10770 + }, + { + "epoch": 1.182846474851746, + "grad_norm": 2.2770557403564453, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7322567701339722, + "num_tokens": 274599222.0, + "step": 10771 + }, + { + "epoch": 1.1829562925543597, + "grad_norm": 2.0090532302856445, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7286567091941833, + "num_tokens": 274628610.0, + "step": 10772 + }, + { + "epoch": 1.1830661102569735, + "grad_norm": 2.543841600418091, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7099980115890503, + "num_tokens": 274648330.0, + "step": 10773 + }, + { + "epoch": 1.183175927959587, + "grad_norm": 2.2912790775299072, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7160266637802124, + "num_tokens": 274677176.0, + "step": 10774 + }, + { + "epoch": 1.1832857456622008, + "grad_norm": 2.402967929840088, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7057620286941528, + "num_tokens": 274699856.0, + "step": 10775 + }, + { + "epoch": 1.1833955633648143, + "grad_norm": 1.9174598455429077, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6947680115699768, + "num_tokens": 274735062.0, + "step": 10776 + }, + { + "epoch": 1.183505381067428, + "grad_norm": 2.353240489959717, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7109150886535645, + "num_tokens": 274758833.0, + "step": 10777 + }, + { + "epoch": 1.1836151987700418, + "grad_norm": 2.403813123703003, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7024675011634827, + "num_tokens": 274782294.0, + "step": 10778 + }, + { + "epoch": 1.1837250164726554, + "grad_norm": 2.2283308506011963, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7280350923538208, + "num_tokens": 274808040.0, + "step": 10779 + }, + { + "epoch": 1.1838348341752691, + "grad_norm": 2.004051685333252, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7214677333831787, + "num_tokens": 274839277.0, + "step": 10780 + }, + { + "epoch": 1.1839446518778827, + "grad_norm": 2.318821430206299, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7039758563041687, + "num_tokens": 274864270.0, + "step": 10781 + }, + { + "epoch": 1.1840544695804964, + "grad_norm": 1.9888945817947388, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6898180246353149, + "num_tokens": 274897933.0, + "step": 10782 + }, + { + "epoch": 1.18416428728311, + "grad_norm": 2.272580862045288, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.6969364285469055, + "num_tokens": 274924525.0, + "step": 10783 + }, + { + "epoch": 1.1842741049857237, + "grad_norm": 2.5901050567626953, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7080200910568237, + "num_tokens": 274945172.0, + "step": 10784 + }, + { + "epoch": 1.1843839226883373, + "grad_norm": 2.1399528980255127, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7232967615127563, + "num_tokens": 274971575.0, + "step": 10785 + }, + { + "epoch": 1.184493740390951, + "grad_norm": 2.422740936279297, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7247521877288818, + "num_tokens": 274993572.0, + "step": 10786 + }, + { + "epoch": 1.1846035580935648, + "grad_norm": 1.9616146087646484, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.726915180683136, + "num_tokens": 275026761.0, + "step": 10787 + }, + { + "epoch": 1.1847133757961783, + "grad_norm": 2.2524607181549072, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7144064903259277, + "num_tokens": 275052293.0, + "step": 10788 + }, + { + "epoch": 1.184823193498792, + "grad_norm": 2.3393332958221436, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7043483257293701, + "num_tokens": 275077168.0, + "step": 10789 + }, + { + "epoch": 1.1849330112014056, + "grad_norm": 2.425767421722412, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7352619767189026, + "num_tokens": 275099675.0, + "step": 10790 + }, + { + "epoch": 1.1850428289040194, + "grad_norm": 2.070822238922119, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.715904712677002, + "num_tokens": 275128828.0, + "step": 10791 + }, + { + "epoch": 1.185152646606633, + "grad_norm": 2.366527795791626, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7298800349235535, + "num_tokens": 275151068.0, + "step": 10792 + }, + { + "epoch": 1.1852624643092466, + "grad_norm": 2.2300384044647217, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.707150936126709, + "num_tokens": 275176270.0, + "step": 10793 + }, + { + "epoch": 1.1853722820118604, + "grad_norm": 2.3319125175476074, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7138607501983643, + "num_tokens": 275200065.0, + "step": 10794 + }, + { + "epoch": 1.185482099714474, + "grad_norm": 2.349595069885254, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7246929407119751, + "num_tokens": 275225508.0, + "step": 10795 + }, + { + "epoch": 1.1855919174170877, + "grad_norm": 2.189091205596924, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.715376079082489, + "num_tokens": 275250928.0, + "step": 10796 + }, + { + "epoch": 1.1857017351197012, + "grad_norm": 2.389599084854126, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7307509183883667, + "num_tokens": 275275422.0, + "step": 10797 + }, + { + "epoch": 1.185811552822315, + "grad_norm": 2.1653640270233154, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7251087427139282, + "num_tokens": 275302564.0, + "step": 10798 + }, + { + "epoch": 1.1859213705249285, + "grad_norm": 2.254126787185669, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7063244581222534, + "num_tokens": 275328828.0, + "step": 10799 + }, + { + "epoch": 1.1860311882275423, + "grad_norm": 2.2564175128936768, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7163645625114441, + "num_tokens": 275356623.0, + "step": 10800 + }, + { + "epoch": 1.186141005930156, + "grad_norm": 2.20683217048645, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7329545021057129, + "num_tokens": 275383511.0, + "step": 10801 + }, + { + "epoch": 1.1862508236327696, + "grad_norm": 2.196972370147705, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.6974928379058838, + "num_tokens": 275413556.0, + "step": 10802 + }, + { + "epoch": 1.1863606413353833, + "grad_norm": 2.403557538986206, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.722515881061554, + "num_tokens": 275436465.0, + "step": 10803 + }, + { + "epoch": 1.1864704590379969, + "grad_norm": 2.0426104068756104, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.6977992057800293, + "num_tokens": 275466889.0, + "step": 10804 + }, + { + "epoch": 1.1865802767406106, + "grad_norm": 2.320878505706787, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7368611693382263, + "num_tokens": 275491915.0, + "step": 10805 + }, + { + "epoch": 1.1866900944432242, + "grad_norm": 2.362265110015869, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7341766357421875, + "num_tokens": 275513856.0, + "step": 10806 + }, + { + "epoch": 1.186799912145838, + "grad_norm": 2.041030168533325, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7338963150978088, + "num_tokens": 275544703.0, + "step": 10807 + }, + { + "epoch": 1.1869097298484517, + "grad_norm": 2.143747329711914, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7125867605209351, + "num_tokens": 275573345.0, + "step": 10808 + }, + { + "epoch": 1.1870195475510652, + "grad_norm": 2.4549472332000732, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.7438133955001831, + "num_tokens": 275596038.0, + "step": 10809 + }, + { + "epoch": 1.187129365253679, + "grad_norm": 2.280778646469116, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7231919765472412, + "num_tokens": 275621651.0, + "step": 10810 + }, + { + "epoch": 1.1872391829562925, + "grad_norm": 2.086915969848633, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7175149321556091, + "num_tokens": 275650327.0, + "step": 10811 + }, + { + "epoch": 1.1873490006589063, + "grad_norm": 2.206568717956543, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.705059289932251, + "num_tokens": 275678005.0, + "step": 10812 + }, + { + "epoch": 1.1874588183615198, + "grad_norm": 2.3717586994171143, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7335096597671509, + "num_tokens": 275701759.0, + "step": 10813 + }, + { + "epoch": 1.1875686360641335, + "grad_norm": 2.2973546981811523, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7351846694946289, + "num_tokens": 275725673.0, + "step": 10814 + }, + { + "epoch": 1.1876784537667473, + "grad_norm": 2.44303822517395, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7182567715644836, + "num_tokens": 275747316.0, + "step": 10815 + }, + { + "epoch": 1.1877882714693608, + "grad_norm": 2.4396064281463623, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7419060468673706, + "num_tokens": 275769284.0, + "step": 10816 + }, + { + "epoch": 1.1878980891719746, + "grad_norm": 2.53989577293396, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7130745649337769, + "num_tokens": 275790307.0, + "step": 10817 + }, + { + "epoch": 1.1880079068745881, + "grad_norm": 2.342909097671509, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7235522866249084, + "num_tokens": 275815006.0, + "step": 10818 + }, + { + "epoch": 1.1881177245772019, + "grad_norm": 2.3747596740722656, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7239146828651428, + "num_tokens": 275838924.0, + "step": 10819 + }, + { + "epoch": 1.1882275422798154, + "grad_norm": 2.3436050415039062, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7423415184020996, + "num_tokens": 275861600.0, + "step": 10820 + }, + { + "epoch": 1.1883373599824292, + "grad_norm": 1.866037130355835, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7112352848052979, + "num_tokens": 275896561.0, + "step": 10821 + }, + { + "epoch": 1.1884471776850427, + "grad_norm": 2.134697914123535, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.727946400642395, + "num_tokens": 275922751.0, + "step": 10822 + }, + { + "epoch": 1.1885569953876565, + "grad_norm": 2.428114175796509, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7375532388687134, + "num_tokens": 275944308.0, + "step": 10823 + }, + { + "epoch": 1.1886668130902702, + "grad_norm": 2.059434175491333, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7071207761764526, + "num_tokens": 275974553.0, + "step": 10824 + }, + { + "epoch": 1.1887766307928838, + "grad_norm": 2.2397594451904297, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7430667877197266, + "num_tokens": 275998289.0, + "step": 10825 + }, + { + "epoch": 1.1888864484954975, + "grad_norm": 2.4461565017700195, + "learning_rate": 1e-06, + "loss": 0.7851, + "mean_token_accuracy": 0.746192216873169, + "num_tokens": 276018034.0, + "step": 10826 + }, + { + "epoch": 1.188996266198111, + "grad_norm": 2.417534589767456, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7035438418388367, + "num_tokens": 276040373.0, + "step": 10827 + }, + { + "epoch": 1.1891060839007248, + "grad_norm": 2.698064088821411, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.741923451423645, + "num_tokens": 276059324.0, + "step": 10828 + }, + { + "epoch": 1.1892159016033386, + "grad_norm": 2.386781692504883, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7424684762954712, + "num_tokens": 276081615.0, + "step": 10829 + }, + { + "epoch": 1.189325719305952, + "grad_norm": 1.9357497692108154, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7048999667167664, + "num_tokens": 276114006.0, + "step": 10830 + }, + { + "epoch": 1.1894355370085659, + "grad_norm": 1.9659147262573242, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.711735725402832, + "num_tokens": 276147061.0, + "step": 10831 + }, + { + "epoch": 1.1895453547111794, + "grad_norm": 2.355300188064575, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.6918858289718628, + "num_tokens": 276171305.0, + "step": 10832 + }, + { + "epoch": 1.1896551724137931, + "grad_norm": 2.00344181060791, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7162561416625977, + "num_tokens": 276204889.0, + "step": 10833 + }, + { + "epoch": 1.1897649901164067, + "grad_norm": 2.294478416442871, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.716913104057312, + "num_tokens": 276228382.0, + "step": 10834 + }, + { + "epoch": 1.1898748078190204, + "grad_norm": 2.3633298873901367, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7138600945472717, + "num_tokens": 276251777.0, + "step": 10835 + }, + { + "epoch": 1.189984625521634, + "grad_norm": 2.054069757461548, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.6896630525588989, + "num_tokens": 276284758.0, + "step": 10836 + }, + { + "epoch": 1.1900944432242477, + "grad_norm": 2.3069841861724854, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.723615288734436, + "num_tokens": 276311692.0, + "step": 10837 + }, + { + "epoch": 1.1902042609268615, + "grad_norm": 2.5153887271881104, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7267840504646301, + "num_tokens": 276332672.0, + "step": 10838 + }, + { + "epoch": 1.190314078629475, + "grad_norm": 2.178596019744873, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7095355987548828, + "num_tokens": 276359631.0, + "step": 10839 + }, + { + "epoch": 1.1904238963320888, + "grad_norm": 2.158863067626953, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.708203911781311, + "num_tokens": 276388332.0, + "step": 10840 + }, + { + "epoch": 1.1905337140347023, + "grad_norm": 2.149806022644043, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7151334285736084, + "num_tokens": 276415580.0, + "step": 10841 + }, + { + "epoch": 1.190643531737316, + "grad_norm": 2.3024401664733887, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7208741903305054, + "num_tokens": 276439331.0, + "step": 10842 + }, + { + "epoch": 1.1907533494399298, + "grad_norm": 2.2789816856384277, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7110379934310913, + "num_tokens": 276463871.0, + "step": 10843 + }, + { + "epoch": 1.1908631671425434, + "grad_norm": 2.1448898315429688, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.7000637650489807, + "num_tokens": 276492804.0, + "step": 10844 + }, + { + "epoch": 1.1909729848451571, + "grad_norm": 2.0510122776031494, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7178624868392944, + "num_tokens": 276522542.0, + "step": 10845 + }, + { + "epoch": 1.1910828025477707, + "grad_norm": 2.099989891052246, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.745465099811554, + "num_tokens": 276548448.0, + "step": 10846 + }, + { + "epoch": 1.1911926202503844, + "grad_norm": 2.0859851837158203, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7242699265480042, + "num_tokens": 276578041.0, + "step": 10847 + }, + { + "epoch": 1.191302437952998, + "grad_norm": 2.1834781169891357, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7361645102500916, + "num_tokens": 276604653.0, + "step": 10848 + }, + { + "epoch": 1.1914122556556117, + "grad_norm": 2.361297607421875, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7143380641937256, + "num_tokens": 276628415.0, + "step": 10849 + }, + { + "epoch": 1.1915220733582252, + "grad_norm": 2.175917863845825, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7405051589012146, + "num_tokens": 276653264.0, + "step": 10850 + }, + { + "epoch": 1.191631891060839, + "grad_norm": 2.176579475402832, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7077437043190002, + "num_tokens": 276682631.0, + "step": 10851 + }, + { + "epoch": 1.1917417087634528, + "grad_norm": 2.312086343765259, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7079464197158813, + "num_tokens": 276707803.0, + "step": 10852 + }, + { + "epoch": 1.1918515264660663, + "grad_norm": 2.3706772327423096, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7308278679847717, + "num_tokens": 276731453.0, + "step": 10853 + }, + { + "epoch": 1.19196134416868, + "grad_norm": 2.14912748336792, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7182310223579407, + "num_tokens": 276759611.0, + "step": 10854 + }, + { + "epoch": 1.1920711618712936, + "grad_norm": 1.9462692737579346, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7268466353416443, + "num_tokens": 276788128.0, + "step": 10855 + }, + { + "epoch": 1.1921809795739073, + "grad_norm": 2.224257230758667, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7264159917831421, + "num_tokens": 276815020.0, + "step": 10856 + }, + { + "epoch": 1.192290797276521, + "grad_norm": 2.067396640777588, + "learning_rate": 1e-06, + "loss": 1.0319, + "mean_token_accuracy": 0.682633638381958, + "num_tokens": 276845820.0, + "step": 10857 + }, + { + "epoch": 1.1924006149791346, + "grad_norm": 2.388521671295166, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7135202884674072, + "num_tokens": 276868925.0, + "step": 10858 + }, + { + "epoch": 1.1925104326817484, + "grad_norm": 2.4244871139526367, + "learning_rate": 1e-06, + "loss": 0.811, + "mean_token_accuracy": 0.7502114772796631, + "num_tokens": 276889011.0, + "step": 10859 + }, + { + "epoch": 1.192620250384362, + "grad_norm": 2.3338406085968018, + "learning_rate": 1e-06, + "loss": 1.0294, + "mean_token_accuracy": 0.6830011606216431, + "num_tokens": 276915895.0, + "step": 10860 + }, + { + "epoch": 1.1927300680869757, + "grad_norm": 2.285973310470581, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7180366516113281, + "num_tokens": 276940346.0, + "step": 10861 + }, + { + "epoch": 1.1928398857895892, + "grad_norm": 2.133146047592163, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.6994268894195557, + "num_tokens": 276967537.0, + "step": 10862 + }, + { + "epoch": 1.192949703492203, + "grad_norm": 2.462580919265747, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7136474847793579, + "num_tokens": 276990004.0, + "step": 10863 + }, + { + "epoch": 1.1930595211948165, + "grad_norm": 2.3816802501678467, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7397757768630981, + "num_tokens": 277012282.0, + "step": 10864 + }, + { + "epoch": 1.1931693388974303, + "grad_norm": 2.1477291584014893, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7311639189720154, + "num_tokens": 277037649.0, + "step": 10865 + }, + { + "epoch": 1.193279156600044, + "grad_norm": 2.045257329940796, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7098987102508545, + "num_tokens": 277068313.0, + "step": 10866 + }, + { + "epoch": 1.1933889743026576, + "grad_norm": 2.1162664890289307, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6946840286254883, + "num_tokens": 277098739.0, + "step": 10867 + }, + { + "epoch": 1.1934987920052713, + "grad_norm": 2.22094988822937, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7112237215042114, + "num_tokens": 277127742.0, + "step": 10868 + }, + { + "epoch": 1.1936086097078848, + "grad_norm": 2.4389896392822266, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7223215103149414, + "num_tokens": 277148493.0, + "step": 10869 + }, + { + "epoch": 1.1937184274104986, + "grad_norm": 2.2415475845336914, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7223286628723145, + "num_tokens": 277174981.0, + "step": 10870 + }, + { + "epoch": 1.1938282451131121, + "grad_norm": 2.2486703395843506, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.6908338069915771, + "num_tokens": 277200677.0, + "step": 10871 + }, + { + "epoch": 1.193938062815726, + "grad_norm": 2.32462215423584, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7188267707824707, + "num_tokens": 277223092.0, + "step": 10872 + }, + { + "epoch": 1.1940478805183394, + "grad_norm": 2.186318874359131, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7184540033340454, + "num_tokens": 277250173.0, + "step": 10873 + }, + { + "epoch": 1.1941576982209532, + "grad_norm": 2.031485080718994, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7082939147949219, + "num_tokens": 277281183.0, + "step": 10874 + }, + { + "epoch": 1.194267515923567, + "grad_norm": 2.3071773052215576, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7086279988288879, + "num_tokens": 277307295.0, + "step": 10875 + }, + { + "epoch": 1.1943773336261805, + "grad_norm": 2.577284097671509, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7073646783828735, + "num_tokens": 277327722.0, + "step": 10876 + }, + { + "epoch": 1.1944871513287942, + "grad_norm": 2.2362799644470215, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7245845794677734, + "num_tokens": 277353051.0, + "step": 10877 + }, + { + "epoch": 1.1945969690314078, + "grad_norm": 2.3409194946289062, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7185465097427368, + "num_tokens": 277376815.0, + "step": 10878 + }, + { + "epoch": 1.1947067867340215, + "grad_norm": 2.1284737586975098, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7111936807632446, + "num_tokens": 277405624.0, + "step": 10879 + }, + { + "epoch": 1.1948166044366353, + "grad_norm": 2.4085195064544678, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7185937166213989, + "num_tokens": 277429746.0, + "step": 10880 + }, + { + "epoch": 1.1949264221392488, + "grad_norm": 2.205833673477173, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6953032612800598, + "num_tokens": 277458631.0, + "step": 10881 + }, + { + "epoch": 1.1950362398418626, + "grad_norm": 2.2529048919677734, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7339069843292236, + "num_tokens": 277481677.0, + "step": 10882 + }, + { + "epoch": 1.1951460575444761, + "grad_norm": 2.5319364070892334, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7227048873901367, + "num_tokens": 277502611.0, + "step": 10883 + }, + { + "epoch": 1.1952558752470899, + "grad_norm": 2.316709518432617, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.721264123916626, + "num_tokens": 277526486.0, + "step": 10884 + }, + { + "epoch": 1.1953656929497034, + "grad_norm": 2.4946486949920654, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.71226966381073, + "num_tokens": 277548529.0, + "step": 10885 + }, + { + "epoch": 1.1954755106523172, + "grad_norm": 2.1120777130126953, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7255088090896606, + "num_tokens": 277574996.0, + "step": 10886 + }, + { + "epoch": 1.1955853283549307, + "grad_norm": 2.875685453414917, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7351605892181396, + "num_tokens": 277592193.0, + "step": 10887 + }, + { + "epoch": 1.1956951460575445, + "grad_norm": 2.2003190517425537, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7110027074813843, + "num_tokens": 277619719.0, + "step": 10888 + }, + { + "epoch": 1.1958049637601582, + "grad_norm": 2.142892599105835, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7210797667503357, + "num_tokens": 277645715.0, + "step": 10889 + }, + { + "epoch": 1.1959147814627717, + "grad_norm": 2.0964601039886475, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7228800654411316, + "num_tokens": 277672795.0, + "step": 10890 + }, + { + "epoch": 1.1960245991653855, + "grad_norm": 2.1851425170898438, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7514396905899048, + "num_tokens": 277700497.0, + "step": 10891 + }, + { + "epoch": 1.196134416867999, + "grad_norm": 2.2832744121551514, + "learning_rate": 1e-06, + "loss": 0.7782, + "mean_token_accuracy": 0.748363196849823, + "num_tokens": 277722934.0, + "step": 10892 + }, + { + "epoch": 1.1962442345706128, + "grad_norm": 2.373734951019287, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7307989597320557, + "num_tokens": 277744629.0, + "step": 10893 + }, + { + "epoch": 1.1963540522732266, + "grad_norm": 2.117093801498413, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7079784274101257, + "num_tokens": 277772947.0, + "step": 10894 + }, + { + "epoch": 1.19646386997584, + "grad_norm": 2.397343635559082, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7149467468261719, + "num_tokens": 277797166.0, + "step": 10895 + }, + { + "epoch": 1.1965736876784538, + "grad_norm": 2.1882641315460205, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7172068357467651, + "num_tokens": 277821195.0, + "step": 10896 + }, + { + "epoch": 1.1966835053810674, + "grad_norm": 2.2183775901794434, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7196837663650513, + "num_tokens": 277846728.0, + "step": 10897 + }, + { + "epoch": 1.1967933230836811, + "grad_norm": 2.195861577987671, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7279454469680786, + "num_tokens": 277870319.0, + "step": 10898 + }, + { + "epoch": 1.1969031407862947, + "grad_norm": 1.9910539388656616, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7023042440414429, + "num_tokens": 277902625.0, + "step": 10899 + }, + { + "epoch": 1.1970129584889084, + "grad_norm": 2.187222719192505, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.72649747133255, + "num_tokens": 277928991.0, + "step": 10900 + }, + { + "epoch": 1.197122776191522, + "grad_norm": 2.1509745121002197, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7020198106765747, + "num_tokens": 277954499.0, + "step": 10901 + }, + { + "epoch": 1.1972325938941357, + "grad_norm": 2.3682539463043213, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7560079097747803, + "num_tokens": 277976163.0, + "step": 10902 + }, + { + "epoch": 1.1973424115967495, + "grad_norm": 2.33065128326416, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7395257949829102, + "num_tokens": 277999614.0, + "step": 10903 + }, + { + "epoch": 1.197452229299363, + "grad_norm": 2.45285701751709, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7078089714050293, + "num_tokens": 278021734.0, + "step": 10904 + }, + { + "epoch": 1.1975620470019768, + "grad_norm": 2.4152681827545166, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7248767614364624, + "num_tokens": 278043645.0, + "step": 10905 + }, + { + "epoch": 1.1976718647045903, + "grad_norm": 2.762056350708008, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7257200479507446, + "num_tokens": 278063192.0, + "step": 10906 + }, + { + "epoch": 1.197781682407204, + "grad_norm": 2.061028003692627, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7325058579444885, + "num_tokens": 278091165.0, + "step": 10907 + }, + { + "epoch": 1.1978915001098178, + "grad_norm": 2.506720781326294, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7262356877326965, + "num_tokens": 278113039.0, + "step": 10908 + }, + { + "epoch": 1.1980013178124314, + "grad_norm": 2.2619569301605225, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7219287753105164, + "num_tokens": 278139640.0, + "step": 10909 + }, + { + "epoch": 1.1981111355150451, + "grad_norm": 2.3249752521514893, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.6973350048065186, + "num_tokens": 278163346.0, + "step": 10910 + }, + { + "epoch": 1.1982209532176586, + "grad_norm": 2.251955032348633, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7338738441467285, + "num_tokens": 278189288.0, + "step": 10911 + }, + { + "epoch": 1.1983307709202724, + "grad_norm": 1.9536956548690796, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7169244289398193, + "num_tokens": 278221813.0, + "step": 10912 + }, + { + "epoch": 1.198440588622886, + "grad_norm": 2.40401554107666, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7281495332717896, + "num_tokens": 278244853.0, + "step": 10913 + }, + { + "epoch": 1.1985504063254997, + "grad_norm": 1.9746297597885132, + "learning_rate": 1e-06, + "loss": 1.0114, + "mean_token_accuracy": 0.6988910436630249, + "num_tokens": 278276337.0, + "step": 10914 + }, + { + "epoch": 1.1986602240281132, + "grad_norm": 1.94016695022583, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7050936222076416, + "num_tokens": 278310729.0, + "step": 10915 + }, + { + "epoch": 1.198770041730727, + "grad_norm": 2.2571921348571777, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7271946668624878, + "num_tokens": 278335282.0, + "step": 10916 + }, + { + "epoch": 1.1988798594333407, + "grad_norm": 2.1970536708831787, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7251425981521606, + "num_tokens": 278361186.0, + "step": 10917 + }, + { + "epoch": 1.1989896771359543, + "grad_norm": 2.151336669921875, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7140012979507446, + "num_tokens": 278389481.0, + "step": 10918 + }, + { + "epoch": 1.199099494838568, + "grad_norm": 2.9282827377319336, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7281755208969116, + "num_tokens": 278406108.0, + "step": 10919 + }, + { + "epoch": 1.1992093125411816, + "grad_norm": 2.186055898666382, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7045665383338928, + "num_tokens": 278433771.0, + "step": 10920 + }, + { + "epoch": 1.1993191302437953, + "grad_norm": 2.022801160812378, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7194536328315735, + "num_tokens": 278462546.0, + "step": 10921 + }, + { + "epoch": 1.199428947946409, + "grad_norm": 2.263437509536743, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7216897010803223, + "num_tokens": 278487354.0, + "step": 10922 + }, + { + "epoch": 1.1995387656490226, + "grad_norm": 2.142216444015503, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7186156511306763, + "num_tokens": 278515054.0, + "step": 10923 + }, + { + "epoch": 1.1996485833516364, + "grad_norm": 2.2261905670166016, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.736164927482605, + "num_tokens": 278538410.0, + "step": 10924 + }, + { + "epoch": 1.19975840105425, + "grad_norm": 2.4266116619110107, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7330648899078369, + "num_tokens": 278561663.0, + "step": 10925 + }, + { + "epoch": 1.1998682187568637, + "grad_norm": 2.0701608657836914, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7358255386352539, + "num_tokens": 278590433.0, + "step": 10926 + }, + { + "epoch": 1.1999780364594772, + "grad_norm": 2.118317127227783, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.747994065284729, + "num_tokens": 278617951.0, + "step": 10927 + }, + { + "epoch": 1.200087854162091, + "grad_norm": 2.397149085998535, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7063362002372742, + "num_tokens": 278642485.0, + "step": 10928 + }, + { + "epoch": 1.2001976718647045, + "grad_norm": 2.0569379329681396, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7091568112373352, + "num_tokens": 278672614.0, + "step": 10929 + }, + { + "epoch": 1.2003074895673183, + "grad_norm": 2.2173359394073486, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.6992686986923218, + "num_tokens": 278698799.0, + "step": 10930 + }, + { + "epoch": 1.200417307269932, + "grad_norm": 2.6673378944396973, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7322980761528015, + "num_tokens": 278718242.0, + "step": 10931 + }, + { + "epoch": 1.2005271249725455, + "grad_norm": 2.3700947761535645, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7279470562934875, + "num_tokens": 278741132.0, + "step": 10932 + }, + { + "epoch": 1.2006369426751593, + "grad_norm": 2.236189126968384, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7207038402557373, + "num_tokens": 278767387.0, + "step": 10933 + }, + { + "epoch": 1.2007467603777728, + "grad_norm": 2.1874747276306152, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7136040329933167, + "num_tokens": 278794654.0, + "step": 10934 + }, + { + "epoch": 1.2008565780803866, + "grad_norm": 2.601457357406616, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6969177722930908, + "num_tokens": 278820662.0, + "step": 10935 + }, + { + "epoch": 1.2009663957830001, + "grad_norm": 2.1254613399505615, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.703021764755249, + "num_tokens": 278849630.0, + "step": 10936 + }, + { + "epoch": 1.2010762134856139, + "grad_norm": 2.517271041870117, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7310478091239929, + "num_tokens": 278870378.0, + "step": 10937 + }, + { + "epoch": 1.2011860311882274, + "grad_norm": 2.362372398376465, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7178663015365601, + "num_tokens": 278894008.0, + "step": 10938 + }, + { + "epoch": 1.2012958488908412, + "grad_norm": 2.117705821990967, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7329553961753845, + "num_tokens": 278920620.0, + "step": 10939 + }, + { + "epoch": 1.201405666593455, + "grad_norm": 2.085376024246216, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.697835385799408, + "num_tokens": 278952107.0, + "step": 10940 + }, + { + "epoch": 1.2015154842960685, + "grad_norm": 2.455273389816284, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7068042755126953, + "num_tokens": 278975932.0, + "step": 10941 + }, + { + "epoch": 1.2016253019986822, + "grad_norm": 2.031815528869629, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7096587419509888, + "num_tokens": 279008271.0, + "step": 10942 + }, + { + "epoch": 1.2017351197012958, + "grad_norm": 2.4519753456115723, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7339602708816528, + "num_tokens": 279029619.0, + "step": 10943 + }, + { + "epoch": 1.2018449374039095, + "grad_norm": 2.2180678844451904, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7302287817001343, + "num_tokens": 279055004.0, + "step": 10944 + }, + { + "epoch": 1.2019547551065233, + "grad_norm": 2.6686747074127197, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7239689230918884, + "num_tokens": 279075228.0, + "step": 10945 + }, + { + "epoch": 1.2020645728091368, + "grad_norm": 2.30543851852417, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7260669469833374, + "num_tokens": 279098296.0, + "step": 10946 + }, + { + "epoch": 1.2021743905117506, + "grad_norm": 2.2889623641967773, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7413880825042725, + "num_tokens": 279122173.0, + "step": 10947 + }, + { + "epoch": 1.202284208214364, + "grad_norm": 2.5595149993896484, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.72398442029953, + "num_tokens": 279144559.0, + "step": 10948 + }, + { + "epoch": 1.2023940259169779, + "grad_norm": 2.2760019302368164, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7102103233337402, + "num_tokens": 279170157.0, + "step": 10949 + }, + { + "epoch": 1.2025038436195914, + "grad_norm": 2.1992413997650146, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7270089387893677, + "num_tokens": 279195731.0, + "step": 10950 + }, + { + "epoch": 1.2026136613222052, + "grad_norm": 2.097898244857788, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7331926226615906, + "num_tokens": 279223536.0, + "step": 10951 + }, + { + "epoch": 1.2027234790248187, + "grad_norm": 2.0046463012695312, + "learning_rate": 1e-06, + "loss": 1.0238, + "mean_token_accuracy": 0.6864317655563354, + "num_tokens": 279255884.0, + "step": 10952 + }, + { + "epoch": 1.2028332967274324, + "grad_norm": 2.3104748725891113, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7198294401168823, + "num_tokens": 279280919.0, + "step": 10953 + }, + { + "epoch": 1.2029431144300462, + "grad_norm": 2.3296091556549072, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7422910928726196, + "num_tokens": 279304808.0, + "step": 10954 + }, + { + "epoch": 1.2030529321326597, + "grad_norm": 2.2901716232299805, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7336251139640808, + "num_tokens": 279331195.0, + "step": 10955 + }, + { + "epoch": 1.2031627498352735, + "grad_norm": 2.2063486576080322, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7176646590232849, + "num_tokens": 279358352.0, + "step": 10956 + }, + { + "epoch": 1.203272567537887, + "grad_norm": 2.1771509647369385, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.707552433013916, + "num_tokens": 279386011.0, + "step": 10957 + }, + { + "epoch": 1.2033823852405008, + "grad_norm": 2.059713840484619, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7105117440223694, + "num_tokens": 279417504.0, + "step": 10958 + }, + { + "epoch": 1.2034922029431145, + "grad_norm": 2.3032398223876953, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7347410321235657, + "num_tokens": 279441149.0, + "step": 10959 + }, + { + "epoch": 1.203602020645728, + "grad_norm": 2.081731081008911, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7054196000099182, + "num_tokens": 279469350.0, + "step": 10960 + }, + { + "epoch": 1.2037118383483418, + "grad_norm": 2.298635482788086, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7338821887969971, + "num_tokens": 279494209.0, + "step": 10961 + }, + { + "epoch": 1.2038216560509554, + "grad_norm": 2.1787867546081543, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.711250901222229, + "num_tokens": 279522221.0, + "step": 10962 + }, + { + "epoch": 1.2039314737535691, + "grad_norm": 2.4402260780334473, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7150496244430542, + "num_tokens": 279543730.0, + "step": 10963 + }, + { + "epoch": 1.2040412914561827, + "grad_norm": 2.412503480911255, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7185053825378418, + "num_tokens": 279565802.0, + "step": 10964 + }, + { + "epoch": 1.2041511091587964, + "grad_norm": 2.372650146484375, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7173614501953125, + "num_tokens": 279588953.0, + "step": 10965 + }, + { + "epoch": 1.20426092686141, + "grad_norm": 2.6788458824157715, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7278669476509094, + "num_tokens": 279609402.0, + "step": 10966 + }, + { + "epoch": 1.2043707445640237, + "grad_norm": 2.450554609298706, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.724433958530426, + "num_tokens": 279631832.0, + "step": 10967 + }, + { + "epoch": 1.2044805622666375, + "grad_norm": 2.421154499053955, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7275218963623047, + "num_tokens": 279654359.0, + "step": 10968 + }, + { + "epoch": 1.204590379969251, + "grad_norm": 2.1662750244140625, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7199981808662415, + "num_tokens": 279681298.0, + "step": 10969 + }, + { + "epoch": 1.2047001976718648, + "grad_norm": 2.3092446327209473, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7037777900695801, + "num_tokens": 279706278.0, + "step": 10970 + }, + { + "epoch": 1.2048100153744783, + "grad_norm": 2.374255657196045, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7331667542457581, + "num_tokens": 279730055.0, + "step": 10971 + }, + { + "epoch": 1.204919833077092, + "grad_norm": 2.3310604095458984, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7246156334877014, + "num_tokens": 279754395.0, + "step": 10972 + }, + { + "epoch": 1.2050296507797058, + "grad_norm": 2.1307456493377686, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7337088584899902, + "num_tokens": 279781911.0, + "step": 10973 + }, + { + "epoch": 1.2051394684823193, + "grad_norm": 2.7103354930877686, + "learning_rate": 1e-06, + "loss": 0.7255, + "mean_token_accuracy": 0.7663427591323853, + "num_tokens": 279800781.0, + "step": 10974 + }, + { + "epoch": 1.205249286184933, + "grad_norm": 2.552208185195923, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7355011701583862, + "num_tokens": 279819430.0, + "step": 10975 + }, + { + "epoch": 1.2053591038875466, + "grad_norm": 2.219426393508911, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7261264324188232, + "num_tokens": 279844992.0, + "step": 10976 + }, + { + "epoch": 1.2054689215901604, + "grad_norm": 2.2746849060058594, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7081720232963562, + "num_tokens": 279872287.0, + "step": 10977 + }, + { + "epoch": 1.205578739292774, + "grad_norm": 2.151624917984009, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7003538608551025, + "num_tokens": 279899261.0, + "step": 10978 + }, + { + "epoch": 1.2056885569953877, + "grad_norm": 2.072514533996582, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7053349614143372, + "num_tokens": 279930126.0, + "step": 10979 + }, + { + "epoch": 1.2057983746980012, + "grad_norm": 2.255155324935913, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.725094199180603, + "num_tokens": 279956084.0, + "step": 10980 + }, + { + "epoch": 1.205908192400615, + "grad_norm": 2.0030581951141357, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7305943965911865, + "num_tokens": 279985241.0, + "step": 10981 + }, + { + "epoch": 1.2060180101032287, + "grad_norm": 2.169273853302002, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7125641107559204, + "num_tokens": 280011928.0, + "step": 10982 + }, + { + "epoch": 1.2061278278058423, + "grad_norm": 2.255373477935791, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7167750597000122, + "num_tokens": 280037878.0, + "step": 10983 + }, + { + "epoch": 1.206237645508456, + "grad_norm": 2.0934083461761475, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7126094698905945, + "num_tokens": 280065589.0, + "step": 10984 + }, + { + "epoch": 1.2063474632110696, + "grad_norm": 2.37149715423584, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7238706350326538, + "num_tokens": 280088851.0, + "step": 10985 + }, + { + "epoch": 1.2064572809136833, + "grad_norm": 2.2636168003082275, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7118691802024841, + "num_tokens": 280113457.0, + "step": 10986 + }, + { + "epoch": 1.2065670986162969, + "grad_norm": 2.2923965454101562, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7271246910095215, + "num_tokens": 280138131.0, + "step": 10987 + }, + { + "epoch": 1.2066769163189106, + "grad_norm": 2.0751326084136963, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7332859039306641, + "num_tokens": 280164979.0, + "step": 10988 + }, + { + "epoch": 1.2067867340215244, + "grad_norm": 2.408596992492676, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.735595166683197, + "num_tokens": 280186096.0, + "step": 10989 + }, + { + "epoch": 1.206896551724138, + "grad_norm": 2.5096638202667236, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7184168100357056, + "num_tokens": 280208354.0, + "step": 10990 + }, + { + "epoch": 1.2070063694267517, + "grad_norm": 2.43449330329895, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7174456119537354, + "num_tokens": 280230636.0, + "step": 10991 + }, + { + "epoch": 1.2071161871293652, + "grad_norm": 2.272233486175537, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7089368104934692, + "num_tokens": 280254759.0, + "step": 10992 + }, + { + "epoch": 1.207226004831979, + "grad_norm": 2.3942346572875977, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7288022041320801, + "num_tokens": 280276926.0, + "step": 10993 + }, + { + "epoch": 1.2073358225345925, + "grad_norm": 2.2258458137512207, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.720672607421875, + "num_tokens": 280302720.0, + "step": 10994 + }, + { + "epoch": 1.2074456402372062, + "grad_norm": 2.4819703102111816, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7305006980895996, + "num_tokens": 280326566.0, + "step": 10995 + }, + { + "epoch": 1.20755545793982, + "grad_norm": 2.554046154022217, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7422232031822205, + "num_tokens": 280347562.0, + "step": 10996 + }, + { + "epoch": 1.2076652756424335, + "grad_norm": 2.304448366165161, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.7096725106239319, + "num_tokens": 280372800.0, + "step": 10997 + }, + { + "epoch": 1.2077750933450473, + "grad_norm": 2.198796033859253, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7213698625564575, + "num_tokens": 280399281.0, + "step": 10998 + }, + { + "epoch": 1.2078849110476608, + "grad_norm": 2.621755599975586, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7331042289733887, + "num_tokens": 280419253.0, + "step": 10999 + }, + { + "epoch": 1.2079947287502746, + "grad_norm": 2.1436593532562256, + "learning_rate": 1e-06, + "loss": 0.7935, + "mean_token_accuracy": 0.7501246333122253, + "num_tokens": 280444643.0, + "step": 11000 + }, + { + "epoch": 1.2081045464528881, + "grad_norm": 2.2593932151794434, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7262779474258423, + "num_tokens": 280469942.0, + "step": 11001 + }, + { + "epoch": 1.2082143641555019, + "grad_norm": 2.3433077335357666, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7245615720748901, + "num_tokens": 280493384.0, + "step": 11002 + }, + { + "epoch": 1.2083241818581154, + "grad_norm": 2.175926446914673, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7207603454589844, + "num_tokens": 280521328.0, + "step": 11003 + }, + { + "epoch": 1.2084339995607292, + "grad_norm": 2.274456739425659, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.6987206935882568, + "num_tokens": 280547177.0, + "step": 11004 + }, + { + "epoch": 1.208543817263343, + "grad_norm": 2.321720600128174, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.717748761177063, + "num_tokens": 280573332.0, + "step": 11005 + }, + { + "epoch": 1.2086536349659565, + "grad_norm": 2.214827299118042, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7303550243377686, + "num_tokens": 280598399.0, + "step": 11006 + }, + { + "epoch": 1.2087634526685702, + "grad_norm": 1.944864273071289, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7094365358352661, + "num_tokens": 280628936.0, + "step": 11007 + }, + { + "epoch": 1.2088732703711838, + "grad_norm": 2.2257485389709473, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7423330545425415, + "num_tokens": 280653025.0, + "step": 11008 + }, + { + "epoch": 1.2089830880737975, + "grad_norm": 2.1957836151123047, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7150769829750061, + "num_tokens": 280681022.0, + "step": 11009 + }, + { + "epoch": 1.2090929057764113, + "grad_norm": 2.4201231002807617, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7304928302764893, + "num_tokens": 280705485.0, + "step": 11010 + }, + { + "epoch": 1.2092027234790248, + "grad_norm": 2.134833812713623, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7371178865432739, + "num_tokens": 280731757.0, + "step": 11011 + }, + { + "epoch": 1.2093125411816386, + "grad_norm": 2.20707368850708, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7220526933670044, + "num_tokens": 280756279.0, + "step": 11012 + }, + { + "epoch": 1.209422358884252, + "grad_norm": 2.1735692024230957, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7163302898406982, + "num_tokens": 280782688.0, + "step": 11013 + }, + { + "epoch": 1.2095321765868658, + "grad_norm": 2.084115982055664, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.711462140083313, + "num_tokens": 280811091.0, + "step": 11014 + }, + { + "epoch": 1.2096419942894794, + "grad_norm": 2.169588088989258, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7083563208580017, + "num_tokens": 280837718.0, + "step": 11015 + }, + { + "epoch": 1.2097518119920931, + "grad_norm": 2.2188873291015625, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7299531698226929, + "num_tokens": 280863139.0, + "step": 11016 + }, + { + "epoch": 1.2098616296947067, + "grad_norm": 2.380265474319458, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7558356523513794, + "num_tokens": 280885238.0, + "step": 11017 + }, + { + "epoch": 1.2099714473973204, + "grad_norm": 2.4056589603424072, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7116794586181641, + "num_tokens": 280907175.0, + "step": 11018 + }, + { + "epoch": 1.2100812650999342, + "grad_norm": 1.9544155597686768, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7122538089752197, + "num_tokens": 280938999.0, + "step": 11019 + }, + { + "epoch": 1.2101910828025477, + "grad_norm": 2.3270277976989746, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7211666703224182, + "num_tokens": 280964544.0, + "step": 11020 + }, + { + "epoch": 1.2103009005051615, + "grad_norm": 2.2449684143066406, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6903090476989746, + "num_tokens": 280992335.0, + "step": 11021 + }, + { + "epoch": 1.210410718207775, + "grad_norm": 2.153533935546875, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.744911789894104, + "num_tokens": 281018219.0, + "step": 11022 + }, + { + "epoch": 1.2105205359103888, + "grad_norm": 2.0791285037994385, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.696979820728302, + "num_tokens": 281046606.0, + "step": 11023 + }, + { + "epoch": 1.2106303536130025, + "grad_norm": 2.4256842136383057, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7207076549530029, + "num_tokens": 281068088.0, + "step": 11024 + }, + { + "epoch": 1.210740171315616, + "grad_norm": 2.2727978229522705, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7268221378326416, + "num_tokens": 281092366.0, + "step": 11025 + }, + { + "epoch": 1.2108499890182298, + "grad_norm": 2.307399034500122, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7181915640830994, + "num_tokens": 281117182.0, + "step": 11026 + }, + { + "epoch": 1.2109598067208434, + "grad_norm": 2.098172664642334, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7085863351821899, + "num_tokens": 281145983.0, + "step": 11027 + }, + { + "epoch": 1.2110696244234571, + "grad_norm": 2.4221439361572266, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7454429268836975, + "num_tokens": 281168518.0, + "step": 11028 + }, + { + "epoch": 1.2111794421260706, + "grad_norm": 2.485208749771118, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7329022884368896, + "num_tokens": 281188367.0, + "step": 11029 + }, + { + "epoch": 1.2112892598286844, + "grad_norm": 2.4689204692840576, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7203273773193359, + "num_tokens": 281211629.0, + "step": 11030 + }, + { + "epoch": 1.211399077531298, + "grad_norm": 1.9134165048599243, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6902605295181274, + "num_tokens": 281246447.0, + "step": 11031 + }, + { + "epoch": 1.2115088952339117, + "grad_norm": 2.353609085083008, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7140673398971558, + "num_tokens": 281270713.0, + "step": 11032 + }, + { + "epoch": 1.2116187129365255, + "grad_norm": 2.2580063343048096, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7387559413909912, + "num_tokens": 281294699.0, + "step": 11033 + }, + { + "epoch": 1.211728530639139, + "grad_norm": 1.9999897480010986, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.702322244644165, + "num_tokens": 281328229.0, + "step": 11034 + }, + { + "epoch": 1.2118383483417527, + "grad_norm": 2.1661462783813477, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7049212455749512, + "num_tokens": 281356027.0, + "step": 11035 + }, + { + "epoch": 1.2119481660443663, + "grad_norm": 2.4499800205230713, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7181152701377869, + "num_tokens": 281379056.0, + "step": 11036 + }, + { + "epoch": 1.21205798374698, + "grad_norm": 2.179690361022949, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7145246267318726, + "num_tokens": 281406472.0, + "step": 11037 + }, + { + "epoch": 1.2121678014495938, + "grad_norm": 2.423060894012451, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7203695178031921, + "num_tokens": 281429810.0, + "step": 11038 + }, + { + "epoch": 1.2122776191522073, + "grad_norm": 2.6638662815093994, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7293366193771362, + "num_tokens": 281449876.0, + "step": 11039 + }, + { + "epoch": 1.212387436854821, + "grad_norm": 2.2325384616851807, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7295993566513062, + "num_tokens": 281475673.0, + "step": 11040 + }, + { + "epoch": 1.2124972545574346, + "grad_norm": 2.3431668281555176, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7244282960891724, + "num_tokens": 281498847.0, + "step": 11041 + }, + { + "epoch": 1.2126070722600484, + "grad_norm": 2.222733974456787, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.706723690032959, + "num_tokens": 281525554.0, + "step": 11042 + }, + { + "epoch": 1.212716889962662, + "grad_norm": 1.9837448596954346, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7300654053688049, + "num_tokens": 281557073.0, + "step": 11043 + }, + { + "epoch": 1.2128267076652757, + "grad_norm": 2.4450912475585938, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7086083889007568, + "num_tokens": 281580564.0, + "step": 11044 + }, + { + "epoch": 1.2129365253678892, + "grad_norm": 2.461652994155884, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7358474731445312, + "num_tokens": 281600708.0, + "step": 11045 + }, + { + "epoch": 1.213046343070503, + "grad_norm": 2.3581061363220215, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.726954996585846, + "num_tokens": 281622070.0, + "step": 11046 + }, + { + "epoch": 1.2131561607731167, + "grad_norm": 2.0179734230041504, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7085686326026917, + "num_tokens": 281651515.0, + "step": 11047 + }, + { + "epoch": 1.2132659784757303, + "grad_norm": 2.189260244369507, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.700921893119812, + "num_tokens": 281678216.0, + "step": 11048 + }, + { + "epoch": 1.213375796178344, + "grad_norm": 2.278290271759033, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7290096282958984, + "num_tokens": 281702660.0, + "step": 11049 + }, + { + "epoch": 1.2134856138809575, + "grad_norm": 3.0352532863616943, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7276570200920105, + "num_tokens": 281717866.0, + "step": 11050 + }, + { + "epoch": 1.2135954315835713, + "grad_norm": 2.4246466159820557, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.6960830688476562, + "num_tokens": 281741637.0, + "step": 11051 + }, + { + "epoch": 1.2137052492861848, + "grad_norm": 2.347111940383911, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7159572839736938, + "num_tokens": 281766876.0, + "step": 11052 + }, + { + "epoch": 1.2138150669887986, + "grad_norm": 2.195383310317993, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.6875342726707458, + "num_tokens": 281795304.0, + "step": 11053 + }, + { + "epoch": 1.2139248846914121, + "grad_norm": 1.9757543802261353, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7287135124206543, + "num_tokens": 281826045.0, + "step": 11054 + }, + { + "epoch": 1.2140347023940259, + "grad_norm": 2.0695691108703613, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7114706039428711, + "num_tokens": 281855512.0, + "step": 11055 + }, + { + "epoch": 1.2141445200966396, + "grad_norm": 2.3917524814605713, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7203611135482788, + "num_tokens": 281878065.0, + "step": 11056 + }, + { + "epoch": 1.2142543377992532, + "grad_norm": 2.3172149658203125, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7277096509933472, + "num_tokens": 281902583.0, + "step": 11057 + }, + { + "epoch": 1.214364155501867, + "grad_norm": 2.3625669479370117, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7039040923118591, + "num_tokens": 281926188.0, + "step": 11058 + }, + { + "epoch": 1.2144739732044805, + "grad_norm": 2.389498472213745, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7088381052017212, + "num_tokens": 281949827.0, + "step": 11059 + }, + { + "epoch": 1.2145837909070942, + "grad_norm": 2.1059958934783936, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6953029632568359, + "num_tokens": 281978382.0, + "step": 11060 + }, + { + "epoch": 1.214693608609708, + "grad_norm": 2.398771286010742, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7230939269065857, + "num_tokens": 282002226.0, + "step": 11061 + }, + { + "epoch": 1.2148034263123215, + "grad_norm": 2.312662124633789, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7115023136138916, + "num_tokens": 282027315.0, + "step": 11062 + }, + { + "epoch": 1.2149132440149353, + "grad_norm": 2.283358097076416, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7276439666748047, + "num_tokens": 282050979.0, + "step": 11063 + }, + { + "epoch": 1.2150230617175488, + "grad_norm": 2.550252676010132, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7343469858169556, + "num_tokens": 282072909.0, + "step": 11064 + }, + { + "epoch": 1.2151328794201626, + "grad_norm": 2.198131799697876, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7288882732391357, + "num_tokens": 282099551.0, + "step": 11065 + }, + { + "epoch": 1.215242697122776, + "grad_norm": 2.3685216903686523, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7129082679748535, + "num_tokens": 282123478.0, + "step": 11066 + }, + { + "epoch": 1.2153525148253899, + "grad_norm": 2.4885950088500977, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7115515470504761, + "num_tokens": 282145555.0, + "step": 11067 + }, + { + "epoch": 1.2154623325280034, + "grad_norm": 2.4720611572265625, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7352975606918335, + "num_tokens": 282168378.0, + "step": 11068 + }, + { + "epoch": 1.2155721502306172, + "grad_norm": 2.2571468353271484, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7087720632553101, + "num_tokens": 282196051.0, + "step": 11069 + }, + { + "epoch": 1.215681967933231, + "grad_norm": 2.7149181365966797, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.722443163394928, + "num_tokens": 282214634.0, + "step": 11070 + }, + { + "epoch": 1.2157917856358444, + "grad_norm": 2.167809247970581, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7072998285293579, + "num_tokens": 282242632.0, + "step": 11071 + }, + { + "epoch": 1.2159016033384582, + "grad_norm": 2.149405002593994, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7155810594558716, + "num_tokens": 282271411.0, + "step": 11072 + }, + { + "epoch": 1.2160114210410717, + "grad_norm": 2.3353805541992188, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7251808047294617, + "num_tokens": 282295631.0, + "step": 11073 + }, + { + "epoch": 1.2161212387436855, + "grad_norm": 2.2157647609710693, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7168225049972534, + "num_tokens": 282322630.0, + "step": 11074 + }, + { + "epoch": 1.2162310564462993, + "grad_norm": 2.568542957305908, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7372277975082397, + "num_tokens": 282344329.0, + "step": 11075 + }, + { + "epoch": 1.2163408741489128, + "grad_norm": 2.3685498237609863, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7185540199279785, + "num_tokens": 282368159.0, + "step": 11076 + }, + { + "epoch": 1.2164506918515265, + "grad_norm": 2.501821994781494, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.7409192323684692, + "num_tokens": 282389136.0, + "step": 11077 + }, + { + "epoch": 1.21656050955414, + "grad_norm": 2.1203885078430176, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7327942252159119, + "num_tokens": 282415625.0, + "step": 11078 + }, + { + "epoch": 1.2166703272567538, + "grad_norm": 2.1766550540924072, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7343053817749023, + "num_tokens": 282441269.0, + "step": 11079 + }, + { + "epoch": 1.2167801449593674, + "grad_norm": 2.6363861560821533, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7212783098220825, + "num_tokens": 282462491.0, + "step": 11080 + }, + { + "epoch": 1.2168899626619811, + "grad_norm": 2.392444610595703, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7226022481918335, + "num_tokens": 282487813.0, + "step": 11081 + }, + { + "epoch": 1.2169997803645947, + "grad_norm": 2.552394390106201, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7429132461547852, + "num_tokens": 282508045.0, + "step": 11082 + }, + { + "epoch": 1.2171095980672084, + "grad_norm": 2.4368011951446533, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7100664377212524, + "num_tokens": 282532458.0, + "step": 11083 + }, + { + "epoch": 1.2172194157698222, + "grad_norm": 2.2308130264282227, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.710623025894165, + "num_tokens": 282559064.0, + "step": 11084 + }, + { + "epoch": 1.2173292334724357, + "grad_norm": 2.322854995727539, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7097182869911194, + "num_tokens": 282584275.0, + "step": 11085 + }, + { + "epoch": 1.2174390511750495, + "grad_norm": 2.4006505012512207, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7169598937034607, + "num_tokens": 282607213.0, + "step": 11086 + }, + { + "epoch": 1.217548868877663, + "grad_norm": 2.417379379272461, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7231712341308594, + "num_tokens": 282629983.0, + "step": 11087 + }, + { + "epoch": 1.2176586865802768, + "grad_norm": 2.3066213130950928, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7390333414077759, + "num_tokens": 282655272.0, + "step": 11088 + }, + { + "epoch": 1.2177685042828905, + "grad_norm": 2.496269702911377, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7349480390548706, + "num_tokens": 282675594.0, + "step": 11089 + }, + { + "epoch": 1.217878321985504, + "grad_norm": 2.4042227268218994, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7338712215423584, + "num_tokens": 282699802.0, + "step": 11090 + }, + { + "epoch": 1.2179881396881178, + "grad_norm": 2.4957542419433594, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7216694355010986, + "num_tokens": 282720403.0, + "step": 11091 + }, + { + "epoch": 1.2180979573907313, + "grad_norm": 2.370408296585083, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.733116626739502, + "num_tokens": 282742604.0, + "step": 11092 + }, + { + "epoch": 1.218207775093345, + "grad_norm": 2.215101480484009, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7116120457649231, + "num_tokens": 282769031.0, + "step": 11093 + }, + { + "epoch": 1.2183175927959586, + "grad_norm": 1.8513604402542114, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7030009031295776, + "num_tokens": 282803421.0, + "step": 11094 + }, + { + "epoch": 1.2184274104985724, + "grad_norm": 2.5753185749053955, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7282220721244812, + "num_tokens": 282823645.0, + "step": 11095 + }, + { + "epoch": 1.218537228201186, + "grad_norm": 2.4379987716674805, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7090802192687988, + "num_tokens": 282847014.0, + "step": 11096 + }, + { + "epoch": 1.2186470459037997, + "grad_norm": 2.401743173599243, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7078419923782349, + "num_tokens": 282871048.0, + "step": 11097 + }, + { + "epoch": 1.2187568636064134, + "grad_norm": 2.4329798221588135, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7134207487106323, + "num_tokens": 282896615.0, + "step": 11098 + }, + { + "epoch": 1.218866681309027, + "grad_norm": 2.5111522674560547, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7334416508674622, + "num_tokens": 282916937.0, + "step": 11099 + }, + { + "epoch": 1.2189764990116407, + "grad_norm": 2.5292093753814697, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7331675291061401, + "num_tokens": 282939119.0, + "step": 11100 + }, + { + "epoch": 1.2190863167142543, + "grad_norm": 2.4440934658050537, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7107405662536621, + "num_tokens": 282962421.0, + "step": 11101 + }, + { + "epoch": 1.219196134416868, + "grad_norm": 2.291870355606079, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7274824380874634, + "num_tokens": 282988314.0, + "step": 11102 + }, + { + "epoch": 1.2193059521194818, + "grad_norm": 2.3451356887817383, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.699771523475647, + "num_tokens": 283014455.0, + "step": 11103 + }, + { + "epoch": 1.2194157698220953, + "grad_norm": 2.1944005489349365, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.71929532289505, + "num_tokens": 283040205.0, + "step": 11104 + }, + { + "epoch": 1.219525587524709, + "grad_norm": 2.2345385551452637, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.6929867267608643, + "num_tokens": 283064433.0, + "step": 11105 + }, + { + "epoch": 1.2196354052273226, + "grad_norm": 2.433598041534424, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7071266174316406, + "num_tokens": 283087197.0, + "step": 11106 + }, + { + "epoch": 1.2197452229299364, + "grad_norm": 2.0315730571746826, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7183578014373779, + "num_tokens": 283116280.0, + "step": 11107 + }, + { + "epoch": 1.21985504063255, + "grad_norm": 2.438321113586426, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7122551202774048, + "num_tokens": 283140434.0, + "step": 11108 + }, + { + "epoch": 1.2199648583351637, + "grad_norm": 2.1332194805145264, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7092052698135376, + "num_tokens": 283170897.0, + "step": 11109 + }, + { + "epoch": 1.2200746760377772, + "grad_norm": 2.6120972633361816, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7229220867156982, + "num_tokens": 283190901.0, + "step": 11110 + }, + { + "epoch": 1.220184493740391, + "grad_norm": 2.4266483783721924, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7256805300712585, + "num_tokens": 283213129.0, + "step": 11111 + }, + { + "epoch": 1.2202943114430047, + "grad_norm": 2.5925862789154053, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7012203931808472, + "num_tokens": 283235195.0, + "step": 11112 + }, + { + "epoch": 1.2204041291456182, + "grad_norm": 2.2620086669921875, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.734660267829895, + "num_tokens": 283260147.0, + "step": 11113 + }, + { + "epoch": 1.220513946848232, + "grad_norm": 2.2122154235839844, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7048000693321228, + "num_tokens": 283288644.0, + "step": 11114 + }, + { + "epoch": 1.2206237645508455, + "grad_norm": 2.1775643825531006, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7162802815437317, + "num_tokens": 283315662.0, + "step": 11115 + }, + { + "epoch": 1.2207335822534593, + "grad_norm": 2.5681755542755127, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7199462652206421, + "num_tokens": 283335325.0, + "step": 11116 + }, + { + "epoch": 1.2208433999560728, + "grad_norm": 2.2368924617767334, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7278670072555542, + "num_tokens": 283361413.0, + "step": 11117 + }, + { + "epoch": 1.2209532176586866, + "grad_norm": 2.5601298809051514, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.715337872505188, + "num_tokens": 283383596.0, + "step": 11118 + }, + { + "epoch": 1.2210630353613001, + "grad_norm": 2.373093843460083, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.733481228351593, + "num_tokens": 283406801.0, + "step": 11119 + }, + { + "epoch": 1.2211728530639139, + "grad_norm": 2.4561357498168945, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7081364393234253, + "num_tokens": 283429609.0, + "step": 11120 + }, + { + "epoch": 1.2212826707665276, + "grad_norm": 2.057649612426758, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7216874361038208, + "num_tokens": 283460242.0, + "step": 11121 + }, + { + "epoch": 1.2213924884691412, + "grad_norm": 2.1803479194641113, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7024332284927368, + "num_tokens": 283487421.0, + "step": 11122 + }, + { + "epoch": 1.221502306171755, + "grad_norm": 2.2949423789978027, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7263647317886353, + "num_tokens": 283512926.0, + "step": 11123 + }, + { + "epoch": 1.2216121238743685, + "grad_norm": 2.2620413303375244, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7120955586433411, + "num_tokens": 283539111.0, + "step": 11124 + }, + { + "epoch": 1.2217219415769822, + "grad_norm": 2.3622727394104004, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.6991720199584961, + "num_tokens": 283563357.0, + "step": 11125 + }, + { + "epoch": 1.221831759279596, + "grad_norm": 2.1033637523651123, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.7479457855224609, + "num_tokens": 283591550.0, + "step": 11126 + }, + { + "epoch": 1.2219415769822095, + "grad_norm": 2.197075843811035, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.697762668132782, + "num_tokens": 283620160.0, + "step": 11127 + }, + { + "epoch": 1.2220513946848233, + "grad_norm": 2.3487203121185303, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7149367332458496, + "num_tokens": 283644800.0, + "step": 11128 + }, + { + "epoch": 1.2221612123874368, + "grad_norm": 2.1393024921417236, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7091655135154724, + "num_tokens": 283674004.0, + "step": 11129 + }, + { + "epoch": 1.2222710300900506, + "grad_norm": 2.598217725753784, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7140762805938721, + "num_tokens": 283695529.0, + "step": 11130 + }, + { + "epoch": 1.222380847792664, + "grad_norm": 2.495147705078125, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7264555096626282, + "num_tokens": 283717380.0, + "step": 11131 + }, + { + "epoch": 1.2224906654952779, + "grad_norm": 2.2863786220550537, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6897382140159607, + "num_tokens": 283744250.0, + "step": 11132 + }, + { + "epoch": 1.2226004831978914, + "grad_norm": 2.2221081256866455, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7351019978523254, + "num_tokens": 283769336.0, + "step": 11133 + }, + { + "epoch": 1.2227103009005051, + "grad_norm": 2.383549690246582, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7289566993713379, + "num_tokens": 283792267.0, + "step": 11134 + }, + { + "epoch": 1.222820118603119, + "grad_norm": 2.369814395904541, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7295185327529907, + "num_tokens": 283815758.0, + "step": 11135 + }, + { + "epoch": 1.2229299363057324, + "grad_norm": 2.161276340484619, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7024117708206177, + "num_tokens": 283844624.0, + "step": 11136 + }, + { + "epoch": 1.2230397540083462, + "grad_norm": 1.871704339981079, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7202917337417603, + "num_tokens": 283877970.0, + "step": 11137 + }, + { + "epoch": 1.2231495717109597, + "grad_norm": 1.901994228363037, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6883041858673096, + "num_tokens": 283912592.0, + "step": 11138 + }, + { + "epoch": 1.2232593894135735, + "grad_norm": 2.3408846855163574, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7025007009506226, + "num_tokens": 283935970.0, + "step": 11139 + }, + { + "epoch": 1.2233692071161872, + "grad_norm": 2.208749532699585, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.6990125179290771, + "num_tokens": 283962296.0, + "step": 11140 + }, + { + "epoch": 1.2234790248188008, + "grad_norm": 2.3195340633392334, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7241528630256653, + "num_tokens": 283985155.0, + "step": 11141 + }, + { + "epoch": 1.2235888425214145, + "grad_norm": 2.1408820152282715, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7041275501251221, + "num_tokens": 284012946.0, + "step": 11142 + }, + { + "epoch": 1.223698660224028, + "grad_norm": 2.3938581943511963, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7144103050231934, + "num_tokens": 284035728.0, + "step": 11143 + }, + { + "epoch": 1.2238084779266418, + "grad_norm": 2.249861717224121, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7298753261566162, + "num_tokens": 284061448.0, + "step": 11144 + }, + { + "epoch": 1.2239182956292554, + "grad_norm": 2.3143553733825684, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7182933688163757, + "num_tokens": 284087129.0, + "step": 11145 + }, + { + "epoch": 1.2240281133318691, + "grad_norm": 2.538013458251953, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7249448299407959, + "num_tokens": 284110256.0, + "step": 11146 + }, + { + "epoch": 1.2241379310344827, + "grad_norm": 2.027890920639038, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7379600405693054, + "num_tokens": 284137254.0, + "step": 11147 + }, + { + "epoch": 1.2242477487370964, + "grad_norm": 2.2991840839385986, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.745224118232727, + "num_tokens": 284161014.0, + "step": 11148 + }, + { + "epoch": 1.2243575664397102, + "grad_norm": 2.417454957962036, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7175333499908447, + "num_tokens": 284185852.0, + "step": 11149 + }, + { + "epoch": 1.2244673841423237, + "grad_norm": 2.216092109680176, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7199913263320923, + "num_tokens": 284211633.0, + "step": 11150 + }, + { + "epoch": 1.2245772018449375, + "grad_norm": 2.2685699462890625, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7079087495803833, + "num_tokens": 284237889.0, + "step": 11151 + }, + { + "epoch": 1.224687019547551, + "grad_norm": 2.0055360794067383, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.719650149345398, + "num_tokens": 284269081.0, + "step": 11152 + }, + { + "epoch": 1.2247968372501647, + "grad_norm": 2.299729108810425, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7266805171966553, + "num_tokens": 284293008.0, + "step": 11153 + }, + { + "epoch": 1.2249066549527785, + "grad_norm": 2.1617581844329834, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7234988212585449, + "num_tokens": 284319390.0, + "step": 11154 + }, + { + "epoch": 1.225016472655392, + "grad_norm": 2.080368757247925, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7168358564376831, + "num_tokens": 284348428.0, + "step": 11155 + }, + { + "epoch": 1.2251262903580058, + "grad_norm": 2.3124256134033203, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.7081645727157593, + "num_tokens": 284371667.0, + "step": 11156 + }, + { + "epoch": 1.2252361080606193, + "grad_norm": 2.2889060974121094, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6971611380577087, + "num_tokens": 284397069.0, + "step": 11157 + }, + { + "epoch": 1.225345925763233, + "grad_norm": 2.2345409393310547, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7197890281677246, + "num_tokens": 284423898.0, + "step": 11158 + }, + { + "epoch": 1.2254557434658466, + "grad_norm": 2.1895155906677246, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7103161215782166, + "num_tokens": 284449378.0, + "step": 11159 + }, + { + "epoch": 1.2255655611684604, + "grad_norm": 2.340538263320923, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7350970506668091, + "num_tokens": 284472642.0, + "step": 11160 + }, + { + "epoch": 1.225675378871074, + "grad_norm": 2.297412157058716, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7356545925140381, + "num_tokens": 284495268.0, + "step": 11161 + }, + { + "epoch": 1.2257851965736877, + "grad_norm": 2.207113027572632, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7171177268028259, + "num_tokens": 284520502.0, + "step": 11162 + }, + { + "epoch": 1.2258950142763014, + "grad_norm": 2.2768421173095703, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7337387800216675, + "num_tokens": 284545104.0, + "step": 11163 + }, + { + "epoch": 1.226004831978915, + "grad_norm": 2.490093946456909, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7371501922607422, + "num_tokens": 284565739.0, + "step": 11164 + }, + { + "epoch": 1.2261146496815287, + "grad_norm": 2.53021502494812, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7127695083618164, + "num_tokens": 284586541.0, + "step": 11165 + }, + { + "epoch": 1.2262244673841423, + "grad_norm": 2.1369473934173584, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7165851593017578, + "num_tokens": 284616241.0, + "step": 11166 + }, + { + "epoch": 1.226334285086756, + "grad_norm": 2.6357898712158203, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7252757549285889, + "num_tokens": 284638269.0, + "step": 11167 + }, + { + "epoch": 1.2264441027893696, + "grad_norm": 2.0817599296569824, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7210673689842224, + "num_tokens": 284667137.0, + "step": 11168 + }, + { + "epoch": 1.2265539204919833, + "grad_norm": 2.024289131164551, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7043188810348511, + "num_tokens": 284696226.0, + "step": 11169 + }, + { + "epoch": 1.226663738194597, + "grad_norm": 2.231025457382202, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7092667818069458, + "num_tokens": 284722383.0, + "step": 11170 + }, + { + "epoch": 1.2267735558972106, + "grad_norm": 2.3684465885162354, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7238105535507202, + "num_tokens": 284746783.0, + "step": 11171 + }, + { + "epoch": 1.2268833735998244, + "grad_norm": 2.231675148010254, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7198964357376099, + "num_tokens": 284774010.0, + "step": 11172 + }, + { + "epoch": 1.226993191302438, + "grad_norm": 2.1051268577575684, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6949008107185364, + "num_tokens": 284804103.0, + "step": 11173 + }, + { + "epoch": 1.2271030090050516, + "grad_norm": 2.2843375205993652, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7174983024597168, + "num_tokens": 284830405.0, + "step": 11174 + }, + { + "epoch": 1.2272128267076652, + "grad_norm": 2.4065208435058594, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7144500017166138, + "num_tokens": 284855686.0, + "step": 11175 + }, + { + "epoch": 1.227322644410279, + "grad_norm": 1.98502516746521, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7269219160079956, + "num_tokens": 284885924.0, + "step": 11176 + }, + { + "epoch": 1.2274324621128927, + "grad_norm": 2.2038795948028564, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7058092355728149, + "num_tokens": 284913202.0, + "step": 11177 + }, + { + "epoch": 1.2275422798155062, + "grad_norm": 2.2369844913482666, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7052249908447266, + "num_tokens": 284942438.0, + "step": 11178 + }, + { + "epoch": 1.22765209751812, + "grad_norm": 2.1244561672210693, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.697675347328186, + "num_tokens": 284973058.0, + "step": 11179 + }, + { + "epoch": 1.2277619152207335, + "grad_norm": 2.1837873458862305, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7317144870758057, + "num_tokens": 284998595.0, + "step": 11180 + }, + { + "epoch": 1.2278717329233473, + "grad_norm": 2.0799005031585693, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7003298997879028, + "num_tokens": 285028334.0, + "step": 11181 + }, + { + "epoch": 1.2279815506259608, + "grad_norm": 2.4387967586517334, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.718009889125824, + "num_tokens": 285051269.0, + "step": 11182 + }, + { + "epoch": 1.2280913683285746, + "grad_norm": 2.3295581340789795, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7154151201248169, + "num_tokens": 285076062.0, + "step": 11183 + }, + { + "epoch": 1.228201186031188, + "grad_norm": 2.341142177581787, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7061322331428528, + "num_tokens": 285100842.0, + "step": 11184 + }, + { + "epoch": 1.2283110037338019, + "grad_norm": 2.6390883922576904, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7228176593780518, + "num_tokens": 285119817.0, + "step": 11185 + }, + { + "epoch": 1.2284208214364156, + "grad_norm": 2.064368724822998, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7205016016960144, + "num_tokens": 285149854.0, + "step": 11186 + }, + { + "epoch": 1.2285306391390292, + "grad_norm": 2.3219645023345947, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7130652666091919, + "num_tokens": 285173962.0, + "step": 11187 + }, + { + "epoch": 1.228640456841643, + "grad_norm": 2.616460084915161, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7101187109947205, + "num_tokens": 285196010.0, + "step": 11188 + }, + { + "epoch": 1.2287502745442564, + "grad_norm": 2.4488987922668457, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7296178340911865, + "num_tokens": 285217567.0, + "step": 11189 + }, + { + "epoch": 1.2288600922468702, + "grad_norm": 2.218730926513672, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7129876017570496, + "num_tokens": 285243837.0, + "step": 11190 + }, + { + "epoch": 1.228969909949484, + "grad_norm": 2.3747806549072266, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7176129817962646, + "num_tokens": 285268757.0, + "step": 11191 + }, + { + "epoch": 1.2290797276520975, + "grad_norm": 2.28375244140625, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7179316878318787, + "num_tokens": 285294209.0, + "step": 11192 + }, + { + "epoch": 1.2291895453547113, + "grad_norm": 2.216475009918213, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.6984639167785645, + "num_tokens": 285319447.0, + "step": 11193 + }, + { + "epoch": 1.2292993630573248, + "grad_norm": 2.16048264503479, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7334835529327393, + "num_tokens": 285345079.0, + "step": 11194 + }, + { + "epoch": 1.2294091807599385, + "grad_norm": 2.055168390274048, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7160689830780029, + "num_tokens": 285371737.0, + "step": 11195 + }, + { + "epoch": 1.229518998462552, + "grad_norm": 2.274339437484741, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7055029273033142, + "num_tokens": 285397418.0, + "step": 11196 + }, + { + "epoch": 1.2296288161651658, + "grad_norm": 2.370549201965332, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7404452562332153, + "num_tokens": 285420444.0, + "step": 11197 + }, + { + "epoch": 1.2297386338677794, + "grad_norm": 2.2200019359588623, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7110909223556519, + "num_tokens": 285446393.0, + "step": 11198 + }, + { + "epoch": 1.2298484515703931, + "grad_norm": 2.605926036834717, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7171629667282104, + "num_tokens": 285467223.0, + "step": 11199 + }, + { + "epoch": 1.2299582692730069, + "grad_norm": 2.31674861907959, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7151936292648315, + "num_tokens": 285491224.0, + "step": 11200 + }, + { + "epoch": 1.2300680869756204, + "grad_norm": 2.1554269790649414, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7168041467666626, + "num_tokens": 285520177.0, + "step": 11201 + }, + { + "epoch": 1.2301779046782342, + "grad_norm": 2.2855706214904785, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7099796533584595, + "num_tokens": 285544398.0, + "step": 11202 + }, + { + "epoch": 1.2302877223808477, + "grad_norm": 2.2359697818756104, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7204725742340088, + "num_tokens": 285569750.0, + "step": 11203 + }, + { + "epoch": 1.2303975400834615, + "grad_norm": 2.3155367374420166, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7164356112480164, + "num_tokens": 285594274.0, + "step": 11204 + }, + { + "epoch": 1.2305073577860752, + "grad_norm": 1.992428183555603, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.6988492012023926, + "num_tokens": 285629174.0, + "step": 11205 + }, + { + "epoch": 1.2306171754886888, + "grad_norm": 2.264345645904541, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.731654167175293, + "num_tokens": 285653936.0, + "step": 11206 + }, + { + "epoch": 1.2307269931913025, + "grad_norm": 2.3893449306488037, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7004422545433044, + "num_tokens": 285677998.0, + "step": 11207 + }, + { + "epoch": 1.230836810893916, + "grad_norm": 2.0227248668670654, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7070268392562866, + "num_tokens": 285708499.0, + "step": 11208 + }, + { + "epoch": 1.2309466285965298, + "grad_norm": 2.1011626720428467, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7371824979782104, + "num_tokens": 285734763.0, + "step": 11209 + }, + { + "epoch": 1.2310564462991433, + "grad_norm": 2.263463020324707, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7073255777359009, + "num_tokens": 285761285.0, + "step": 11210 + }, + { + "epoch": 1.231166264001757, + "grad_norm": 2.655268430709839, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7152304649353027, + "num_tokens": 285780262.0, + "step": 11211 + }, + { + "epoch": 1.2312760817043706, + "grad_norm": 2.298910140991211, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7401976585388184, + "num_tokens": 285803961.0, + "step": 11212 + }, + { + "epoch": 1.2313858994069844, + "grad_norm": 2.3431169986724854, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7303987741470337, + "num_tokens": 285828187.0, + "step": 11213 + }, + { + "epoch": 1.2314957171095982, + "grad_norm": 2.140848159790039, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7099233865737915, + "num_tokens": 285854478.0, + "step": 11214 + }, + { + "epoch": 1.2316055348122117, + "grad_norm": 2.1903319358825684, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.6978862285614014, + "num_tokens": 285881149.0, + "step": 11215 + }, + { + "epoch": 1.2317153525148254, + "grad_norm": 2.381929397583008, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7076553106307983, + "num_tokens": 285904765.0, + "step": 11216 + }, + { + "epoch": 1.231825170217439, + "grad_norm": 2.0125977993011475, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7206387519836426, + "num_tokens": 285936805.0, + "step": 11217 + }, + { + "epoch": 1.2319349879200527, + "grad_norm": 2.3221242427825928, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7278957962989807, + "num_tokens": 285959405.0, + "step": 11218 + }, + { + "epoch": 1.2320448056226665, + "grad_norm": 2.515906810760498, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7273783683776855, + "num_tokens": 285980702.0, + "step": 11219 + }, + { + "epoch": 1.23215462332528, + "grad_norm": 2.050539970397949, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7301499843597412, + "num_tokens": 286009390.0, + "step": 11220 + }, + { + "epoch": 1.2322644410278938, + "grad_norm": 2.455965518951416, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7287198305130005, + "num_tokens": 286031139.0, + "step": 11221 + }, + { + "epoch": 1.2323742587305073, + "grad_norm": 2.4161112308502197, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.740276038646698, + "num_tokens": 286054945.0, + "step": 11222 + }, + { + "epoch": 1.232484076433121, + "grad_norm": 2.7427196502685547, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7349627614021301, + "num_tokens": 286073223.0, + "step": 11223 + }, + { + "epoch": 1.2325938941357346, + "grad_norm": 2.3246893882751465, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7085442543029785, + "num_tokens": 286098369.0, + "step": 11224 + }, + { + "epoch": 1.2327037118383484, + "grad_norm": 2.062190055847168, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7186402082443237, + "num_tokens": 286127849.0, + "step": 11225 + }, + { + "epoch": 1.232813529540962, + "grad_norm": 2.1394004821777344, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7132437825202942, + "num_tokens": 286158552.0, + "step": 11226 + }, + { + "epoch": 1.2329233472435757, + "grad_norm": 2.3040311336517334, + "learning_rate": 1e-06, + "loss": 1.0295, + "mean_token_accuracy": 0.6927962303161621, + "num_tokens": 286184755.0, + "step": 11227 + }, + { + "epoch": 1.2330331649461894, + "grad_norm": 2.2145700454711914, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7192983627319336, + "num_tokens": 286210040.0, + "step": 11228 + }, + { + "epoch": 1.233142982648803, + "grad_norm": 2.2451581954956055, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7125279903411865, + "num_tokens": 286236198.0, + "step": 11229 + }, + { + "epoch": 1.2332528003514167, + "grad_norm": 2.6866683959960938, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7375297546386719, + "num_tokens": 286255297.0, + "step": 11230 + }, + { + "epoch": 1.2333626180540302, + "grad_norm": 2.535093069076538, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7040011882781982, + "num_tokens": 286276543.0, + "step": 11231 + }, + { + "epoch": 1.233472435756644, + "grad_norm": 2.25215744972229, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7176752090454102, + "num_tokens": 286301937.0, + "step": 11232 + }, + { + "epoch": 1.2335822534592575, + "grad_norm": 2.2062644958496094, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7111848592758179, + "num_tokens": 286327156.0, + "step": 11233 + }, + { + "epoch": 1.2336920711618713, + "grad_norm": 2.2974753379821777, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.733816385269165, + "num_tokens": 286353079.0, + "step": 11234 + }, + { + "epoch": 1.2338018888644848, + "grad_norm": 2.185241222381592, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7258789539337158, + "num_tokens": 286379272.0, + "step": 11235 + }, + { + "epoch": 1.2339117065670986, + "grad_norm": 2.5161125659942627, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.696373462677002, + "num_tokens": 286402935.0, + "step": 11236 + }, + { + "epoch": 1.2340215242697123, + "grad_norm": 2.421290636062622, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7025997638702393, + "num_tokens": 286427597.0, + "step": 11237 + }, + { + "epoch": 1.2341313419723259, + "grad_norm": 2.1042728424072266, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7120476365089417, + "num_tokens": 286457683.0, + "step": 11238 + }, + { + "epoch": 1.2342411596749396, + "grad_norm": 2.404784679412842, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7156987190246582, + "num_tokens": 286480933.0, + "step": 11239 + }, + { + "epoch": 1.2343509773775532, + "grad_norm": 2.716447353363037, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7278131246566772, + "num_tokens": 286498501.0, + "step": 11240 + }, + { + "epoch": 1.234460795080167, + "grad_norm": 2.406221628189087, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7069442272186279, + "num_tokens": 286523693.0, + "step": 11241 + }, + { + "epoch": 1.2345706127827807, + "grad_norm": 2.0773661136627197, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.724715530872345, + "num_tokens": 286551392.0, + "step": 11242 + }, + { + "epoch": 1.2346804304853942, + "grad_norm": 2.183776617050171, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7121713757514954, + "num_tokens": 286579113.0, + "step": 11243 + }, + { + "epoch": 1.234790248188008, + "grad_norm": 2.3690502643585205, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.6974643468856812, + "num_tokens": 286602644.0, + "step": 11244 + }, + { + "epoch": 1.2349000658906215, + "grad_norm": 2.3065803050994873, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7284796833992004, + "num_tokens": 286625424.0, + "step": 11245 + }, + { + "epoch": 1.2350098835932353, + "grad_norm": 2.297661542892456, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7355036735534668, + "num_tokens": 286649938.0, + "step": 11246 + }, + { + "epoch": 1.2351197012958488, + "grad_norm": 2.045999765396118, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7084248661994934, + "num_tokens": 286681020.0, + "step": 11247 + }, + { + "epoch": 1.2352295189984626, + "grad_norm": 2.1123433113098145, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7049106359481812, + "num_tokens": 286708052.0, + "step": 11248 + }, + { + "epoch": 1.235339336701076, + "grad_norm": 2.296116590499878, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7143243551254272, + "num_tokens": 286731879.0, + "step": 11249 + }, + { + "epoch": 1.2354491544036899, + "grad_norm": 2.2243824005126953, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.729107141494751, + "num_tokens": 286757989.0, + "step": 11250 + }, + { + "epoch": 1.2355589721063036, + "grad_norm": 2.057258129119873, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7296086549758911, + "num_tokens": 286789293.0, + "step": 11251 + }, + { + "epoch": 1.2356687898089171, + "grad_norm": 2.43467378616333, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6981437802314758, + "num_tokens": 286812733.0, + "step": 11252 + }, + { + "epoch": 1.235778607511531, + "grad_norm": 2.05610990524292, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7067534923553467, + "num_tokens": 286842720.0, + "step": 11253 + }, + { + "epoch": 1.2358884252141444, + "grad_norm": 2.2806386947631836, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7018907070159912, + "num_tokens": 286868264.0, + "step": 11254 + }, + { + "epoch": 1.2359982429167582, + "grad_norm": 2.441148519515991, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7088432312011719, + "num_tokens": 286893072.0, + "step": 11255 + }, + { + "epoch": 1.236108060619372, + "grad_norm": 2.700122117996216, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.728951096534729, + "num_tokens": 286911037.0, + "step": 11256 + }, + { + "epoch": 1.2362178783219855, + "grad_norm": 2.26953125, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7105988264083862, + "num_tokens": 286935619.0, + "step": 11257 + }, + { + "epoch": 1.2363276960245992, + "grad_norm": 2.135990619659424, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7105220556259155, + "num_tokens": 286964254.0, + "step": 11258 + }, + { + "epoch": 1.2364375137272128, + "grad_norm": 2.261697292327881, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7148607969284058, + "num_tokens": 286989952.0, + "step": 11259 + }, + { + "epoch": 1.2365473314298265, + "grad_norm": 2.1773340702056885, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7098256349563599, + "num_tokens": 287015965.0, + "step": 11260 + }, + { + "epoch": 1.23665714913244, + "grad_norm": 2.240853786468506, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7043310403823853, + "num_tokens": 287042562.0, + "step": 11261 + }, + { + "epoch": 1.2367669668350538, + "grad_norm": 2.1137096881866455, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7157418727874756, + "num_tokens": 287072019.0, + "step": 11262 + }, + { + "epoch": 1.2368767845376674, + "grad_norm": 2.2288010120391846, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7088686227798462, + "num_tokens": 287100763.0, + "step": 11263 + }, + { + "epoch": 1.2369866022402811, + "grad_norm": 2.3116230964660645, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.732528805732727, + "num_tokens": 287124901.0, + "step": 11264 + }, + { + "epoch": 1.2370964199428949, + "grad_norm": 2.1455259323120117, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.694935142993927, + "num_tokens": 287152390.0, + "step": 11265 + }, + { + "epoch": 1.2372062376455084, + "grad_norm": 2.3735718727111816, + "learning_rate": 1e-06, + "loss": 0.7715, + "mean_token_accuracy": 0.7603986859321594, + "num_tokens": 287175019.0, + "step": 11266 + }, + { + "epoch": 1.2373160553481222, + "grad_norm": 2.390923261642456, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7144507169723511, + "num_tokens": 287197814.0, + "step": 11267 + }, + { + "epoch": 1.2374258730507357, + "grad_norm": 2.330934762954712, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7229169607162476, + "num_tokens": 287221077.0, + "step": 11268 + }, + { + "epoch": 1.2375356907533495, + "grad_norm": 2.2810704708099365, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7217615842819214, + "num_tokens": 287246947.0, + "step": 11269 + }, + { + "epoch": 1.2376455084559632, + "grad_norm": 2.247426986694336, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7192519903182983, + "num_tokens": 287271508.0, + "step": 11270 + }, + { + "epoch": 1.2377553261585768, + "grad_norm": 2.192904233932495, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7091515064239502, + "num_tokens": 287298419.0, + "step": 11271 + }, + { + "epoch": 1.2378651438611905, + "grad_norm": 2.442232847213745, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7158471345901489, + "num_tokens": 287321621.0, + "step": 11272 + }, + { + "epoch": 1.237974961563804, + "grad_norm": 2.2918307781219482, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7218230366706848, + "num_tokens": 287346709.0, + "step": 11273 + }, + { + "epoch": 1.2380847792664178, + "grad_norm": 2.4567742347717285, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7263038158416748, + "num_tokens": 287368418.0, + "step": 11274 + }, + { + "epoch": 1.2381945969690313, + "grad_norm": 2.376633644104004, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.729267954826355, + "num_tokens": 287390618.0, + "step": 11275 + }, + { + "epoch": 1.238304414671645, + "grad_norm": 2.1311869621276855, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.712165355682373, + "num_tokens": 287417899.0, + "step": 11276 + }, + { + "epoch": 1.2384142323742586, + "grad_norm": 2.220477819442749, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7177362442016602, + "num_tokens": 287443962.0, + "step": 11277 + }, + { + "epoch": 1.2385240500768724, + "grad_norm": 2.6209850311279297, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.705018937587738, + "num_tokens": 287464021.0, + "step": 11278 + }, + { + "epoch": 1.2386338677794861, + "grad_norm": 2.636770009994507, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7437052130699158, + "num_tokens": 287483604.0, + "step": 11279 + }, + { + "epoch": 1.2387436854820997, + "grad_norm": 2.500178813934326, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7207143306732178, + "num_tokens": 287504959.0, + "step": 11280 + }, + { + "epoch": 1.2388535031847134, + "grad_norm": 2.032400131225586, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.714206337928772, + "num_tokens": 287536455.0, + "step": 11281 + }, + { + "epoch": 1.238963320887327, + "grad_norm": 2.051352024078369, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7202959060668945, + "num_tokens": 287563998.0, + "step": 11282 + }, + { + "epoch": 1.2390731385899407, + "grad_norm": 2.4858975410461426, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7103620767593384, + "num_tokens": 287585085.0, + "step": 11283 + }, + { + "epoch": 1.2391829562925545, + "grad_norm": 2.213334560394287, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7239176034927368, + "num_tokens": 287610535.0, + "step": 11284 + }, + { + "epoch": 1.239292773995168, + "grad_norm": 2.376112461090088, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7075142860412598, + "num_tokens": 287636166.0, + "step": 11285 + }, + { + "epoch": 1.2394025916977818, + "grad_norm": 2.1419878005981445, + "learning_rate": 1e-06, + "loss": 1.039, + "mean_token_accuracy": 0.6787771582603455, + "num_tokens": 287666833.0, + "step": 11286 + }, + { + "epoch": 1.2395124094003953, + "grad_norm": 2.2870607376098633, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7070609331130981, + "num_tokens": 287692066.0, + "step": 11287 + }, + { + "epoch": 1.239622227103009, + "grad_norm": 2.4572389125823975, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7301188707351685, + "num_tokens": 287713427.0, + "step": 11288 + }, + { + "epoch": 1.2397320448056226, + "grad_norm": 2.1833648681640625, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.713598370552063, + "num_tokens": 287740454.0, + "step": 11289 + }, + { + "epoch": 1.2398418625082364, + "grad_norm": 2.3121941089630127, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7250069975852966, + "num_tokens": 287769249.0, + "step": 11290 + }, + { + "epoch": 1.23995168021085, + "grad_norm": 2.020024538040161, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7341405153274536, + "num_tokens": 287796856.0, + "step": 11291 + }, + { + "epoch": 1.2400614979134637, + "grad_norm": 2.407137155532837, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7274477481842041, + "num_tokens": 287820207.0, + "step": 11292 + }, + { + "epoch": 1.2401713156160774, + "grad_norm": 2.3013198375701904, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7279641628265381, + "num_tokens": 287845278.0, + "step": 11293 + }, + { + "epoch": 1.240281133318691, + "grad_norm": 2.4097917079925537, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7047096490859985, + "num_tokens": 287871132.0, + "step": 11294 + }, + { + "epoch": 1.2403909510213047, + "grad_norm": 2.197113275527954, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6932398080825806, + "num_tokens": 287899614.0, + "step": 11295 + }, + { + "epoch": 1.2405007687239182, + "grad_norm": 2.441572904586792, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7291474342346191, + "num_tokens": 287922919.0, + "step": 11296 + }, + { + "epoch": 1.240610586426532, + "grad_norm": 2.366194725036621, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7483623027801514, + "num_tokens": 287947865.0, + "step": 11297 + }, + { + "epoch": 1.2407204041291455, + "grad_norm": 2.0162734985351562, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7059214115142822, + "num_tokens": 287979292.0, + "step": 11298 + }, + { + "epoch": 1.2408302218317593, + "grad_norm": 2.259227752685547, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7164934873580933, + "num_tokens": 288004101.0, + "step": 11299 + }, + { + "epoch": 1.2409400395343728, + "grad_norm": 2.2411155700683594, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7275286316871643, + "num_tokens": 288030413.0, + "step": 11300 + }, + { + "epoch": 1.2410498572369866, + "grad_norm": 2.6348910331726074, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7211357355117798, + "num_tokens": 288050897.0, + "step": 11301 + }, + { + "epoch": 1.2411596749396003, + "grad_norm": 2.3381998538970947, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7278153896331787, + "num_tokens": 288073371.0, + "step": 11302 + }, + { + "epoch": 1.2412694926422139, + "grad_norm": 2.3879876136779785, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.705898106098175, + "num_tokens": 288100508.0, + "step": 11303 + }, + { + "epoch": 1.2413793103448276, + "grad_norm": 2.1970229148864746, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.6929582357406616, + "num_tokens": 288128748.0, + "step": 11304 + }, + { + "epoch": 1.2414891280474412, + "grad_norm": 2.3665902614593506, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.716646671295166, + "num_tokens": 288151838.0, + "step": 11305 + }, + { + "epoch": 1.241598945750055, + "grad_norm": 2.4009175300598145, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7378892302513123, + "num_tokens": 288173744.0, + "step": 11306 + }, + { + "epoch": 1.2417087634526687, + "grad_norm": 2.4444048404693604, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7164067625999451, + "num_tokens": 288195438.0, + "step": 11307 + }, + { + "epoch": 1.2418185811552822, + "grad_norm": 2.3453242778778076, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7330424785614014, + "num_tokens": 288219113.0, + "step": 11308 + }, + { + "epoch": 1.241928398857896, + "grad_norm": 2.1479086875915527, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7212426662445068, + "num_tokens": 288245692.0, + "step": 11309 + }, + { + "epoch": 1.2420382165605095, + "grad_norm": 2.1755099296569824, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7219060659408569, + "num_tokens": 288271585.0, + "step": 11310 + }, + { + "epoch": 1.2421480342631233, + "grad_norm": 2.476268768310547, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.7043755054473877, + "num_tokens": 288294093.0, + "step": 11311 + }, + { + "epoch": 1.2422578519657368, + "grad_norm": 2.3677332401275635, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7194643616676331, + "num_tokens": 288319012.0, + "step": 11312 + }, + { + "epoch": 1.2423676696683505, + "grad_norm": 2.2040483951568604, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7333937287330627, + "num_tokens": 288347324.0, + "step": 11313 + }, + { + "epoch": 1.242477487370964, + "grad_norm": 1.9663876295089722, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7100803852081299, + "num_tokens": 288376152.0, + "step": 11314 + }, + { + "epoch": 1.2425873050735778, + "grad_norm": 2.3750579357147217, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7282865047454834, + "num_tokens": 288398912.0, + "step": 11315 + }, + { + "epoch": 1.2426971227761916, + "grad_norm": 2.1498847007751465, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7197004556655884, + "num_tokens": 288426328.0, + "step": 11316 + }, + { + "epoch": 1.2428069404788051, + "grad_norm": 2.2598369121551514, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7421127557754517, + "num_tokens": 288449344.0, + "step": 11317 + }, + { + "epoch": 1.242916758181419, + "grad_norm": 2.3560683727264404, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7166207432746887, + "num_tokens": 288472185.0, + "step": 11318 + }, + { + "epoch": 1.2430265758840324, + "grad_norm": 2.229553461074829, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.741549015045166, + "num_tokens": 288495812.0, + "step": 11319 + }, + { + "epoch": 1.2431363935866462, + "grad_norm": 2.0123984813690186, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.6973258852958679, + "num_tokens": 288527283.0, + "step": 11320 + }, + { + "epoch": 1.24324621128926, + "grad_norm": 2.162980079650879, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7250120639801025, + "num_tokens": 288552812.0, + "step": 11321 + }, + { + "epoch": 1.2433560289918735, + "grad_norm": 2.6825389862060547, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7260158061981201, + "num_tokens": 288572756.0, + "step": 11322 + }, + { + "epoch": 1.2434658466944872, + "grad_norm": 2.070375680923462, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7241896986961365, + "num_tokens": 288603190.0, + "step": 11323 + }, + { + "epoch": 1.2435756643971008, + "grad_norm": 2.0698344707489014, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7000223398208618, + "num_tokens": 288633049.0, + "step": 11324 + }, + { + "epoch": 1.2436854820997145, + "grad_norm": 2.4257965087890625, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7216530442237854, + "num_tokens": 288656258.0, + "step": 11325 + }, + { + "epoch": 1.243795299802328, + "grad_norm": 2.155385732650757, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7324225902557373, + "num_tokens": 288685537.0, + "step": 11326 + }, + { + "epoch": 1.2439051175049418, + "grad_norm": 2.257744312286377, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7168474197387695, + "num_tokens": 288712672.0, + "step": 11327 + }, + { + "epoch": 1.2440149352075554, + "grad_norm": 2.2818474769592285, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.725957989692688, + "num_tokens": 288736518.0, + "step": 11328 + }, + { + "epoch": 1.244124752910169, + "grad_norm": 2.277183771133423, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7155107259750366, + "num_tokens": 288761242.0, + "step": 11329 + }, + { + "epoch": 1.2442345706127829, + "grad_norm": 2.2387077808380127, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7293220162391663, + "num_tokens": 288787814.0, + "step": 11330 + }, + { + "epoch": 1.2443443883153964, + "grad_norm": 2.5734903812408447, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7423496246337891, + "num_tokens": 288808821.0, + "step": 11331 + }, + { + "epoch": 1.2444542060180102, + "grad_norm": 2.1607697010040283, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7201390862464905, + "num_tokens": 288836672.0, + "step": 11332 + }, + { + "epoch": 1.2445640237206237, + "grad_norm": 2.206268548965454, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7263785600662231, + "num_tokens": 288863677.0, + "step": 11333 + }, + { + "epoch": 1.2446738414232374, + "grad_norm": 2.303800106048584, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7241297960281372, + "num_tokens": 288886160.0, + "step": 11334 + }, + { + "epoch": 1.2447836591258512, + "grad_norm": 2.2828173637390137, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7205811738967896, + "num_tokens": 288913204.0, + "step": 11335 + }, + { + "epoch": 1.2448934768284647, + "grad_norm": 2.300689220428467, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7279816269874573, + "num_tokens": 288939346.0, + "step": 11336 + }, + { + "epoch": 1.2450032945310785, + "grad_norm": 2.1812174320220947, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7273955345153809, + "num_tokens": 288966462.0, + "step": 11337 + }, + { + "epoch": 1.245113112233692, + "grad_norm": 2.2800612449645996, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7028875350952148, + "num_tokens": 288992049.0, + "step": 11338 + }, + { + "epoch": 1.2452229299363058, + "grad_norm": 2.340073823928833, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7264045476913452, + "num_tokens": 289013291.0, + "step": 11339 + }, + { + "epoch": 1.2453327476389193, + "grad_norm": 2.419306755065918, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7255764007568359, + "num_tokens": 289037478.0, + "step": 11340 + }, + { + "epoch": 1.245442565341533, + "grad_norm": 2.1871535778045654, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7071841955184937, + "num_tokens": 289063121.0, + "step": 11341 + }, + { + "epoch": 1.2455523830441466, + "grad_norm": 2.0181119441986084, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7005094289779663, + "num_tokens": 289092627.0, + "step": 11342 + }, + { + "epoch": 1.2456622007467604, + "grad_norm": 2.6437931060791016, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7484172582626343, + "num_tokens": 289112344.0, + "step": 11343 + }, + { + "epoch": 1.2457720184493741, + "grad_norm": 2.567861795425415, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.6987975835800171, + "num_tokens": 289135534.0, + "step": 11344 + }, + { + "epoch": 1.2458818361519877, + "grad_norm": 2.2435762882232666, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7137348651885986, + "num_tokens": 289161144.0, + "step": 11345 + }, + { + "epoch": 1.2459916538546014, + "grad_norm": 2.3537442684173584, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7140972018241882, + "num_tokens": 289184175.0, + "step": 11346 + }, + { + "epoch": 1.246101471557215, + "grad_norm": 2.3964152336120605, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.73891282081604, + "num_tokens": 289207746.0, + "step": 11347 + }, + { + "epoch": 1.2462112892598287, + "grad_norm": 2.14925479888916, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7234920263290405, + "num_tokens": 289235097.0, + "step": 11348 + }, + { + "epoch": 1.2463211069624422, + "grad_norm": 2.3400909900665283, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6949387788772583, + "num_tokens": 289261214.0, + "step": 11349 + }, + { + "epoch": 1.246430924665056, + "grad_norm": 2.073254108428955, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7020006775856018, + "num_tokens": 289291229.0, + "step": 11350 + }, + { + "epoch": 1.2465407423676698, + "grad_norm": 2.14739727973938, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.71848464012146, + "num_tokens": 289318071.0, + "step": 11351 + }, + { + "epoch": 1.2466505600702833, + "grad_norm": 2.2030210494995117, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7080733776092529, + "num_tokens": 289344530.0, + "step": 11352 + }, + { + "epoch": 1.246760377772897, + "grad_norm": 2.291879415512085, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7154509425163269, + "num_tokens": 289370119.0, + "step": 11353 + }, + { + "epoch": 1.2468701954755106, + "grad_norm": 2.329164505004883, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7155745029449463, + "num_tokens": 289394028.0, + "step": 11354 + }, + { + "epoch": 1.2469800131781243, + "grad_norm": 2.5475473403930664, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7335432767868042, + "num_tokens": 289413839.0, + "step": 11355 + }, + { + "epoch": 1.2470898308807379, + "grad_norm": 2.1244924068450928, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7085940837860107, + "num_tokens": 289443180.0, + "step": 11356 + }, + { + "epoch": 1.2471996485833516, + "grad_norm": 1.9926618337631226, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6983314156532288, + "num_tokens": 289476512.0, + "step": 11357 + }, + { + "epoch": 1.2473094662859654, + "grad_norm": 2.2215023040771484, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7220852375030518, + "num_tokens": 289502834.0, + "step": 11358 + }, + { + "epoch": 1.247419283988579, + "grad_norm": 2.0328586101531982, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.722762405872345, + "num_tokens": 289530782.0, + "step": 11359 + }, + { + "epoch": 1.2475291016911927, + "grad_norm": 2.0916149616241455, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.70322185754776, + "num_tokens": 289560632.0, + "step": 11360 + }, + { + "epoch": 1.2476389193938062, + "grad_norm": 2.401824951171875, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7411733865737915, + "num_tokens": 289583311.0, + "step": 11361 + }, + { + "epoch": 1.24774873709642, + "grad_norm": 2.0660951137542725, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7117582559585571, + "num_tokens": 289611526.0, + "step": 11362 + }, + { + "epoch": 1.2478585547990335, + "grad_norm": 2.4574053287506104, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7327381372451782, + "num_tokens": 289632891.0, + "step": 11363 + }, + { + "epoch": 1.2479683725016473, + "grad_norm": 2.3695926666259766, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7279943227767944, + "num_tokens": 289656842.0, + "step": 11364 + }, + { + "epoch": 1.2480781902042608, + "grad_norm": 2.6601650714874268, + "learning_rate": 1e-06, + "loss": 0.7988, + "mean_token_accuracy": 0.7422602772712708, + "num_tokens": 289675614.0, + "step": 11365 + }, + { + "epoch": 1.2481880079068746, + "grad_norm": 2.200836420059204, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7120629549026489, + "num_tokens": 289700621.0, + "step": 11366 + }, + { + "epoch": 1.2482978256094883, + "grad_norm": 2.106205940246582, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7095892429351807, + "num_tokens": 289730799.0, + "step": 11367 + }, + { + "epoch": 1.2484076433121019, + "grad_norm": 2.11989688873291, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7325742244720459, + "num_tokens": 289757896.0, + "step": 11368 + }, + { + "epoch": 1.2485174610147156, + "grad_norm": 2.5709683895111084, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7303581833839417, + "num_tokens": 289779011.0, + "step": 11369 + }, + { + "epoch": 1.2486272787173291, + "grad_norm": 2.3541126251220703, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7060662508010864, + "num_tokens": 289803163.0, + "step": 11370 + }, + { + "epoch": 1.248737096419943, + "grad_norm": 2.101405143737793, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7080447673797607, + "num_tokens": 289833135.0, + "step": 11371 + }, + { + "epoch": 1.2488469141225567, + "grad_norm": 2.166170835494995, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7272924780845642, + "num_tokens": 289861412.0, + "step": 11372 + }, + { + "epoch": 1.2489567318251702, + "grad_norm": 2.1952030658721924, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7122011780738831, + "num_tokens": 289887470.0, + "step": 11373 + }, + { + "epoch": 1.249066549527784, + "grad_norm": 2.32738995552063, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7481757402420044, + "num_tokens": 289910649.0, + "step": 11374 + }, + { + "epoch": 1.2491763672303975, + "grad_norm": 2.4682095050811768, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7223483324050903, + "num_tokens": 289932351.0, + "step": 11375 + }, + { + "epoch": 1.2492861849330112, + "grad_norm": 2.200554132461548, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6895381808280945, + "num_tokens": 289959818.0, + "step": 11376 + }, + { + "epoch": 1.2493960026356248, + "grad_norm": 2.438964605331421, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7469856142997742, + "num_tokens": 289980874.0, + "step": 11377 + }, + { + "epoch": 1.2495058203382385, + "grad_norm": 2.547869920730591, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7241495847702026, + "num_tokens": 290002228.0, + "step": 11378 + }, + { + "epoch": 1.249615638040852, + "grad_norm": 2.1009111404418945, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.713229775428772, + "num_tokens": 290029464.0, + "step": 11379 + }, + { + "epoch": 1.2497254557434658, + "grad_norm": 2.1999318599700928, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7316261529922485, + "num_tokens": 290054668.0, + "step": 11380 + }, + { + "epoch": 1.2498352734460796, + "grad_norm": 2.2551143169403076, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7079830765724182, + "num_tokens": 290079981.0, + "step": 11381 + }, + { + "epoch": 1.2499450911486931, + "grad_norm": 2.0817811489105225, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7104586958885193, + "num_tokens": 290106616.0, + "step": 11382 + }, + { + "epoch": 1.2500549088513069, + "grad_norm": 2.1388094425201416, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7052751779556274, + "num_tokens": 290134764.0, + "step": 11383 + }, + { + "epoch": 1.2501647265539204, + "grad_norm": 2.09905743598938, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7186050415039062, + "num_tokens": 290160168.0, + "step": 11384 + }, + { + "epoch": 1.2502745442565342, + "grad_norm": 2.3199901580810547, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7206629514694214, + "num_tokens": 290183560.0, + "step": 11385 + }, + { + "epoch": 1.250384361959148, + "grad_norm": 2.6560251712799072, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7188827395439148, + "num_tokens": 290204323.0, + "step": 11386 + }, + { + "epoch": 1.2504941796617615, + "grad_norm": 2.3914906978607178, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7169166207313538, + "num_tokens": 290226180.0, + "step": 11387 + }, + { + "epoch": 1.2506039973643752, + "grad_norm": 2.4301633834838867, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7194495797157288, + "num_tokens": 290248156.0, + "step": 11388 + }, + { + "epoch": 1.2507138150669888, + "grad_norm": 2.3358166217803955, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7500742673873901, + "num_tokens": 290272511.0, + "step": 11389 + }, + { + "epoch": 1.2508236327696025, + "grad_norm": 2.0029163360595703, + "learning_rate": 1e-06, + "loss": 1.0336, + "mean_token_accuracy": 0.6842816472053528, + "num_tokens": 290304795.0, + "step": 11390 + }, + { + "epoch": 1.250933450472216, + "grad_norm": 2.1440727710723877, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.708019495010376, + "num_tokens": 290335310.0, + "step": 11391 + }, + { + "epoch": 1.2510432681748298, + "grad_norm": 2.327169895172119, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7288115620613098, + "num_tokens": 290358591.0, + "step": 11392 + }, + { + "epoch": 1.2511530858774433, + "grad_norm": 2.598100423812866, + "learning_rate": 1e-06, + "loss": 0.8024, + "mean_token_accuracy": 0.744172990322113, + "num_tokens": 290377123.0, + "step": 11393 + }, + { + "epoch": 1.251262903580057, + "grad_norm": 2.216942071914673, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7376560568809509, + "num_tokens": 290402586.0, + "step": 11394 + }, + { + "epoch": 1.2513727212826709, + "grad_norm": 2.5169739723205566, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7333092093467712, + "num_tokens": 290423480.0, + "step": 11395 + }, + { + "epoch": 1.2514825389852844, + "grad_norm": 2.5128848552703857, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.713912844657898, + "num_tokens": 290447311.0, + "step": 11396 + }, + { + "epoch": 1.2515923566878981, + "grad_norm": 2.472120523452759, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.743313193321228, + "num_tokens": 290469326.0, + "step": 11397 + }, + { + "epoch": 1.2517021743905117, + "grad_norm": 2.1066646575927734, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.713215172290802, + "num_tokens": 290498655.0, + "step": 11398 + }, + { + "epoch": 1.2518119920931254, + "grad_norm": 2.2501165866851807, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7288707494735718, + "num_tokens": 290525379.0, + "step": 11399 + }, + { + "epoch": 1.2519218097957392, + "grad_norm": 2.379310369491577, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.731380045413971, + "num_tokens": 290548746.0, + "step": 11400 + }, + { + "epoch": 1.2520316274983527, + "grad_norm": 2.215226173400879, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7090505957603455, + "num_tokens": 290575189.0, + "step": 11401 + }, + { + "epoch": 1.2521414452009663, + "grad_norm": 2.2744271755218506, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7053474187850952, + "num_tokens": 290601019.0, + "step": 11402 + }, + { + "epoch": 1.25225126290358, + "grad_norm": 2.391878366470337, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7072433233261108, + "num_tokens": 290625826.0, + "step": 11403 + }, + { + "epoch": 1.2523610806061938, + "grad_norm": 2.449679374694824, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7319883108139038, + "num_tokens": 290647029.0, + "step": 11404 + }, + { + "epoch": 1.2524708983088073, + "grad_norm": 2.140523910522461, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7206902503967285, + "num_tokens": 290673578.0, + "step": 11405 + }, + { + "epoch": 1.252580716011421, + "grad_norm": 2.305478811264038, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7334710955619812, + "num_tokens": 290697899.0, + "step": 11406 + }, + { + "epoch": 1.2526905337140346, + "grad_norm": 2.347536563873291, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7347339987754822, + "num_tokens": 290722654.0, + "step": 11407 + }, + { + "epoch": 1.2528003514166484, + "grad_norm": 2.177938222885132, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7389981746673584, + "num_tokens": 290750442.0, + "step": 11408 + }, + { + "epoch": 1.2529101691192621, + "grad_norm": 2.483999729156494, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7270302772521973, + "num_tokens": 290771237.0, + "step": 11409 + }, + { + "epoch": 1.2530199868218757, + "grad_norm": 2.3220160007476807, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7242649793624878, + "num_tokens": 290795386.0, + "step": 11410 + }, + { + "epoch": 1.2531298045244894, + "grad_norm": 2.1685729026794434, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7227346301078796, + "num_tokens": 290821486.0, + "step": 11411 + }, + { + "epoch": 1.253239622227103, + "grad_norm": 2.5172550678253174, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.6988829970359802, + "num_tokens": 290843314.0, + "step": 11412 + }, + { + "epoch": 1.2533494399297167, + "grad_norm": 2.229794502258301, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.733919620513916, + "num_tokens": 290869024.0, + "step": 11413 + }, + { + "epoch": 1.2534592576323305, + "grad_norm": 2.214834690093994, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7209996581077576, + "num_tokens": 290895859.0, + "step": 11414 + }, + { + "epoch": 1.253569075334944, + "grad_norm": 2.058901071548462, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7273650169372559, + "num_tokens": 290925423.0, + "step": 11415 + }, + { + "epoch": 1.2536788930375575, + "grad_norm": 2.10756516456604, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7082480788230896, + "num_tokens": 290956031.0, + "step": 11416 + }, + { + "epoch": 1.2537887107401713, + "grad_norm": 2.0783822536468506, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.715286135673523, + "num_tokens": 290984627.0, + "step": 11417 + }, + { + "epoch": 1.253898528442785, + "grad_norm": 2.203217029571533, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7274339199066162, + "num_tokens": 291010656.0, + "step": 11418 + }, + { + "epoch": 1.2540083461453986, + "grad_norm": 2.2867014408111572, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7220590710639954, + "num_tokens": 291035760.0, + "step": 11419 + }, + { + "epoch": 1.2541181638480123, + "grad_norm": 2.104964017868042, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7168843746185303, + "num_tokens": 291063340.0, + "step": 11420 + }, + { + "epoch": 1.2542279815506259, + "grad_norm": 2.1072022914886475, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7198503613471985, + "num_tokens": 291092327.0, + "step": 11421 + }, + { + "epoch": 1.2543377992532396, + "grad_norm": 2.07714581489563, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7126027941703796, + "num_tokens": 291122245.0, + "step": 11422 + }, + { + "epoch": 1.2544476169558534, + "grad_norm": 2.160144090652466, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7212777733802795, + "num_tokens": 291148897.0, + "step": 11423 + }, + { + "epoch": 1.254557434658467, + "grad_norm": 2.104762077331543, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7235537767410278, + "num_tokens": 291176325.0, + "step": 11424 + }, + { + "epoch": 1.2546672523610807, + "grad_norm": 2.3871896266937256, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7316580414772034, + "num_tokens": 291198314.0, + "step": 11425 + }, + { + "epoch": 1.2547770700636942, + "grad_norm": 2.5033366680145264, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7041342258453369, + "num_tokens": 291220951.0, + "step": 11426 + }, + { + "epoch": 1.254886887766308, + "grad_norm": 2.198880195617676, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7115265130996704, + "num_tokens": 291248399.0, + "step": 11427 + }, + { + "epoch": 1.2549967054689217, + "grad_norm": 2.5630595684051514, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7285903692245483, + "num_tokens": 291267815.0, + "step": 11428 + }, + { + "epoch": 1.2551065231715353, + "grad_norm": 2.3420956134796143, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7248910665512085, + "num_tokens": 291293729.0, + "step": 11429 + }, + { + "epoch": 1.2552163408741488, + "grad_norm": 2.0732603073120117, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.6951390504837036, + "num_tokens": 291325153.0, + "step": 11430 + }, + { + "epoch": 1.2553261585767626, + "grad_norm": 2.6398673057556152, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7263190746307373, + "num_tokens": 291343876.0, + "step": 11431 + }, + { + "epoch": 1.2554359762793763, + "grad_norm": 2.426055908203125, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7334646582603455, + "num_tokens": 291366676.0, + "step": 11432 + }, + { + "epoch": 1.2555457939819898, + "grad_norm": 2.1261982917785645, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7333943843841553, + "num_tokens": 291394507.0, + "step": 11433 + }, + { + "epoch": 1.2556556116846036, + "grad_norm": 2.16835880279541, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7182642221450806, + "num_tokens": 291420321.0, + "step": 11434 + }, + { + "epoch": 1.2557654293872171, + "grad_norm": 2.2439959049224854, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.726121187210083, + "num_tokens": 291444973.0, + "step": 11435 + }, + { + "epoch": 1.255875247089831, + "grad_norm": 1.9973111152648926, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.692101240158081, + "num_tokens": 291479610.0, + "step": 11436 + }, + { + "epoch": 1.2559850647924446, + "grad_norm": 2.2204530239105225, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7300201654434204, + "num_tokens": 291504756.0, + "step": 11437 + }, + { + "epoch": 1.2560948824950582, + "grad_norm": 2.120332956314087, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7094094157218933, + "num_tokens": 291533794.0, + "step": 11438 + }, + { + "epoch": 1.256204700197672, + "grad_norm": 2.185715913772583, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7059133052825928, + "num_tokens": 291563556.0, + "step": 11439 + }, + { + "epoch": 1.2563145179002855, + "grad_norm": 2.5593225955963135, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7063605785369873, + "num_tokens": 291584460.0, + "step": 11440 + }, + { + "epoch": 1.2564243356028992, + "grad_norm": 2.2314846515655518, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7173677086830139, + "num_tokens": 291608812.0, + "step": 11441 + }, + { + "epoch": 1.2565341533055128, + "grad_norm": 2.585690498352051, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7272700071334839, + "num_tokens": 291629423.0, + "step": 11442 + }, + { + "epoch": 1.2566439710081265, + "grad_norm": 2.089761972427368, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7147325277328491, + "num_tokens": 291658521.0, + "step": 11443 + }, + { + "epoch": 1.25675378871074, + "grad_norm": 2.6337103843688965, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7476017475128174, + "num_tokens": 291676426.0, + "step": 11444 + }, + { + "epoch": 1.2568636064133538, + "grad_norm": 2.2279720306396484, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7320285439491272, + "num_tokens": 291700639.0, + "step": 11445 + }, + { + "epoch": 1.2569734241159676, + "grad_norm": 2.0393645763397217, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7350548505783081, + "num_tokens": 291729464.0, + "step": 11446 + }, + { + "epoch": 1.257083241818581, + "grad_norm": 2.2834911346435547, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7236019968986511, + "num_tokens": 291758823.0, + "step": 11447 + }, + { + "epoch": 1.2571930595211949, + "grad_norm": 2.163892984390259, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7024291753768921, + "num_tokens": 291786908.0, + "step": 11448 + }, + { + "epoch": 1.2573028772238084, + "grad_norm": 2.109347105026245, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7238223552703857, + "num_tokens": 291814717.0, + "step": 11449 + }, + { + "epoch": 1.2574126949264222, + "grad_norm": 2.4230294227600098, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7210391759872437, + "num_tokens": 291835995.0, + "step": 11450 + }, + { + "epoch": 1.257522512629036, + "grad_norm": 1.9532006978988647, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7133987545967102, + "num_tokens": 291867578.0, + "step": 11451 + }, + { + "epoch": 1.2576323303316495, + "grad_norm": 2.466555595397949, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7329961061477661, + "num_tokens": 291889374.0, + "step": 11452 + }, + { + "epoch": 1.257742148034263, + "grad_norm": 2.3537559509277344, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7101120948791504, + "num_tokens": 291914201.0, + "step": 11453 + }, + { + "epoch": 1.2578519657368767, + "grad_norm": 2.801811456680298, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7413996458053589, + "num_tokens": 291931935.0, + "step": 11454 + }, + { + "epoch": 1.2579617834394905, + "grad_norm": 2.078806161880493, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7203037142753601, + "num_tokens": 291959755.0, + "step": 11455 + }, + { + "epoch": 1.258071601142104, + "grad_norm": 2.221864700317383, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7290864586830139, + "num_tokens": 291986425.0, + "step": 11456 + }, + { + "epoch": 1.2581814188447178, + "grad_norm": 2.3689115047454834, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7227456569671631, + "num_tokens": 292011674.0, + "step": 11457 + }, + { + "epoch": 1.2582912365473313, + "grad_norm": 2.101459503173828, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7293391823768616, + "num_tokens": 292040445.0, + "step": 11458 + }, + { + "epoch": 1.258401054249945, + "grad_norm": 1.9470137357711792, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7348474264144897, + "num_tokens": 292072431.0, + "step": 11459 + }, + { + "epoch": 1.2585108719525588, + "grad_norm": 1.9951013326644897, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7073236703872681, + "num_tokens": 292105139.0, + "step": 11460 + }, + { + "epoch": 1.2586206896551724, + "grad_norm": 2.0803327560424805, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7340809106826782, + "num_tokens": 292133752.0, + "step": 11461 + }, + { + "epoch": 1.2587305073577861, + "grad_norm": 2.339226484298706, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.706389307975769, + "num_tokens": 292156837.0, + "step": 11462 + }, + { + "epoch": 1.2588403250603997, + "grad_norm": 2.3108415603637695, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7218025922775269, + "num_tokens": 292182973.0, + "step": 11463 + }, + { + "epoch": 1.2589501427630134, + "grad_norm": 1.970477819442749, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6928446888923645, + "num_tokens": 292216444.0, + "step": 11464 + }, + { + "epoch": 1.2590599604656272, + "grad_norm": 2.585768699645996, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7475703954696655, + "num_tokens": 292235913.0, + "step": 11465 + }, + { + "epoch": 1.2591697781682407, + "grad_norm": 2.2955501079559326, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7343933582305908, + "num_tokens": 292260073.0, + "step": 11466 + }, + { + "epoch": 1.2592795958708543, + "grad_norm": 1.997610092163086, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.6978659629821777, + "num_tokens": 292290060.0, + "step": 11467 + }, + { + "epoch": 1.259389413573468, + "grad_norm": 2.292123794555664, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7410092353820801, + "num_tokens": 292311495.0, + "step": 11468 + }, + { + "epoch": 1.2594992312760818, + "grad_norm": 2.122297525405884, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7165563106536865, + "num_tokens": 292336587.0, + "step": 11469 + }, + { + "epoch": 1.2596090489786953, + "grad_norm": 2.3209314346313477, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.694992184638977, + "num_tokens": 292361312.0, + "step": 11470 + }, + { + "epoch": 1.259718866681309, + "grad_norm": 2.1951744556427, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7211813926696777, + "num_tokens": 292389064.0, + "step": 11471 + }, + { + "epoch": 1.2598286843839226, + "grad_norm": 2.0995423793792725, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7332643866539001, + "num_tokens": 292417240.0, + "step": 11472 + }, + { + "epoch": 1.2599385020865363, + "grad_norm": 2.0800344944000244, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7123582363128662, + "num_tokens": 292443977.0, + "step": 11473 + }, + { + "epoch": 1.26004831978915, + "grad_norm": 2.051171064376831, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7097810506820679, + "num_tokens": 292472488.0, + "step": 11474 + }, + { + "epoch": 1.2601581374917636, + "grad_norm": 2.1888859272003174, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7288601398468018, + "num_tokens": 292498695.0, + "step": 11475 + }, + { + "epoch": 1.2602679551943774, + "grad_norm": 2.2134125232696533, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7034849524497986, + "num_tokens": 292527674.0, + "step": 11476 + }, + { + "epoch": 1.260377772896991, + "grad_norm": 2.4401490688323975, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7409192323684692, + "num_tokens": 292549813.0, + "step": 11477 + }, + { + "epoch": 1.2604875905996047, + "grad_norm": 2.0930252075195312, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7250609397888184, + "num_tokens": 292578514.0, + "step": 11478 + }, + { + "epoch": 1.2605974083022184, + "grad_norm": 2.5481176376342773, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7110483646392822, + "num_tokens": 292600962.0, + "step": 11479 + }, + { + "epoch": 1.260707226004832, + "grad_norm": 2.3505852222442627, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7348870635032654, + "num_tokens": 292623404.0, + "step": 11480 + }, + { + "epoch": 1.2608170437074455, + "grad_norm": 2.6507668495178223, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7157579660415649, + "num_tokens": 292644487.0, + "step": 11481 + }, + { + "epoch": 1.2609268614100593, + "grad_norm": 2.2171950340270996, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7172259092330933, + "num_tokens": 292672029.0, + "step": 11482 + }, + { + "epoch": 1.261036679112673, + "grad_norm": 2.237347364425659, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7007513046264648, + "num_tokens": 292698839.0, + "step": 11483 + }, + { + "epoch": 1.2611464968152866, + "grad_norm": 2.131122589111328, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.714210033416748, + "num_tokens": 292726933.0, + "step": 11484 + }, + { + "epoch": 1.2612563145179003, + "grad_norm": 2.2482690811157227, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7055795788764954, + "num_tokens": 292753332.0, + "step": 11485 + }, + { + "epoch": 1.2613661322205139, + "grad_norm": 2.4353606700897217, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7292706370353699, + "num_tokens": 292776322.0, + "step": 11486 + }, + { + "epoch": 1.2614759499231276, + "grad_norm": 2.2837371826171875, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.743887722492218, + "num_tokens": 292800793.0, + "step": 11487 + }, + { + "epoch": 1.2615857676257414, + "grad_norm": 2.036799669265747, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7181955575942993, + "num_tokens": 292832769.0, + "step": 11488 + }, + { + "epoch": 1.261695585328355, + "grad_norm": 2.354010581970215, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7476845979690552, + "num_tokens": 292852634.0, + "step": 11489 + }, + { + "epoch": 1.2618054030309687, + "grad_norm": 2.400273323059082, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.715185284614563, + "num_tokens": 292875937.0, + "step": 11490 + }, + { + "epoch": 1.2619152207335822, + "grad_norm": 2.5381720066070557, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7255318760871887, + "num_tokens": 292896088.0, + "step": 11491 + }, + { + "epoch": 1.262025038436196, + "grad_norm": 2.3818981647491455, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.7628406286239624, + "num_tokens": 292916848.0, + "step": 11492 + }, + { + "epoch": 1.2621348561388097, + "grad_norm": 2.2175137996673584, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7103815078735352, + "num_tokens": 292943571.0, + "step": 11493 + }, + { + "epoch": 1.2622446738414232, + "grad_norm": 2.2325825691223145, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7031479477882385, + "num_tokens": 292972277.0, + "step": 11494 + }, + { + "epoch": 1.2623544915440368, + "grad_norm": 2.2788286209106445, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7179211378097534, + "num_tokens": 292999426.0, + "step": 11495 + }, + { + "epoch": 1.2624643092466505, + "grad_norm": 2.342561721801758, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7072851657867432, + "num_tokens": 293024674.0, + "step": 11496 + }, + { + "epoch": 1.2625741269492643, + "grad_norm": 2.2199857234954834, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6889159679412842, + "num_tokens": 293050248.0, + "step": 11497 + }, + { + "epoch": 1.2626839446518778, + "grad_norm": 2.561049222946167, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7274460792541504, + "num_tokens": 293070962.0, + "step": 11498 + }, + { + "epoch": 1.2627937623544916, + "grad_norm": 2.352142810821533, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7184784412384033, + "num_tokens": 293093698.0, + "step": 11499 + }, + { + "epoch": 1.2629035800571051, + "grad_norm": 2.3373517990112305, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7143605947494507, + "num_tokens": 293117875.0, + "step": 11500 + }, + { + "epoch": 1.2630133977597189, + "grad_norm": 2.3308749198913574, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7259051203727722, + "num_tokens": 293142562.0, + "step": 11501 + }, + { + "epoch": 1.2631232154623326, + "grad_norm": 2.3288629055023193, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7308673858642578, + "num_tokens": 293167977.0, + "step": 11502 + }, + { + "epoch": 1.2632330331649462, + "grad_norm": 2.500173330307007, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7197953462600708, + "num_tokens": 293190232.0, + "step": 11503 + }, + { + "epoch": 1.26334285086756, + "grad_norm": 2.0052530765533447, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.737794816493988, + "num_tokens": 293219601.0, + "step": 11504 + }, + { + "epoch": 1.2634526685701735, + "grad_norm": 2.273134708404541, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.716478705406189, + "num_tokens": 293244939.0, + "step": 11505 + }, + { + "epoch": 1.2635624862727872, + "grad_norm": 2.415391445159912, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.756756603717804, + "num_tokens": 293267134.0, + "step": 11506 + }, + { + "epoch": 1.2636723039754008, + "grad_norm": 2.422752857208252, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7322185039520264, + "num_tokens": 293288480.0, + "step": 11507 + }, + { + "epoch": 1.2637821216780145, + "grad_norm": 2.1675424575805664, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7119181156158447, + "num_tokens": 293315786.0, + "step": 11508 + }, + { + "epoch": 1.263891939380628, + "grad_norm": 2.3863251209259033, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7129421830177307, + "num_tokens": 293338199.0, + "step": 11509 + }, + { + "epoch": 1.2640017570832418, + "grad_norm": 2.199554920196533, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7042255401611328, + "num_tokens": 293366343.0, + "step": 11510 + }, + { + "epoch": 1.2641115747858556, + "grad_norm": 2.7749249935150146, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7436866164207458, + "num_tokens": 293385245.0, + "step": 11511 + }, + { + "epoch": 1.264221392488469, + "grad_norm": 2.7096595764160156, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7383863925933838, + "num_tokens": 293402235.0, + "step": 11512 + }, + { + "epoch": 1.2643312101910829, + "grad_norm": 1.938297152519226, + "learning_rate": 1e-06, + "loss": 1.0228, + "mean_token_accuracy": 0.6919341087341309, + "num_tokens": 293437158.0, + "step": 11513 + }, + { + "epoch": 1.2644410278936964, + "grad_norm": 2.3763487339019775, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7225828170776367, + "num_tokens": 293461864.0, + "step": 11514 + }, + { + "epoch": 1.2645508455963101, + "grad_norm": 2.8226919174194336, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7285861372947693, + "num_tokens": 293479382.0, + "step": 11515 + }, + { + "epoch": 1.264660663298924, + "grad_norm": 2.4366705417633057, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7092402577400208, + "num_tokens": 293503672.0, + "step": 11516 + }, + { + "epoch": 1.2647704810015374, + "grad_norm": 2.0983216762542725, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7327035665512085, + "num_tokens": 293530667.0, + "step": 11517 + }, + { + "epoch": 1.264880298704151, + "grad_norm": 2.393278121948242, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7002567052841187, + "num_tokens": 293553971.0, + "step": 11518 + }, + { + "epoch": 1.2649901164067647, + "grad_norm": 2.319697141647339, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7212052345275879, + "num_tokens": 293578777.0, + "step": 11519 + }, + { + "epoch": 1.2650999341093785, + "grad_norm": 2.335087776184082, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6946845650672913, + "num_tokens": 293604500.0, + "step": 11520 + }, + { + "epoch": 1.265209751811992, + "grad_norm": 2.478043794631958, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7353209853172302, + "num_tokens": 293626058.0, + "step": 11521 + }, + { + "epoch": 1.2653195695146058, + "grad_norm": 2.199951410293579, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.710523784160614, + "num_tokens": 293651739.0, + "step": 11522 + }, + { + "epoch": 1.2654293872172193, + "grad_norm": 1.9944428205490112, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7125276327133179, + "num_tokens": 293683854.0, + "step": 11523 + }, + { + "epoch": 1.265539204919833, + "grad_norm": 2.448789119720459, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.7363004684448242, + "num_tokens": 293703801.0, + "step": 11524 + }, + { + "epoch": 1.2656490226224468, + "grad_norm": 2.203491687774658, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7121647596359253, + "num_tokens": 293731276.0, + "step": 11525 + }, + { + "epoch": 1.2657588403250604, + "grad_norm": 2.274773597717285, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7369987964630127, + "num_tokens": 293754411.0, + "step": 11526 + }, + { + "epoch": 1.2658686580276741, + "grad_norm": 2.2026331424713135, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6916504502296448, + "num_tokens": 293780871.0, + "step": 11527 + }, + { + "epoch": 1.2659784757302877, + "grad_norm": 2.484971761703491, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7373166084289551, + "num_tokens": 293802008.0, + "step": 11528 + }, + { + "epoch": 1.2660882934329014, + "grad_norm": 2.569988965988159, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7520071864128113, + "num_tokens": 293820379.0, + "step": 11529 + }, + { + "epoch": 1.2661981111355152, + "grad_norm": 2.375588893890381, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.737819492816925, + "num_tokens": 293843480.0, + "step": 11530 + }, + { + "epoch": 1.2663079288381287, + "grad_norm": 2.292001724243164, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.718950629234314, + "num_tokens": 293868551.0, + "step": 11531 + }, + { + "epoch": 1.2664177465407422, + "grad_norm": 2.037700891494751, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7115005254745483, + "num_tokens": 293898789.0, + "step": 11532 + }, + { + "epoch": 1.266527564243356, + "grad_norm": 2.297682523727417, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7262529730796814, + "num_tokens": 293924512.0, + "step": 11533 + }, + { + "epoch": 1.2666373819459698, + "grad_norm": 2.5292675495147705, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.73432457447052, + "num_tokens": 293949811.0, + "step": 11534 + }, + { + "epoch": 1.2667471996485833, + "grad_norm": 2.1388819217681885, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7108336687088013, + "num_tokens": 293977489.0, + "step": 11535 + }, + { + "epoch": 1.266857017351197, + "grad_norm": 2.1344711780548096, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.737925112247467, + "num_tokens": 294008476.0, + "step": 11536 + }, + { + "epoch": 1.2669668350538106, + "grad_norm": 2.10094952583313, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.7012161016464233, + "num_tokens": 294038518.0, + "step": 11537 + }, + { + "epoch": 1.2670766527564243, + "grad_norm": 1.9983516931533813, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7180708050727844, + "num_tokens": 294070466.0, + "step": 11538 + }, + { + "epoch": 1.267186470459038, + "grad_norm": 2.2184221744537354, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7190299034118652, + "num_tokens": 294095968.0, + "step": 11539 + }, + { + "epoch": 1.2672962881616516, + "grad_norm": 2.377180814743042, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.729401707649231, + "num_tokens": 294117976.0, + "step": 11540 + }, + { + "epoch": 1.2674061058642654, + "grad_norm": 2.3209068775177, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7241633534431458, + "num_tokens": 294143596.0, + "step": 11541 + }, + { + "epoch": 1.267515923566879, + "grad_norm": 2.4736483097076416, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7379769086837769, + "num_tokens": 294165520.0, + "step": 11542 + }, + { + "epoch": 1.2676257412694927, + "grad_norm": 2.1827824115753174, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7253401875495911, + "num_tokens": 294191885.0, + "step": 11543 + }, + { + "epoch": 1.2677355589721064, + "grad_norm": 2.2518670558929443, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7238602042198181, + "num_tokens": 294217348.0, + "step": 11544 + }, + { + "epoch": 1.26784537667472, + "grad_norm": 2.111860990524292, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7215397953987122, + "num_tokens": 294243389.0, + "step": 11545 + }, + { + "epoch": 1.2679551943773335, + "grad_norm": 2.074894905090332, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7114353179931641, + "num_tokens": 294272863.0, + "step": 11546 + }, + { + "epoch": 1.2680650120799473, + "grad_norm": 2.1549930572509766, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7129282355308533, + "num_tokens": 294301097.0, + "step": 11547 + }, + { + "epoch": 1.268174829782561, + "grad_norm": 2.060298442840576, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7049375772476196, + "num_tokens": 294334788.0, + "step": 11548 + }, + { + "epoch": 1.2682846474851746, + "grad_norm": 2.227555274963379, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7209814786911011, + "num_tokens": 294359812.0, + "step": 11549 + }, + { + "epoch": 1.2683944651877883, + "grad_norm": 2.320847511291504, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7059282064437866, + "num_tokens": 294386319.0, + "step": 11550 + }, + { + "epoch": 1.2685042828904018, + "grad_norm": 2.162449836730957, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7381119132041931, + "num_tokens": 294413080.0, + "step": 11551 + }, + { + "epoch": 1.2686141005930156, + "grad_norm": 2.4213826656341553, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7586044073104858, + "num_tokens": 294433629.0, + "step": 11552 + }, + { + "epoch": 1.2687239182956294, + "grad_norm": 2.3509459495544434, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7398524284362793, + "num_tokens": 294455652.0, + "step": 11553 + }, + { + "epoch": 1.268833735998243, + "grad_norm": 2.2280690670013428, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7284287214279175, + "num_tokens": 294481764.0, + "step": 11554 + }, + { + "epoch": 1.2689435537008567, + "grad_norm": 2.5029783248901367, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7108749151229858, + "num_tokens": 294504110.0, + "step": 11555 + }, + { + "epoch": 1.2690533714034702, + "grad_norm": 2.093083381652832, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7172213792800903, + "num_tokens": 294531284.0, + "step": 11556 + }, + { + "epoch": 1.269163189106084, + "grad_norm": 2.261165142059326, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7082729339599609, + "num_tokens": 294557241.0, + "step": 11557 + }, + { + "epoch": 1.2692730068086975, + "grad_norm": 2.4452600479125977, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7332258224487305, + "num_tokens": 294578741.0, + "step": 11558 + }, + { + "epoch": 1.2693828245113112, + "grad_norm": 2.3501131534576416, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.716468095779419, + "num_tokens": 294605289.0, + "step": 11559 + }, + { + "epoch": 1.2694926422139248, + "grad_norm": 2.2344002723693848, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7405794858932495, + "num_tokens": 294631248.0, + "step": 11560 + }, + { + "epoch": 1.2696024599165385, + "grad_norm": 2.4747109413146973, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7052591443061829, + "num_tokens": 294652586.0, + "step": 11561 + }, + { + "epoch": 1.2697122776191523, + "grad_norm": 2.2645263671875, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7150789499282837, + "num_tokens": 294677712.0, + "step": 11562 + }, + { + "epoch": 1.2698220953217658, + "grad_norm": 2.2273123264312744, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7303478717803955, + "num_tokens": 294704811.0, + "step": 11563 + }, + { + "epoch": 1.2699319130243796, + "grad_norm": 2.174471616744995, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7395976185798645, + "num_tokens": 294730740.0, + "step": 11564 + }, + { + "epoch": 1.2700417307269931, + "grad_norm": 2.3341987133026123, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7230989933013916, + "num_tokens": 294753218.0, + "step": 11565 + }, + { + "epoch": 1.2701515484296069, + "grad_norm": 2.1221508979797363, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7133510112762451, + "num_tokens": 294782877.0, + "step": 11566 + }, + { + "epoch": 1.2702613661322206, + "grad_norm": 2.4900715351104736, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7230736613273621, + "num_tokens": 294806054.0, + "step": 11567 + }, + { + "epoch": 1.2703711838348342, + "grad_norm": 2.5406250953674316, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7368390560150146, + "num_tokens": 294825848.0, + "step": 11568 + }, + { + "epoch": 1.270481001537448, + "grad_norm": 2.3607022762298584, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7169463634490967, + "num_tokens": 294848193.0, + "step": 11569 + }, + { + "epoch": 1.2705908192400615, + "grad_norm": 2.1286823749542236, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7064783573150635, + "num_tokens": 294876092.0, + "step": 11570 + }, + { + "epoch": 1.2707006369426752, + "grad_norm": 2.1788511276245117, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7328541874885559, + "num_tokens": 294902060.0, + "step": 11571 + }, + { + "epoch": 1.2708104546452887, + "grad_norm": 2.5938477516174316, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7243468761444092, + "num_tokens": 294922444.0, + "step": 11572 + }, + { + "epoch": 1.2709202723479025, + "grad_norm": 2.2646641731262207, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7358701229095459, + "num_tokens": 294947614.0, + "step": 11573 + }, + { + "epoch": 1.271030090050516, + "grad_norm": 2.508791446685791, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7222191095352173, + "num_tokens": 294970016.0, + "step": 11574 + }, + { + "epoch": 1.2711399077531298, + "grad_norm": 2.412041664123535, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7137465476989746, + "num_tokens": 294992348.0, + "step": 11575 + }, + { + "epoch": 1.2712497254557436, + "grad_norm": 1.9639170169830322, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.700170636177063, + "num_tokens": 295022631.0, + "step": 11576 + }, + { + "epoch": 1.271359543158357, + "grad_norm": 2.174726963043213, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7265228033065796, + "num_tokens": 295050017.0, + "step": 11577 + }, + { + "epoch": 1.2714693608609708, + "grad_norm": 2.135420799255371, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7432296872138977, + "num_tokens": 295076850.0, + "step": 11578 + }, + { + "epoch": 1.2715791785635844, + "grad_norm": 2.401026725769043, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7201593518257141, + "num_tokens": 295100234.0, + "step": 11579 + }, + { + "epoch": 1.2716889962661981, + "grad_norm": 2.2838916778564453, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7387479543685913, + "num_tokens": 295125046.0, + "step": 11580 + }, + { + "epoch": 1.271798813968812, + "grad_norm": 2.2938365936279297, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7400472164154053, + "num_tokens": 295148063.0, + "step": 11581 + }, + { + "epoch": 1.2719086316714254, + "grad_norm": 2.104865550994873, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7202697992324829, + "num_tokens": 295175231.0, + "step": 11582 + }, + { + "epoch": 1.272018449374039, + "grad_norm": 2.209416389465332, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7259705066680908, + "num_tokens": 295203088.0, + "step": 11583 + }, + { + "epoch": 1.2721282670766527, + "grad_norm": 2.3318839073181152, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7366260886192322, + "num_tokens": 295227995.0, + "step": 11584 + }, + { + "epoch": 1.2722380847792665, + "grad_norm": 1.9979041814804077, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7187648415565491, + "num_tokens": 295256948.0, + "step": 11585 + }, + { + "epoch": 1.27234790248188, + "grad_norm": 1.9904351234436035, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7180808782577515, + "num_tokens": 295287799.0, + "step": 11586 + }, + { + "epoch": 1.2724577201844938, + "grad_norm": 2.16692852973938, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.714143693447113, + "num_tokens": 295315908.0, + "step": 11587 + }, + { + "epoch": 1.2725675378871073, + "grad_norm": 2.1808719635009766, + "learning_rate": 1e-06, + "loss": 1.0329, + "mean_token_accuracy": 0.6880031824111938, + "num_tokens": 295342261.0, + "step": 11588 + }, + { + "epoch": 1.272677355589721, + "grad_norm": 2.3266801834106445, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7129287123680115, + "num_tokens": 295367455.0, + "step": 11589 + }, + { + "epoch": 1.2727871732923348, + "grad_norm": 2.11567759513855, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7351611852645874, + "num_tokens": 295394605.0, + "step": 11590 + }, + { + "epoch": 1.2728969909949484, + "grad_norm": 2.2373690605163574, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7084008455276489, + "num_tokens": 295423046.0, + "step": 11591 + }, + { + "epoch": 1.273006808697562, + "grad_norm": 2.108201026916504, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7177219390869141, + "num_tokens": 295451327.0, + "step": 11592 + }, + { + "epoch": 1.2731166264001756, + "grad_norm": 2.3873002529144287, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7153226137161255, + "num_tokens": 295474903.0, + "step": 11593 + }, + { + "epoch": 1.2732264441027894, + "grad_norm": 2.571157693862915, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7246133089065552, + "num_tokens": 295495896.0, + "step": 11594 + }, + { + "epoch": 1.2733362618054032, + "grad_norm": 2.2961347103118896, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7136408090591431, + "num_tokens": 295519336.0, + "step": 11595 + }, + { + "epoch": 1.2734460795080167, + "grad_norm": 2.436737060546875, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7119688391685486, + "num_tokens": 295542708.0, + "step": 11596 + }, + { + "epoch": 1.2735558972106302, + "grad_norm": 2.4575562477111816, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.712265133857727, + "num_tokens": 295565022.0, + "step": 11597 + }, + { + "epoch": 1.273665714913244, + "grad_norm": 2.059584856033325, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7049521207809448, + "num_tokens": 295594619.0, + "step": 11598 + }, + { + "epoch": 1.2737755326158577, + "grad_norm": 2.9515717029571533, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7231377363204956, + "num_tokens": 295610138.0, + "step": 11599 + }, + { + "epoch": 1.2738853503184713, + "grad_norm": 2.372140884399414, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7209981679916382, + "num_tokens": 295633410.0, + "step": 11600 + }, + { + "epoch": 1.273995168021085, + "grad_norm": 2.3402600288391113, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7237783670425415, + "num_tokens": 295656982.0, + "step": 11601 + }, + { + "epoch": 1.2741049857236986, + "grad_norm": 2.6093294620513916, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7359117865562439, + "num_tokens": 295675740.0, + "step": 11602 + }, + { + "epoch": 1.2742148034263123, + "grad_norm": 2.282198667526245, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7334001064300537, + "num_tokens": 295700860.0, + "step": 11603 + }, + { + "epoch": 1.274324621128926, + "grad_norm": 2.4434475898742676, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7125735282897949, + "num_tokens": 295723625.0, + "step": 11604 + }, + { + "epoch": 1.2744344388315396, + "grad_norm": 2.150158166885376, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7135865688323975, + "num_tokens": 295751880.0, + "step": 11605 + }, + { + "epoch": 1.2745442565341534, + "grad_norm": 2.769174098968506, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7196634411811829, + "num_tokens": 295770481.0, + "step": 11606 + }, + { + "epoch": 1.274654074236767, + "grad_norm": 2.272594690322876, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7268246412277222, + "num_tokens": 295795939.0, + "step": 11607 + }, + { + "epoch": 1.2747638919393807, + "grad_norm": 2.2342021465301514, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7061927914619446, + "num_tokens": 295822470.0, + "step": 11608 + }, + { + "epoch": 1.2748737096419944, + "grad_norm": 2.1818394660949707, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7148387432098389, + "num_tokens": 295848509.0, + "step": 11609 + }, + { + "epoch": 1.274983527344608, + "grad_norm": 2.115074396133423, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.709236741065979, + "num_tokens": 295877979.0, + "step": 11610 + }, + { + "epoch": 1.2750933450472215, + "grad_norm": 2.2315194606781006, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7290343642234802, + "num_tokens": 295902799.0, + "step": 11611 + }, + { + "epoch": 1.2752031627498353, + "grad_norm": 1.9556504487991333, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.716850996017456, + "num_tokens": 295933887.0, + "step": 11612 + }, + { + "epoch": 1.275312980452449, + "grad_norm": 2.141568422317505, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.6966985464096069, + "num_tokens": 295959911.0, + "step": 11613 + }, + { + "epoch": 1.2754227981550625, + "grad_norm": 2.33516526222229, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7035574316978455, + "num_tokens": 295983925.0, + "step": 11614 + }, + { + "epoch": 1.2755326158576763, + "grad_norm": 2.189220666885376, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7168890833854675, + "num_tokens": 296011106.0, + "step": 11615 + }, + { + "epoch": 1.2756424335602898, + "grad_norm": 2.35099720954895, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7088630199432373, + "num_tokens": 296035448.0, + "step": 11616 + }, + { + "epoch": 1.2757522512629036, + "grad_norm": 2.3767077922821045, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.728212833404541, + "num_tokens": 296056875.0, + "step": 11617 + }, + { + "epoch": 1.2758620689655173, + "grad_norm": 2.329451322555542, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.711816132068634, + "num_tokens": 296079538.0, + "step": 11618 + }, + { + "epoch": 1.2759718866681309, + "grad_norm": 2.2462470531463623, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.6938109993934631, + "num_tokens": 296105846.0, + "step": 11619 + }, + { + "epoch": 1.2760817043707446, + "grad_norm": 2.7425286769866943, + "learning_rate": 1e-06, + "loss": 0.7564, + "mean_token_accuracy": 0.7527895569801331, + "num_tokens": 296123231.0, + "step": 11620 + }, + { + "epoch": 1.2761915220733582, + "grad_norm": 2.2068450450897217, + "learning_rate": 1e-06, + "loss": 0.7692, + "mean_token_accuracy": 0.7549502849578857, + "num_tokens": 296148775.0, + "step": 11621 + }, + { + "epoch": 1.276301339775972, + "grad_norm": 2.3753890991210938, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7343538999557495, + "num_tokens": 296171972.0, + "step": 11622 + }, + { + "epoch": 1.2764111574785855, + "grad_norm": 2.1866812705993652, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7146158218383789, + "num_tokens": 296199125.0, + "step": 11623 + }, + { + "epoch": 1.2765209751811992, + "grad_norm": 2.285734176635742, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7235115766525269, + "num_tokens": 296224631.0, + "step": 11624 + }, + { + "epoch": 1.2766307928838128, + "grad_norm": 2.5798487663269043, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7321454286575317, + "num_tokens": 296245415.0, + "step": 11625 + }, + { + "epoch": 1.2767406105864265, + "grad_norm": 2.2004120349884033, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7108733057975769, + "num_tokens": 296274182.0, + "step": 11626 + }, + { + "epoch": 1.2768504282890403, + "grad_norm": 2.3077759742736816, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7051618099212646, + "num_tokens": 296297361.0, + "step": 11627 + }, + { + "epoch": 1.2769602459916538, + "grad_norm": 2.138051986694336, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6907541155815125, + "num_tokens": 296324135.0, + "step": 11628 + }, + { + "epoch": 1.2770700636942676, + "grad_norm": 2.2873575687408447, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7182227373123169, + "num_tokens": 296349931.0, + "step": 11629 + }, + { + "epoch": 1.277179881396881, + "grad_norm": 2.304595708847046, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7162781357765198, + "num_tokens": 296375124.0, + "step": 11630 + }, + { + "epoch": 1.2772896990994949, + "grad_norm": 2.191401243209839, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7288322448730469, + "num_tokens": 296400416.0, + "step": 11631 + }, + { + "epoch": 1.2773995168021086, + "grad_norm": 2.6550276279449463, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7328140735626221, + "num_tokens": 296418841.0, + "step": 11632 + }, + { + "epoch": 1.2775093345047221, + "grad_norm": 2.2181551456451416, + "learning_rate": 1e-06, + "loss": 1.0365, + "mean_token_accuracy": 0.6855847239494324, + "num_tokens": 296446750.0, + "step": 11633 + }, + { + "epoch": 1.2776191522073357, + "grad_norm": 2.8624417781829834, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.7424037456512451, + "num_tokens": 296463180.0, + "step": 11634 + }, + { + "epoch": 1.2777289699099494, + "grad_norm": 2.1475369930267334, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7236443758010864, + "num_tokens": 296488364.0, + "step": 11635 + }, + { + "epoch": 1.2778387876125632, + "grad_norm": 2.062947988510132, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7264364957809448, + "num_tokens": 296517149.0, + "step": 11636 + }, + { + "epoch": 1.2779486053151767, + "grad_norm": 2.356964588165283, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7212363481521606, + "num_tokens": 296539503.0, + "step": 11637 + }, + { + "epoch": 1.2780584230177905, + "grad_norm": 2.248472213745117, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.69903165102005, + "num_tokens": 296566529.0, + "step": 11638 + }, + { + "epoch": 1.278168240720404, + "grad_norm": 2.2206788063049316, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7194932699203491, + "num_tokens": 296592509.0, + "step": 11639 + }, + { + "epoch": 1.2782780584230178, + "grad_norm": 2.113090753555298, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7142413854598999, + "num_tokens": 296618529.0, + "step": 11640 + }, + { + "epoch": 1.2783878761256315, + "grad_norm": 1.9858866930007935, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7231768369674683, + "num_tokens": 296649146.0, + "step": 11641 + }, + { + "epoch": 1.278497693828245, + "grad_norm": 2.2198565006256104, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7120842933654785, + "num_tokens": 296676129.0, + "step": 11642 + }, + { + "epoch": 1.2786075115308588, + "grad_norm": 2.197828769683838, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7186952829360962, + "num_tokens": 296701645.0, + "step": 11643 + }, + { + "epoch": 1.2787173292334724, + "grad_norm": 2.2941062450408936, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7055892944335938, + "num_tokens": 296726952.0, + "step": 11644 + }, + { + "epoch": 1.2788271469360861, + "grad_norm": 2.154254913330078, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7375116944313049, + "num_tokens": 296751830.0, + "step": 11645 + }, + { + "epoch": 1.2789369646386999, + "grad_norm": 2.698124647140503, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7345243692398071, + "num_tokens": 296770781.0, + "step": 11646 + }, + { + "epoch": 1.2790467823413134, + "grad_norm": 2.1912291049957275, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7269643545150757, + "num_tokens": 296797101.0, + "step": 11647 + }, + { + "epoch": 1.279156600043927, + "grad_norm": 2.1795079708099365, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7217792868614197, + "num_tokens": 296823894.0, + "step": 11648 + }, + { + "epoch": 1.2792664177465407, + "grad_norm": 2.7477519512176514, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7287513017654419, + "num_tokens": 296842110.0, + "step": 11649 + }, + { + "epoch": 1.2793762354491545, + "grad_norm": 2.338418483734131, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7177999019622803, + "num_tokens": 296866690.0, + "step": 11650 + }, + { + "epoch": 1.279486053151768, + "grad_norm": 2.311244010925293, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7238306403160095, + "num_tokens": 296890136.0, + "step": 11651 + }, + { + "epoch": 1.2795958708543818, + "grad_norm": 2.3699002265930176, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.72974693775177, + "num_tokens": 296916681.0, + "step": 11652 + }, + { + "epoch": 1.2797056885569953, + "grad_norm": 2.5350263118743896, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7119081020355225, + "num_tokens": 296939161.0, + "step": 11653 + }, + { + "epoch": 1.279815506259609, + "grad_norm": 2.205183506011963, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7222750782966614, + "num_tokens": 296963002.0, + "step": 11654 + }, + { + "epoch": 1.2799253239622228, + "grad_norm": 2.208287000656128, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.6956501603126526, + "num_tokens": 296990307.0, + "step": 11655 + }, + { + "epoch": 1.2800351416648363, + "grad_norm": 2.044370412826538, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.7016617059707642, + "num_tokens": 297018880.0, + "step": 11656 + }, + { + "epoch": 1.28014495936745, + "grad_norm": 2.472398281097412, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7423756122589111, + "num_tokens": 297039397.0, + "step": 11657 + }, + { + "epoch": 1.2802547770700636, + "grad_norm": 2.1335299015045166, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7092658877372742, + "num_tokens": 297067737.0, + "step": 11658 + }, + { + "epoch": 1.2803645947726774, + "grad_norm": 2.206332206726074, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7265325784683228, + "num_tokens": 297094212.0, + "step": 11659 + }, + { + "epoch": 1.2804744124752911, + "grad_norm": 2.85603928565979, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7234275341033936, + "num_tokens": 297114474.0, + "step": 11660 + }, + { + "epoch": 1.2805842301779047, + "grad_norm": 2.476153612136841, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.722568154335022, + "num_tokens": 297137428.0, + "step": 11661 + }, + { + "epoch": 1.2806940478805182, + "grad_norm": 2.462293863296509, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7398701310157776, + "num_tokens": 297157850.0, + "step": 11662 + }, + { + "epoch": 1.280803865583132, + "grad_norm": 2.186335563659668, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7016823887825012, + "num_tokens": 297184363.0, + "step": 11663 + }, + { + "epoch": 1.2809136832857457, + "grad_norm": 2.1155459880828857, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.6975805163383484, + "num_tokens": 297212943.0, + "step": 11664 + }, + { + "epoch": 1.2810235009883593, + "grad_norm": 2.4019391536712646, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6908121109008789, + "num_tokens": 297238365.0, + "step": 11665 + }, + { + "epoch": 1.281133318690973, + "grad_norm": 2.0072758197784424, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7256043553352356, + "num_tokens": 297268619.0, + "step": 11666 + }, + { + "epoch": 1.2812431363935866, + "grad_norm": 2.178929567337036, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7202423810958862, + "num_tokens": 297295921.0, + "step": 11667 + }, + { + "epoch": 1.2813529540962003, + "grad_norm": 2.074347972869873, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7038462162017822, + "num_tokens": 297326310.0, + "step": 11668 + }, + { + "epoch": 1.281462771798814, + "grad_norm": 2.231898069381714, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.72216796875, + "num_tokens": 297352297.0, + "step": 11669 + }, + { + "epoch": 1.2815725895014276, + "grad_norm": 2.486008644104004, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7370408773422241, + "num_tokens": 297375453.0, + "step": 11670 + }, + { + "epoch": 1.2816824072040414, + "grad_norm": 2.28936767578125, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.72626131772995, + "num_tokens": 297401035.0, + "step": 11671 + }, + { + "epoch": 1.281792224906655, + "grad_norm": 2.1317951679229736, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7218095064163208, + "num_tokens": 297427843.0, + "step": 11672 + }, + { + "epoch": 1.2819020426092687, + "grad_norm": 2.6297006607055664, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7367892265319824, + "num_tokens": 297447182.0, + "step": 11673 + }, + { + "epoch": 1.2820118603118824, + "grad_norm": 2.72226881980896, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7449137568473816, + "num_tokens": 297465386.0, + "step": 11674 + }, + { + "epoch": 1.282121678014496, + "grad_norm": 2.457274913787842, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7416050434112549, + "num_tokens": 297486811.0, + "step": 11675 + }, + { + "epoch": 1.2822314957171095, + "grad_norm": 2.201220989227295, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.7585033774375916, + "num_tokens": 297509993.0, + "step": 11676 + }, + { + "epoch": 1.2823413134197232, + "grad_norm": 2.308568000793457, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7442650198936462, + "num_tokens": 297533430.0, + "step": 11677 + }, + { + "epoch": 1.282451131122337, + "grad_norm": 2.311246395111084, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.7037590146064758, + "num_tokens": 297562917.0, + "step": 11678 + }, + { + "epoch": 1.2825609488249505, + "grad_norm": 2.32633376121521, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7185313701629639, + "num_tokens": 297587678.0, + "step": 11679 + }, + { + "epoch": 1.2826707665275643, + "grad_norm": 2.1345608234405518, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7180681228637695, + "num_tokens": 297615613.0, + "step": 11680 + }, + { + "epoch": 1.2827805842301778, + "grad_norm": 2.5409975051879883, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7350274324417114, + "num_tokens": 297637110.0, + "step": 11681 + }, + { + "epoch": 1.2828904019327916, + "grad_norm": 2.1523749828338623, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7202125787734985, + "num_tokens": 297665490.0, + "step": 11682 + }, + { + "epoch": 1.2830002196354053, + "grad_norm": 2.045696973800659, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.703449010848999, + "num_tokens": 297694466.0, + "step": 11683 + }, + { + "epoch": 1.2831100373380189, + "grad_norm": 2.43245792388916, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7269162535667419, + "num_tokens": 297717399.0, + "step": 11684 + }, + { + "epoch": 1.2832198550406326, + "grad_norm": 2.468858003616333, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7458267211914062, + "num_tokens": 297738241.0, + "step": 11685 + }, + { + "epoch": 1.2833296727432462, + "grad_norm": 2.3831393718719482, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7156876921653748, + "num_tokens": 297763475.0, + "step": 11686 + }, + { + "epoch": 1.28343949044586, + "grad_norm": 2.355048894882202, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7114059329032898, + "num_tokens": 297787025.0, + "step": 11687 + }, + { + "epoch": 1.2835493081484735, + "grad_norm": 2.410372018814087, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7153385877609253, + "num_tokens": 297810209.0, + "step": 11688 + }, + { + "epoch": 1.2836591258510872, + "grad_norm": 2.15640926361084, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7086246013641357, + "num_tokens": 297836094.0, + "step": 11689 + }, + { + "epoch": 1.2837689435537007, + "grad_norm": 2.3716394901275635, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7152769565582275, + "num_tokens": 297858315.0, + "step": 11690 + }, + { + "epoch": 1.2838787612563145, + "grad_norm": 2.1089065074920654, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7455743551254272, + "num_tokens": 297886086.0, + "step": 11691 + }, + { + "epoch": 1.2839885789589283, + "grad_norm": 2.623971939086914, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7309086322784424, + "num_tokens": 297906455.0, + "step": 11692 + }, + { + "epoch": 1.2840983966615418, + "grad_norm": 1.982554316520691, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7306447625160217, + "num_tokens": 297934320.0, + "step": 11693 + }, + { + "epoch": 1.2842082143641556, + "grad_norm": 2.2945048809051514, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7117851972579956, + "num_tokens": 297959338.0, + "step": 11694 + }, + { + "epoch": 1.284318032066769, + "grad_norm": 2.619597911834717, + "learning_rate": 1e-06, + "loss": 0.8014, + "mean_token_accuracy": 0.7404437065124512, + "num_tokens": 297977644.0, + "step": 11695 + }, + { + "epoch": 1.2844278497693828, + "grad_norm": 2.355741024017334, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7283132076263428, + "num_tokens": 298002919.0, + "step": 11696 + }, + { + "epoch": 1.2845376674719966, + "grad_norm": 2.1276443004608154, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7166548371315002, + "num_tokens": 298031198.0, + "step": 11697 + }, + { + "epoch": 1.2846474851746101, + "grad_norm": 2.3541266918182373, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7172364592552185, + "num_tokens": 298056263.0, + "step": 11698 + }, + { + "epoch": 1.2847573028772237, + "grad_norm": 2.255679130554199, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7119389176368713, + "num_tokens": 298080847.0, + "step": 11699 + }, + { + "epoch": 1.2848671205798374, + "grad_norm": 2.1707401275634766, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7040090560913086, + "num_tokens": 298111861.0, + "step": 11700 + }, + { + "epoch": 1.2849769382824512, + "grad_norm": 2.39139461517334, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7256777286529541, + "num_tokens": 298136035.0, + "step": 11701 + }, + { + "epoch": 1.2850867559850647, + "grad_norm": 2.240351915359497, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7147430777549744, + "num_tokens": 298163099.0, + "step": 11702 + }, + { + "epoch": 1.2851965736876785, + "grad_norm": 2.0433192253112793, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7292400598526001, + "num_tokens": 298193396.0, + "step": 11703 + }, + { + "epoch": 1.285306391390292, + "grad_norm": 2.065791368484497, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7027469873428345, + "num_tokens": 298222102.0, + "step": 11704 + }, + { + "epoch": 1.2854162090929058, + "grad_norm": 2.276064395904541, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7371765971183777, + "num_tokens": 298244470.0, + "step": 11705 + }, + { + "epoch": 1.2855260267955195, + "grad_norm": 1.8826615810394287, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7314601540565491, + "num_tokens": 298277267.0, + "step": 11706 + }, + { + "epoch": 1.285635844498133, + "grad_norm": 2.449432134628296, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7348098754882812, + "num_tokens": 298299101.0, + "step": 11707 + }, + { + "epoch": 1.2857456622007468, + "grad_norm": 2.2759342193603516, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7352041006088257, + "num_tokens": 298325140.0, + "step": 11708 + }, + { + "epoch": 1.2858554799033604, + "grad_norm": 2.129669666290283, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7260384559631348, + "num_tokens": 298352124.0, + "step": 11709 + }, + { + "epoch": 1.2859652976059741, + "grad_norm": 1.9632130861282349, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7211508750915527, + "num_tokens": 298386955.0, + "step": 11710 + }, + { + "epoch": 1.2860751153085879, + "grad_norm": 2.3207669258117676, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7202012538909912, + "num_tokens": 298413066.0, + "step": 11711 + }, + { + "epoch": 1.2861849330112014, + "grad_norm": 2.367340087890625, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7139236927032471, + "num_tokens": 298437682.0, + "step": 11712 + }, + { + "epoch": 1.286294750713815, + "grad_norm": 2.359523057937622, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7152128219604492, + "num_tokens": 298461463.0, + "step": 11713 + }, + { + "epoch": 1.2864045684164287, + "grad_norm": 2.2492384910583496, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.730582594871521, + "num_tokens": 298486616.0, + "step": 11714 + }, + { + "epoch": 1.2865143861190425, + "grad_norm": 2.11613130569458, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.702582836151123, + "num_tokens": 298515449.0, + "step": 11715 + }, + { + "epoch": 1.286624203821656, + "grad_norm": 2.4998281002044678, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7470725774765015, + "num_tokens": 298536391.0, + "step": 11716 + }, + { + "epoch": 1.2867340215242697, + "grad_norm": 2.4093434810638428, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7304002046585083, + "num_tokens": 298558021.0, + "step": 11717 + }, + { + "epoch": 1.2868438392268833, + "grad_norm": 2.278766393661499, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7019352912902832, + "num_tokens": 298583416.0, + "step": 11718 + }, + { + "epoch": 1.286953656929497, + "grad_norm": 2.0281178951263428, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.707909345626831, + "num_tokens": 298612852.0, + "step": 11719 + }, + { + "epoch": 1.2870634746321108, + "grad_norm": 2.3326313495635986, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7392350435256958, + "num_tokens": 298636075.0, + "step": 11720 + }, + { + "epoch": 1.2871732923347243, + "grad_norm": 2.058049201965332, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7186852097511292, + "num_tokens": 298665563.0, + "step": 11721 + }, + { + "epoch": 1.287283110037338, + "grad_norm": 2.339508533477783, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.714601457118988, + "num_tokens": 298689745.0, + "step": 11722 + }, + { + "epoch": 1.2873929277399516, + "grad_norm": 2.214362144470215, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7450723648071289, + "num_tokens": 298716059.0, + "step": 11723 + }, + { + "epoch": 1.2875027454425654, + "grad_norm": 2.361513614654541, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7108175754547119, + "num_tokens": 298739123.0, + "step": 11724 + }, + { + "epoch": 1.2876125631451791, + "grad_norm": 2.0596628189086914, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7079353928565979, + "num_tokens": 298770064.0, + "step": 11725 + }, + { + "epoch": 1.2877223808477927, + "grad_norm": 2.2376086711883545, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7233084440231323, + "num_tokens": 298796881.0, + "step": 11726 + }, + { + "epoch": 1.2878321985504062, + "grad_norm": 2.291076183319092, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7307775020599365, + "num_tokens": 298821483.0, + "step": 11727 + }, + { + "epoch": 1.28794201625302, + "grad_norm": 2.225261926651001, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.6995083093643188, + "num_tokens": 298850101.0, + "step": 11728 + }, + { + "epoch": 1.2880518339556337, + "grad_norm": 2.330625057220459, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7169809341430664, + "num_tokens": 298875826.0, + "step": 11729 + }, + { + "epoch": 1.2881616516582473, + "grad_norm": 2.5079219341278076, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7280626893043518, + "num_tokens": 298897271.0, + "step": 11730 + }, + { + "epoch": 1.288271469360861, + "grad_norm": 2.334528684616089, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7063096165657043, + "num_tokens": 298920314.0, + "step": 11731 + }, + { + "epoch": 1.2883812870634745, + "grad_norm": 2.131762742996216, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7094687223434448, + "num_tokens": 298948772.0, + "step": 11732 + }, + { + "epoch": 1.2884911047660883, + "grad_norm": 2.245335102081299, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.6982362270355225, + "num_tokens": 298975593.0, + "step": 11733 + }, + { + "epoch": 1.288600922468702, + "grad_norm": 2.3388783931732178, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7211069464683533, + "num_tokens": 298999913.0, + "step": 11734 + }, + { + "epoch": 1.2887107401713156, + "grad_norm": 2.371445655822754, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7353833317756653, + "num_tokens": 299023830.0, + "step": 11735 + }, + { + "epoch": 1.2888205578739294, + "grad_norm": 1.9257848262786865, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6947752833366394, + "num_tokens": 299059737.0, + "step": 11736 + }, + { + "epoch": 1.2889303755765429, + "grad_norm": 1.9431415796279907, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7379541397094727, + "num_tokens": 299089871.0, + "step": 11737 + }, + { + "epoch": 1.2890401932791566, + "grad_norm": 2.481325626373291, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7420160174369812, + "num_tokens": 299110633.0, + "step": 11738 + }, + { + "epoch": 1.2891500109817704, + "grad_norm": 2.2192459106445312, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7273393869400024, + "num_tokens": 299137433.0, + "step": 11739 + }, + { + "epoch": 1.289259828684384, + "grad_norm": 2.1984567642211914, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.6946405172348022, + "num_tokens": 299165312.0, + "step": 11740 + }, + { + "epoch": 1.2893696463869975, + "grad_norm": 2.0732009410858154, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.725543200969696, + "num_tokens": 299196913.0, + "step": 11741 + }, + { + "epoch": 1.2894794640896112, + "grad_norm": 2.258314371109009, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7041199207305908, + "num_tokens": 299223497.0, + "step": 11742 + }, + { + "epoch": 1.289589281792225, + "grad_norm": 2.0488617420196533, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7174761295318604, + "num_tokens": 299253919.0, + "step": 11743 + }, + { + "epoch": 1.2896990994948385, + "grad_norm": 2.2271528244018555, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7463310956954956, + "num_tokens": 299279193.0, + "step": 11744 + }, + { + "epoch": 1.2898089171974523, + "grad_norm": 2.1914782524108887, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7321873903274536, + "num_tokens": 299305035.0, + "step": 11745 + }, + { + "epoch": 1.2899187349000658, + "grad_norm": 2.3737077713012695, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7081472873687744, + "num_tokens": 299330829.0, + "step": 11746 + }, + { + "epoch": 1.2900285526026796, + "grad_norm": 2.3987534046173096, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7262969017028809, + "num_tokens": 299353836.0, + "step": 11747 + }, + { + "epoch": 1.2901383703052933, + "grad_norm": 2.0454506874084473, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7170840501785278, + "num_tokens": 299385964.0, + "step": 11748 + }, + { + "epoch": 1.2902481880079069, + "grad_norm": 2.4502975940704346, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7175371050834656, + "num_tokens": 299408319.0, + "step": 11749 + }, + { + "epoch": 1.2903580057105206, + "grad_norm": 2.1727354526519775, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7358007431030273, + "num_tokens": 299434658.0, + "step": 11750 + }, + { + "epoch": 1.2904678234131342, + "grad_norm": 2.3153836727142334, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7087874412536621, + "num_tokens": 299461503.0, + "step": 11751 + }, + { + "epoch": 1.290577641115748, + "grad_norm": 2.139957904815674, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7176055312156677, + "num_tokens": 299488027.0, + "step": 11752 + }, + { + "epoch": 1.2906874588183614, + "grad_norm": 2.348475933074951, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7106605768203735, + "num_tokens": 299511373.0, + "step": 11753 + }, + { + "epoch": 1.2907972765209752, + "grad_norm": 2.036080837249756, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7020642757415771, + "num_tokens": 299540667.0, + "step": 11754 + }, + { + "epoch": 1.2909070942235887, + "grad_norm": 2.644446611404419, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7366379499435425, + "num_tokens": 299559672.0, + "step": 11755 + }, + { + "epoch": 1.2910169119262025, + "grad_norm": 2.5956454277038574, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7314766645431519, + "num_tokens": 299580961.0, + "step": 11756 + }, + { + "epoch": 1.2911267296288162, + "grad_norm": 2.284536361694336, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7122617363929749, + "num_tokens": 299607634.0, + "step": 11757 + }, + { + "epoch": 1.2912365473314298, + "grad_norm": 2.331378698348999, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.6981165409088135, + "num_tokens": 299631591.0, + "step": 11758 + }, + { + "epoch": 1.2913463650340435, + "grad_norm": 2.472282886505127, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7272689938545227, + "num_tokens": 299652526.0, + "step": 11759 + }, + { + "epoch": 1.291456182736657, + "grad_norm": 2.255418539047241, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7210337519645691, + "num_tokens": 299677089.0, + "step": 11760 + }, + { + "epoch": 1.2915660004392708, + "grad_norm": 2.2378060817718506, + "learning_rate": 1e-06, + "loss": 0.7985, + "mean_token_accuracy": 0.7454862594604492, + "num_tokens": 299702041.0, + "step": 11761 + }, + { + "epoch": 1.2916758181418846, + "grad_norm": 2.546700954437256, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7377493977546692, + "num_tokens": 299722359.0, + "step": 11762 + }, + { + "epoch": 1.2917856358444981, + "grad_norm": 2.3012824058532715, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7252166867256165, + "num_tokens": 299745911.0, + "step": 11763 + }, + { + "epoch": 1.2918954535471117, + "grad_norm": 2.174370050430298, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7254929542541504, + "num_tokens": 299772216.0, + "step": 11764 + }, + { + "epoch": 1.2920052712497254, + "grad_norm": 2.223806142807007, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.709458589553833, + "num_tokens": 299798919.0, + "step": 11765 + }, + { + "epoch": 1.2921150889523392, + "grad_norm": 2.2966744899749756, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.695992112159729, + "num_tokens": 299825093.0, + "step": 11766 + }, + { + "epoch": 1.2922249066549527, + "grad_norm": 2.018854856491089, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.6958984136581421, + "num_tokens": 299857220.0, + "step": 11767 + }, + { + "epoch": 1.2923347243575665, + "grad_norm": 2.5867791175842285, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.733851432800293, + "num_tokens": 299878324.0, + "step": 11768 + }, + { + "epoch": 1.29244454206018, + "grad_norm": 2.4660134315490723, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7143220901489258, + "num_tokens": 299901785.0, + "step": 11769 + }, + { + "epoch": 1.2925543597627938, + "grad_norm": 2.4980874061584473, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7299171686172485, + "num_tokens": 299923207.0, + "step": 11770 + }, + { + "epoch": 1.2926641774654075, + "grad_norm": 2.116767406463623, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7213252186775208, + "num_tokens": 299952532.0, + "step": 11771 + }, + { + "epoch": 1.292773995168021, + "grad_norm": 2.229459285736084, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7319716215133667, + "num_tokens": 299979045.0, + "step": 11772 + }, + { + "epoch": 1.2928838128706348, + "grad_norm": 2.362034797668457, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7385978102684021, + "num_tokens": 300002145.0, + "step": 11773 + }, + { + "epoch": 1.2929936305732483, + "grad_norm": 2.26932430267334, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.700707733631134, + "num_tokens": 300027326.0, + "step": 11774 + }, + { + "epoch": 1.293103448275862, + "grad_norm": 2.47348690032959, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7055177688598633, + "num_tokens": 300051027.0, + "step": 11775 + }, + { + "epoch": 1.2932132659784759, + "grad_norm": 2.224029541015625, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7179937958717346, + "num_tokens": 300077072.0, + "step": 11776 + }, + { + "epoch": 1.2933230836810894, + "grad_norm": 2.280266523361206, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7327687740325928, + "num_tokens": 300101162.0, + "step": 11777 + }, + { + "epoch": 1.293432901383703, + "grad_norm": 1.962294578552246, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7316716313362122, + "num_tokens": 300132913.0, + "step": 11778 + }, + { + "epoch": 1.2935427190863167, + "grad_norm": 2.055790424346924, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7471787929534912, + "num_tokens": 300160235.0, + "step": 11779 + }, + { + "epoch": 1.2936525367889304, + "grad_norm": 2.280653715133667, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7224994897842407, + "num_tokens": 300182377.0, + "step": 11780 + }, + { + "epoch": 1.293762354491544, + "grad_norm": 2.1761398315429688, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7338390350341797, + "num_tokens": 300206913.0, + "step": 11781 + }, + { + "epoch": 1.2938721721941577, + "grad_norm": 2.4146335124969482, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7065936923027039, + "num_tokens": 300228890.0, + "step": 11782 + }, + { + "epoch": 1.2939819898967713, + "grad_norm": 2.669360637664795, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7486486434936523, + "num_tokens": 300247733.0, + "step": 11783 + }, + { + "epoch": 1.294091807599385, + "grad_norm": 2.2796950340270996, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.718059241771698, + "num_tokens": 300273266.0, + "step": 11784 + }, + { + "epoch": 1.2942016253019988, + "grad_norm": 2.39841365814209, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7020536661148071, + "num_tokens": 300296891.0, + "step": 11785 + }, + { + "epoch": 1.2943114430046123, + "grad_norm": 2.2774605751037598, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7324085235595703, + "num_tokens": 300321376.0, + "step": 11786 + }, + { + "epoch": 1.294421260707226, + "grad_norm": 2.0772624015808105, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7106776833534241, + "num_tokens": 300350757.0, + "step": 11787 + }, + { + "epoch": 1.2945310784098396, + "grad_norm": 2.7498350143432617, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7376940250396729, + "num_tokens": 300369354.0, + "step": 11788 + }, + { + "epoch": 1.2946408961124534, + "grad_norm": 2.4822685718536377, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7404298782348633, + "num_tokens": 300391055.0, + "step": 11789 + }, + { + "epoch": 1.2947507138150671, + "grad_norm": 2.299696445465088, + "learning_rate": 1e-06, + "loss": 0.7549, + "mean_token_accuracy": 0.7630636692047119, + "num_tokens": 300413900.0, + "step": 11790 + }, + { + "epoch": 1.2948605315176807, + "grad_norm": 2.2381370067596436, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7249746322631836, + "num_tokens": 300441618.0, + "step": 11791 + }, + { + "epoch": 1.2949703492202942, + "grad_norm": 2.4301328659057617, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.7456982731819153, + "num_tokens": 300462856.0, + "step": 11792 + }, + { + "epoch": 1.295080166922908, + "grad_norm": 2.1286230087280273, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.6930220723152161, + "num_tokens": 300491190.0, + "step": 11793 + }, + { + "epoch": 1.2951899846255217, + "grad_norm": 2.350428819656372, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7501394152641296, + "num_tokens": 300512363.0, + "step": 11794 + }, + { + "epoch": 1.2952998023281352, + "grad_norm": 2.13679838180542, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7121738195419312, + "num_tokens": 300539907.0, + "step": 11795 + }, + { + "epoch": 1.295409620030749, + "grad_norm": 2.3367233276367188, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7056564092636108, + "num_tokens": 300563800.0, + "step": 11796 + }, + { + "epoch": 1.2955194377333625, + "grad_norm": 2.3996670246124268, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.721644401550293, + "num_tokens": 300586013.0, + "step": 11797 + }, + { + "epoch": 1.2956292554359763, + "grad_norm": 2.2493364810943604, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7164232134819031, + "num_tokens": 300609697.0, + "step": 11798 + }, + { + "epoch": 1.29573907313859, + "grad_norm": 2.377190589904785, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7423532605171204, + "num_tokens": 300631233.0, + "step": 11799 + }, + { + "epoch": 1.2958488908412036, + "grad_norm": 2.4206981658935547, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.7468277215957642, + "num_tokens": 300652136.0, + "step": 11800 + }, + { + "epoch": 1.2959587085438173, + "grad_norm": 2.6254026889801025, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7325196266174316, + "num_tokens": 300671080.0, + "step": 11801 + }, + { + "epoch": 1.2960685262464309, + "grad_norm": 2.2287046909332275, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7115713953971863, + "num_tokens": 300698158.0, + "step": 11802 + }, + { + "epoch": 1.2961783439490446, + "grad_norm": 2.213279962539673, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6880984306335449, + "num_tokens": 300723572.0, + "step": 11803 + }, + { + "epoch": 1.2962881616516582, + "grad_norm": 2.61287522315979, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7171403765678406, + "num_tokens": 300744031.0, + "step": 11804 + }, + { + "epoch": 1.296397979354272, + "grad_norm": 2.561009168624878, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7286566495895386, + "num_tokens": 300765819.0, + "step": 11805 + }, + { + "epoch": 1.2965077970568855, + "grad_norm": 2.1979429721832275, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7107521891593933, + "num_tokens": 300791957.0, + "step": 11806 + }, + { + "epoch": 1.2966176147594992, + "grad_norm": 1.9847135543823242, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7124646902084351, + "num_tokens": 300822649.0, + "step": 11807 + }, + { + "epoch": 1.296727432462113, + "grad_norm": 1.9687174558639526, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7149760723114014, + "num_tokens": 300854701.0, + "step": 11808 + }, + { + "epoch": 1.2968372501647265, + "grad_norm": 2.3274786472320557, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.722480058670044, + "num_tokens": 300877992.0, + "step": 11809 + }, + { + "epoch": 1.2969470678673403, + "grad_norm": 2.4832403659820557, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7285374402999878, + "num_tokens": 300897633.0, + "step": 11810 + }, + { + "epoch": 1.2970568855699538, + "grad_norm": 2.262993097305298, + "learning_rate": 1e-06, + "loss": 0.8057, + "mean_token_accuracy": 0.7434403300285339, + "num_tokens": 300920483.0, + "step": 11811 + }, + { + "epoch": 1.2971667032725676, + "grad_norm": 2.3466126918792725, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7291394472122192, + "num_tokens": 300942361.0, + "step": 11812 + }, + { + "epoch": 1.2972765209751813, + "grad_norm": 2.2003657817840576, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7235736846923828, + "num_tokens": 300967571.0, + "step": 11813 + }, + { + "epoch": 1.2973863386777948, + "grad_norm": 2.378753900527954, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7101877927780151, + "num_tokens": 300992569.0, + "step": 11814 + }, + { + "epoch": 1.2974961563804086, + "grad_norm": 2.198073148727417, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7352981567382812, + "num_tokens": 301018164.0, + "step": 11815 + }, + { + "epoch": 1.2976059740830221, + "grad_norm": 2.270432949066162, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7074008584022522, + "num_tokens": 301044370.0, + "step": 11816 + }, + { + "epoch": 1.297715791785636, + "grad_norm": 2.402413845062256, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7137333154678345, + "num_tokens": 301067116.0, + "step": 11817 + }, + { + "epoch": 1.2978256094882494, + "grad_norm": 2.40498948097229, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7145429253578186, + "num_tokens": 301089111.0, + "step": 11818 + }, + { + "epoch": 1.2979354271908632, + "grad_norm": 2.20802640914917, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7103714942932129, + "num_tokens": 301116242.0, + "step": 11819 + }, + { + "epoch": 1.2980452448934767, + "grad_norm": 2.099841594696045, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7136309742927551, + "num_tokens": 301146209.0, + "step": 11820 + }, + { + "epoch": 1.2981550625960905, + "grad_norm": 2.01175594329834, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.746515691280365, + "num_tokens": 301175112.0, + "step": 11821 + }, + { + "epoch": 1.2982648802987042, + "grad_norm": 2.2343974113464355, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.750174880027771, + "num_tokens": 301198838.0, + "step": 11822 + }, + { + "epoch": 1.2983746980013178, + "grad_norm": 2.495276689529419, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7248555421829224, + "num_tokens": 301222774.0, + "step": 11823 + }, + { + "epoch": 1.2984845157039315, + "grad_norm": 2.2781496047973633, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.739021897315979, + "num_tokens": 301246116.0, + "step": 11824 + }, + { + "epoch": 1.298594333406545, + "grad_norm": 2.285548210144043, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7073933482170105, + "num_tokens": 301272047.0, + "step": 11825 + }, + { + "epoch": 1.2987041511091588, + "grad_norm": 2.028290271759033, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7108154296875, + "num_tokens": 301302240.0, + "step": 11826 + }, + { + "epoch": 1.2988139688117726, + "grad_norm": 2.2992641925811768, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7286476492881775, + "num_tokens": 301328048.0, + "step": 11827 + }, + { + "epoch": 1.2989237865143861, + "grad_norm": 2.8448431491851807, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7326176166534424, + "num_tokens": 301345291.0, + "step": 11828 + }, + { + "epoch": 1.2990336042169996, + "grad_norm": 2.3896117210388184, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7233607172966003, + "num_tokens": 301368199.0, + "step": 11829 + }, + { + "epoch": 1.2991434219196134, + "grad_norm": 2.1640853881835938, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7019422650337219, + "num_tokens": 301398226.0, + "step": 11830 + }, + { + "epoch": 1.2992532396222272, + "grad_norm": 2.2232866287231445, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7312149405479431, + "num_tokens": 301424376.0, + "step": 11831 + }, + { + "epoch": 1.2993630573248407, + "grad_norm": 2.50362491607666, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7048677206039429, + "num_tokens": 301446124.0, + "step": 11832 + }, + { + "epoch": 1.2994728750274545, + "grad_norm": 2.0995094776153564, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7052854299545288, + "num_tokens": 301476447.0, + "step": 11833 + }, + { + "epoch": 1.299582692730068, + "grad_norm": 2.200096368789673, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7292580008506775, + "num_tokens": 301502373.0, + "step": 11834 + }, + { + "epoch": 1.2996925104326817, + "grad_norm": 2.311004161834717, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7031584978103638, + "num_tokens": 301529627.0, + "step": 11835 + }, + { + "epoch": 1.2998023281352955, + "grad_norm": 2.766414165496826, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7421138882637024, + "num_tokens": 301546594.0, + "step": 11836 + }, + { + "epoch": 1.299912145837909, + "grad_norm": 2.448949098587036, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7135049104690552, + "num_tokens": 301568414.0, + "step": 11837 + }, + { + "epoch": 1.3000219635405228, + "grad_norm": 2.5775952339172363, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7273098230361938, + "num_tokens": 301589302.0, + "step": 11838 + }, + { + "epoch": 1.3001317812431363, + "grad_norm": 2.103550910949707, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7078707814216614, + "num_tokens": 301617623.0, + "step": 11839 + }, + { + "epoch": 1.30024159894575, + "grad_norm": 2.1291542053222656, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7041827440261841, + "num_tokens": 301644238.0, + "step": 11840 + }, + { + "epoch": 1.3003514166483638, + "grad_norm": 2.228581428527832, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7101857662200928, + "num_tokens": 301671204.0, + "step": 11841 + }, + { + "epoch": 1.3004612343509774, + "grad_norm": 2.2198374271392822, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7366154193878174, + "num_tokens": 301697483.0, + "step": 11842 + }, + { + "epoch": 1.300571052053591, + "grad_norm": 2.3049020767211914, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7103844881057739, + "num_tokens": 301722677.0, + "step": 11843 + }, + { + "epoch": 1.3006808697562047, + "grad_norm": 2.3326568603515625, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7330144643783569, + "num_tokens": 301746493.0, + "step": 11844 + }, + { + "epoch": 1.3007906874588184, + "grad_norm": 2.2892885208129883, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7298805713653564, + "num_tokens": 301771979.0, + "step": 11845 + }, + { + "epoch": 1.300900505161432, + "grad_norm": 2.6655433177948, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7251904010772705, + "num_tokens": 301792214.0, + "step": 11846 + }, + { + "epoch": 1.3010103228640457, + "grad_norm": 2.0120255947113037, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7062779664993286, + "num_tokens": 301823873.0, + "step": 11847 + }, + { + "epoch": 1.3011201405666593, + "grad_norm": 2.3068971633911133, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7314490675926208, + "num_tokens": 301847882.0, + "step": 11848 + }, + { + "epoch": 1.301229958269273, + "grad_norm": 2.4605162143707275, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7169467806816101, + "num_tokens": 301871203.0, + "step": 11849 + }, + { + "epoch": 1.3013397759718868, + "grad_norm": 2.099860191345215, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7184539437294006, + "num_tokens": 301900329.0, + "step": 11850 + }, + { + "epoch": 1.3014495936745003, + "grad_norm": 2.1958248615264893, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7200618386268616, + "num_tokens": 301925723.0, + "step": 11851 + }, + { + "epoch": 1.301559411377114, + "grad_norm": 2.428807020187378, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.7080302238464355, + "num_tokens": 301947647.0, + "step": 11852 + }, + { + "epoch": 1.3016692290797276, + "grad_norm": 2.126004219055176, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7323935031890869, + "num_tokens": 301975485.0, + "step": 11853 + }, + { + "epoch": 1.3017790467823414, + "grad_norm": 2.3724231719970703, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7090169191360474, + "num_tokens": 301999757.0, + "step": 11854 + }, + { + "epoch": 1.301888864484955, + "grad_norm": 2.3478643894195557, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7298866510391235, + "num_tokens": 302023209.0, + "step": 11855 + }, + { + "epoch": 1.3019986821875686, + "grad_norm": 2.5423572063446045, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7417844533920288, + "num_tokens": 302043186.0, + "step": 11856 + }, + { + "epoch": 1.3021084998901822, + "grad_norm": 2.1793739795684814, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7126481533050537, + "num_tokens": 302070415.0, + "step": 11857 + }, + { + "epoch": 1.302218317592796, + "grad_norm": 2.7007296085357666, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.724858283996582, + "num_tokens": 302089069.0, + "step": 11858 + }, + { + "epoch": 1.3023281352954097, + "grad_norm": 2.2290608882904053, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7171335220336914, + "num_tokens": 302114040.0, + "step": 11859 + }, + { + "epoch": 1.3024379529980232, + "grad_norm": 2.4617204666137695, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7268499135971069, + "num_tokens": 302135525.0, + "step": 11860 + }, + { + "epoch": 1.302547770700637, + "grad_norm": 2.579827308654785, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.7551188468933105, + "num_tokens": 302154036.0, + "step": 11861 + }, + { + "epoch": 1.3026575884032505, + "grad_norm": 2.2819530963897705, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.6865267753601074, + "num_tokens": 302181578.0, + "step": 11862 + }, + { + "epoch": 1.3027674061058643, + "grad_norm": 2.348318099975586, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.719975471496582, + "num_tokens": 302205993.0, + "step": 11863 + }, + { + "epoch": 1.302877223808478, + "grad_norm": 2.209984540939331, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7121396064758301, + "num_tokens": 302233251.0, + "step": 11864 + }, + { + "epoch": 1.3029870415110916, + "grad_norm": 2.3443002700805664, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7179415225982666, + "num_tokens": 302256893.0, + "step": 11865 + }, + { + "epoch": 1.3030968592137053, + "grad_norm": 2.056427001953125, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7460910677909851, + "num_tokens": 302284614.0, + "step": 11866 + }, + { + "epoch": 1.3032066769163189, + "grad_norm": 2.1343836784362793, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7216606140136719, + "num_tokens": 302314351.0, + "step": 11867 + }, + { + "epoch": 1.3033164946189326, + "grad_norm": 2.3464648723602295, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7319422960281372, + "num_tokens": 302336201.0, + "step": 11868 + }, + { + "epoch": 1.3034263123215462, + "grad_norm": 2.421147108078003, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7375619411468506, + "num_tokens": 302359292.0, + "step": 11869 + }, + { + "epoch": 1.30353613002416, + "grad_norm": 2.260532855987549, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7247800230979919, + "num_tokens": 302383541.0, + "step": 11870 + }, + { + "epoch": 1.3036459477267734, + "grad_norm": 1.9916393756866455, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.720966100692749, + "num_tokens": 302413571.0, + "step": 11871 + }, + { + "epoch": 1.3037557654293872, + "grad_norm": 2.4413161277770996, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7232450246810913, + "num_tokens": 302441066.0, + "step": 11872 + }, + { + "epoch": 1.303865583132001, + "grad_norm": 2.110562324523926, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7245541214942932, + "num_tokens": 302470459.0, + "step": 11873 + }, + { + "epoch": 1.3039754008346145, + "grad_norm": 2.1780142784118652, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7290486693382263, + "num_tokens": 302496566.0, + "step": 11874 + }, + { + "epoch": 1.3040852185372283, + "grad_norm": 2.159809112548828, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7325711846351624, + "num_tokens": 302522712.0, + "step": 11875 + }, + { + "epoch": 1.3041950362398418, + "grad_norm": 2.1391990184783936, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7009813189506531, + "num_tokens": 302551134.0, + "step": 11876 + }, + { + "epoch": 1.3043048539424555, + "grad_norm": 2.2276878356933594, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7191851139068604, + "num_tokens": 302577731.0, + "step": 11877 + }, + { + "epoch": 1.3044146716450693, + "grad_norm": 2.4380908012390137, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7264587879180908, + "num_tokens": 302600934.0, + "step": 11878 + }, + { + "epoch": 1.3045244893476828, + "grad_norm": 2.5641515254974365, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7170546054840088, + "num_tokens": 302622469.0, + "step": 11879 + }, + { + "epoch": 1.3046343070502964, + "grad_norm": 2.1562132835388184, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7027651071548462, + "num_tokens": 302651876.0, + "step": 11880 + }, + { + "epoch": 1.3047441247529101, + "grad_norm": 2.3462064266204834, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7362383604049683, + "num_tokens": 302676435.0, + "step": 11881 + }, + { + "epoch": 1.3048539424555239, + "grad_norm": 2.3522109985351562, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7107066512107849, + "num_tokens": 302699229.0, + "step": 11882 + }, + { + "epoch": 1.3049637601581374, + "grad_norm": 2.489659070968628, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7401347160339355, + "num_tokens": 302719802.0, + "step": 11883 + }, + { + "epoch": 1.3050735778607512, + "grad_norm": 2.3740086555480957, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7244484424591064, + "num_tokens": 302743100.0, + "step": 11884 + }, + { + "epoch": 1.3051833955633647, + "grad_norm": 2.335071086883545, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7103525400161743, + "num_tokens": 302766434.0, + "step": 11885 + }, + { + "epoch": 1.3052932132659785, + "grad_norm": 2.412297248840332, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7356922626495361, + "num_tokens": 302789160.0, + "step": 11886 + }, + { + "epoch": 1.3054030309685922, + "grad_norm": 2.129149913787842, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7015026211738586, + "num_tokens": 302818472.0, + "step": 11887 + }, + { + "epoch": 1.3055128486712058, + "grad_norm": 2.243788480758667, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7273156642913818, + "num_tokens": 302844520.0, + "step": 11888 + }, + { + "epoch": 1.3056226663738195, + "grad_norm": 1.9860796928405762, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7254523038864136, + "num_tokens": 302875071.0, + "step": 11889 + }, + { + "epoch": 1.305732484076433, + "grad_norm": 1.9419785737991333, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.7042153477668762, + "num_tokens": 302908271.0, + "step": 11890 + }, + { + "epoch": 1.3058423017790468, + "grad_norm": 2.7564537525177, + "learning_rate": 1e-06, + "loss": 0.7852, + "mean_token_accuracy": 0.7478258609771729, + "num_tokens": 302927075.0, + "step": 11891 + }, + { + "epoch": 1.3059521194816606, + "grad_norm": 2.1431586742401123, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7456697225570679, + "num_tokens": 302954195.0, + "step": 11892 + }, + { + "epoch": 1.306061937184274, + "grad_norm": 2.1990303993225098, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7188143730163574, + "num_tokens": 302981814.0, + "step": 11893 + }, + { + "epoch": 1.3061717548868876, + "grad_norm": 2.0393929481506348, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7297619581222534, + "num_tokens": 303012081.0, + "step": 11894 + }, + { + "epoch": 1.3062815725895014, + "grad_norm": 2.1864681243896484, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7283107042312622, + "num_tokens": 303038873.0, + "step": 11895 + }, + { + "epoch": 1.3063913902921152, + "grad_norm": 2.4522180557250977, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7236495018005371, + "num_tokens": 303059193.0, + "step": 11896 + }, + { + "epoch": 1.3065012079947287, + "grad_norm": 2.5082294940948486, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7391023635864258, + "num_tokens": 303081078.0, + "step": 11897 + }, + { + "epoch": 1.3066110256973424, + "grad_norm": 2.3986709117889404, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7354673147201538, + "num_tokens": 303107432.0, + "step": 11898 + }, + { + "epoch": 1.306720843399956, + "grad_norm": 2.265255928039551, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.6983996033668518, + "num_tokens": 303131431.0, + "step": 11899 + }, + { + "epoch": 1.3068306611025697, + "grad_norm": 2.1180529594421387, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6949178576469421, + "num_tokens": 303159680.0, + "step": 11900 + }, + { + "epoch": 1.3069404788051835, + "grad_norm": 2.1598875522613525, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7154725193977356, + "num_tokens": 303185172.0, + "step": 11901 + }, + { + "epoch": 1.307050296507797, + "grad_norm": 2.022345542907715, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7080589532852173, + "num_tokens": 303215771.0, + "step": 11902 + }, + { + "epoch": 1.3071601142104108, + "grad_norm": 2.2992794513702393, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7250692844390869, + "num_tokens": 303240670.0, + "step": 11903 + }, + { + "epoch": 1.3072699319130243, + "grad_norm": 2.3526611328125, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7246078848838806, + "num_tokens": 303264328.0, + "step": 11904 + }, + { + "epoch": 1.307379749615638, + "grad_norm": 2.2015128135681152, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7029601335525513, + "num_tokens": 303294321.0, + "step": 11905 + }, + { + "epoch": 1.3074895673182518, + "grad_norm": 2.435556650161743, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7256631851196289, + "num_tokens": 303314636.0, + "step": 11906 + }, + { + "epoch": 1.3075993850208654, + "grad_norm": 2.0594725608825684, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.6991223096847534, + "num_tokens": 303345078.0, + "step": 11907 + }, + { + "epoch": 1.307709202723479, + "grad_norm": 1.9917802810668945, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7071882486343384, + "num_tokens": 303374972.0, + "step": 11908 + }, + { + "epoch": 1.3078190204260927, + "grad_norm": 2.1529107093811035, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7076322436332703, + "num_tokens": 303403292.0, + "step": 11909 + }, + { + "epoch": 1.3079288381287064, + "grad_norm": 2.260012626647949, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7063333988189697, + "num_tokens": 303429547.0, + "step": 11910 + }, + { + "epoch": 1.30803865583132, + "grad_norm": 2.475040912628174, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7289001941680908, + "num_tokens": 303451773.0, + "step": 11911 + }, + { + "epoch": 1.3081484735339337, + "grad_norm": 2.6658401489257812, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7383391261100769, + "num_tokens": 303471696.0, + "step": 11912 + }, + { + "epoch": 1.3082582912365472, + "grad_norm": 2.1717569828033447, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7072780132293701, + "num_tokens": 303500331.0, + "step": 11913 + }, + { + "epoch": 1.308368108939161, + "grad_norm": 2.526026725769043, + "learning_rate": 1e-06, + "loss": 0.7672, + "mean_token_accuracy": 0.7503440380096436, + "num_tokens": 303519371.0, + "step": 11914 + }, + { + "epoch": 1.3084779266417748, + "grad_norm": 1.9767440557479858, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7096755504608154, + "num_tokens": 303550572.0, + "step": 11915 + }, + { + "epoch": 1.3085877443443883, + "grad_norm": 2.3172860145568848, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7053365707397461, + "num_tokens": 303575148.0, + "step": 11916 + }, + { + "epoch": 1.308697562047002, + "grad_norm": 2.364973783493042, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7476233839988708, + "num_tokens": 303598520.0, + "step": 11917 + }, + { + "epoch": 1.3088073797496156, + "grad_norm": 1.962590217590332, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.6976132392883301, + "num_tokens": 303630942.0, + "step": 11918 + }, + { + "epoch": 1.3089171974522293, + "grad_norm": 2.298553466796875, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7235890626907349, + "num_tokens": 303656280.0, + "step": 11919 + }, + { + "epoch": 1.309027015154843, + "grad_norm": 2.2820072174072266, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7362378835678101, + "num_tokens": 303681998.0, + "step": 11920 + }, + { + "epoch": 1.3091368328574566, + "grad_norm": 2.1063714027404785, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7068262100219727, + "num_tokens": 303709852.0, + "step": 11921 + }, + { + "epoch": 1.3092466505600702, + "grad_norm": 2.33373761177063, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7237316370010376, + "num_tokens": 303735864.0, + "step": 11922 + }, + { + "epoch": 1.309356468262684, + "grad_norm": 2.4879887104034424, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7309073209762573, + "num_tokens": 303757696.0, + "step": 11923 + }, + { + "epoch": 1.3094662859652977, + "grad_norm": 2.225126266479492, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7477065324783325, + "num_tokens": 303783439.0, + "step": 11924 + }, + { + "epoch": 1.3095761036679112, + "grad_norm": 2.41508412361145, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7293306589126587, + "num_tokens": 303807140.0, + "step": 11925 + }, + { + "epoch": 1.309685921370525, + "grad_norm": 2.207233428955078, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7112181782722473, + "num_tokens": 303832054.0, + "step": 11926 + }, + { + "epoch": 1.3097957390731385, + "grad_norm": 2.4548959732055664, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7364003658294678, + "num_tokens": 303854564.0, + "step": 11927 + }, + { + "epoch": 1.3099055567757523, + "grad_norm": 2.159893751144409, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7277286052703857, + "num_tokens": 303882041.0, + "step": 11928 + }, + { + "epoch": 1.310015374478366, + "grad_norm": 1.9650205373764038, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.6880312561988831, + "num_tokens": 303914957.0, + "step": 11929 + }, + { + "epoch": 1.3101251921809796, + "grad_norm": 2.160989284515381, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.6990352869033813, + "num_tokens": 303941109.0, + "step": 11930 + }, + { + "epoch": 1.3102350098835933, + "grad_norm": 2.1429200172424316, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7160866856575012, + "num_tokens": 303972058.0, + "step": 11931 + }, + { + "epoch": 1.3103448275862069, + "grad_norm": 2.0087568759918213, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7248401641845703, + "num_tokens": 304002501.0, + "step": 11932 + }, + { + "epoch": 1.3104546452888206, + "grad_norm": 2.1175098419189453, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7086893916130066, + "num_tokens": 304031458.0, + "step": 11933 + }, + { + "epoch": 1.3105644629914341, + "grad_norm": 2.07310152053833, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7075074315071106, + "num_tokens": 304059936.0, + "step": 11934 + }, + { + "epoch": 1.310674280694048, + "grad_norm": 2.1969916820526123, + "learning_rate": 1e-06, + "loss": 1.0451, + "mean_token_accuracy": 0.6883558630943298, + "num_tokens": 304090019.0, + "step": 11935 + }, + { + "epoch": 1.3107840983966614, + "grad_norm": 2.3143763542175293, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7117564082145691, + "num_tokens": 304116451.0, + "step": 11936 + }, + { + "epoch": 1.3108939160992752, + "grad_norm": 2.357940435409546, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.6994990706443787, + "num_tokens": 304140575.0, + "step": 11937 + }, + { + "epoch": 1.311003733801889, + "grad_norm": 2.2096261978149414, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.72036212682724, + "num_tokens": 304166100.0, + "step": 11938 + }, + { + "epoch": 1.3111135515045025, + "grad_norm": 2.449881076812744, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7291665077209473, + "num_tokens": 304189624.0, + "step": 11939 + }, + { + "epoch": 1.3112233692071162, + "grad_norm": 2.234679937362671, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.737783670425415, + "num_tokens": 304215866.0, + "step": 11940 + }, + { + "epoch": 1.3113331869097298, + "grad_norm": 2.0935659408569336, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6954135894775391, + "num_tokens": 304244567.0, + "step": 11941 + }, + { + "epoch": 1.3114430046123435, + "grad_norm": 2.3108928203582764, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7072877883911133, + "num_tokens": 304270248.0, + "step": 11942 + }, + { + "epoch": 1.3115528223149573, + "grad_norm": 2.466930627822876, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7169155478477478, + "num_tokens": 304293637.0, + "step": 11943 + }, + { + "epoch": 1.3116626400175708, + "grad_norm": 2.223137378692627, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.6981298327445984, + "num_tokens": 304319697.0, + "step": 11944 + }, + { + "epoch": 1.3117724577201844, + "grad_norm": 2.2780871391296387, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.701744556427002, + "num_tokens": 304344328.0, + "step": 11945 + }, + { + "epoch": 1.3118822754227981, + "grad_norm": 2.274141311645508, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7008313536643982, + "num_tokens": 304369890.0, + "step": 11946 + }, + { + "epoch": 1.3119920931254119, + "grad_norm": 2.102398633956909, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7133875489234924, + "num_tokens": 304396002.0, + "step": 11947 + }, + { + "epoch": 1.3121019108280254, + "grad_norm": 2.068950653076172, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7053440809249878, + "num_tokens": 304423809.0, + "step": 11948 + }, + { + "epoch": 1.3122117285306392, + "grad_norm": 2.2007150650024414, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6956872344017029, + "num_tokens": 304449373.0, + "step": 11949 + }, + { + "epoch": 1.3123215462332527, + "grad_norm": 2.214489698410034, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7031944990158081, + "num_tokens": 304476871.0, + "step": 11950 + }, + { + "epoch": 1.3124313639358665, + "grad_norm": 2.250972270965576, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7268498539924622, + "num_tokens": 304500934.0, + "step": 11951 + }, + { + "epoch": 1.3125411816384802, + "grad_norm": 2.295116662979126, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7102068662643433, + "num_tokens": 304524594.0, + "step": 11952 + }, + { + "epoch": 1.3126509993410937, + "grad_norm": 2.1575777530670166, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.708450198173523, + "num_tokens": 304549999.0, + "step": 11953 + }, + { + "epoch": 1.3127608170437075, + "grad_norm": 2.2700142860412598, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.6987402439117432, + "num_tokens": 304577937.0, + "step": 11954 + }, + { + "epoch": 1.312870634746321, + "grad_norm": 2.315039873123169, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7200236320495605, + "num_tokens": 304602084.0, + "step": 11955 + }, + { + "epoch": 1.3129804524489348, + "grad_norm": 2.6669797897338867, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7290102243423462, + "num_tokens": 304621056.0, + "step": 11956 + }, + { + "epoch": 1.3130902701515486, + "grad_norm": 2.452089786529541, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7278603315353394, + "num_tokens": 304642936.0, + "step": 11957 + }, + { + "epoch": 1.313200087854162, + "grad_norm": 2.3885257244110107, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.731726884841919, + "num_tokens": 304664164.0, + "step": 11958 + }, + { + "epoch": 1.3133099055567756, + "grad_norm": 2.1336822509765625, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.6976144313812256, + "num_tokens": 304693981.0, + "step": 11959 + }, + { + "epoch": 1.3134197232593894, + "grad_norm": 2.580374240875244, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7206035852432251, + "num_tokens": 304714229.0, + "step": 11960 + }, + { + "epoch": 1.3135295409620031, + "grad_norm": 2.075756549835205, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7058730721473694, + "num_tokens": 304744183.0, + "step": 11961 + }, + { + "epoch": 1.3136393586646167, + "grad_norm": 2.1281704902648926, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7113811373710632, + "num_tokens": 304775824.0, + "step": 11962 + }, + { + "epoch": 1.3137491763672304, + "grad_norm": 2.36686372756958, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.720219075679779, + "num_tokens": 304800381.0, + "step": 11963 + }, + { + "epoch": 1.313858994069844, + "grad_norm": 2.3386263847351074, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7496598958969116, + "num_tokens": 304823433.0, + "step": 11964 + }, + { + "epoch": 1.3139688117724577, + "grad_norm": 2.325127363204956, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.6986567974090576, + "num_tokens": 304849264.0, + "step": 11965 + }, + { + "epoch": 1.3140786294750715, + "grad_norm": 2.0838398933410645, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.703160285949707, + "num_tokens": 304877170.0, + "step": 11966 + }, + { + "epoch": 1.314188447177685, + "grad_norm": 2.2079412937164307, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7342992424964905, + "num_tokens": 304902780.0, + "step": 11967 + }, + { + "epoch": 1.3142982648802988, + "grad_norm": 2.0748512744903564, + "learning_rate": 1e-06, + "loss": 1.0441, + "mean_token_accuracy": 0.6868119239807129, + "num_tokens": 304932612.0, + "step": 11968 + }, + { + "epoch": 1.3144080825829123, + "grad_norm": 2.468637466430664, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7322489023208618, + "num_tokens": 304953859.0, + "step": 11969 + }, + { + "epoch": 1.314517900285526, + "grad_norm": 2.1046645641326904, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7130327224731445, + "num_tokens": 304982899.0, + "step": 11970 + }, + { + "epoch": 1.3146277179881398, + "grad_norm": 2.4357640743255615, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7087527513504028, + "num_tokens": 305005597.0, + "step": 11971 + }, + { + "epoch": 1.3147375356907534, + "grad_norm": 2.442824602127075, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7272107601165771, + "num_tokens": 305027150.0, + "step": 11972 + }, + { + "epoch": 1.314847353393367, + "grad_norm": 2.2118008136749268, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.6958125829696655, + "num_tokens": 305055415.0, + "step": 11973 + }, + { + "epoch": 1.3149571710959806, + "grad_norm": 2.5328447818756104, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7146134376525879, + "num_tokens": 305076538.0, + "step": 11974 + }, + { + "epoch": 1.3150669887985944, + "grad_norm": 2.2383718490600586, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7139621376991272, + "num_tokens": 305100851.0, + "step": 11975 + }, + { + "epoch": 1.315176806501208, + "grad_norm": 2.1620748043060303, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7089231014251709, + "num_tokens": 305128045.0, + "step": 11976 + }, + { + "epoch": 1.3152866242038217, + "grad_norm": 2.2345616817474365, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7291044592857361, + "num_tokens": 305151765.0, + "step": 11977 + }, + { + "epoch": 1.3153964419064352, + "grad_norm": 2.0122556686401367, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7018682360649109, + "num_tokens": 305183085.0, + "step": 11978 + }, + { + "epoch": 1.315506259609049, + "grad_norm": 2.466187000274658, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7145060300827026, + "num_tokens": 305205817.0, + "step": 11979 + }, + { + "epoch": 1.3156160773116627, + "grad_norm": 2.1544148921966553, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7129124402999878, + "num_tokens": 305230968.0, + "step": 11980 + }, + { + "epoch": 1.3157258950142763, + "grad_norm": 1.8703088760375977, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7291549444198608, + "num_tokens": 305264829.0, + "step": 11981 + }, + { + "epoch": 1.31583571271689, + "grad_norm": 2.5550901889801025, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7462204694747925, + "num_tokens": 305284653.0, + "step": 11982 + }, + { + "epoch": 1.3159455304195036, + "grad_norm": 2.2586545944213867, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7252010107040405, + "num_tokens": 305311056.0, + "step": 11983 + }, + { + "epoch": 1.3160553481221173, + "grad_norm": 2.2878689765930176, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7297924160957336, + "num_tokens": 305337064.0, + "step": 11984 + }, + { + "epoch": 1.3161651658247309, + "grad_norm": 2.084745407104492, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7255645990371704, + "num_tokens": 305364857.0, + "step": 11985 + }, + { + "epoch": 1.3162749835273446, + "grad_norm": 2.203885793685913, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7092939019203186, + "num_tokens": 305393205.0, + "step": 11986 + }, + { + "epoch": 1.3163848012299582, + "grad_norm": 2.4459433555603027, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7377872467041016, + "num_tokens": 305414470.0, + "step": 11987 + }, + { + "epoch": 1.316494618932572, + "grad_norm": 2.383542060852051, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7069082260131836, + "num_tokens": 305436835.0, + "step": 11988 + }, + { + "epoch": 1.3166044366351857, + "grad_norm": 2.1649281978607178, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7424423694610596, + "num_tokens": 305465302.0, + "step": 11989 + }, + { + "epoch": 1.3167142543377992, + "grad_norm": 1.979987382888794, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.700482964515686, + "num_tokens": 305496014.0, + "step": 11990 + }, + { + "epoch": 1.316824072040413, + "grad_norm": 2.370779514312744, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7365586757659912, + "num_tokens": 305519858.0, + "step": 11991 + }, + { + "epoch": 1.3169338897430265, + "grad_norm": 2.2766716480255127, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7046812772750854, + "num_tokens": 305546219.0, + "step": 11992 + }, + { + "epoch": 1.3170437074456403, + "grad_norm": 2.1585581302642822, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7103971838951111, + "num_tokens": 305575212.0, + "step": 11993 + }, + { + "epoch": 1.317153525148254, + "grad_norm": 2.413257122039795, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7327591776847839, + "num_tokens": 305598780.0, + "step": 11994 + }, + { + "epoch": 1.3172633428508675, + "grad_norm": 2.0951709747314453, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7253844141960144, + "num_tokens": 305626818.0, + "step": 11995 + }, + { + "epoch": 1.3173731605534813, + "grad_norm": 2.4335925579071045, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7223814725875854, + "num_tokens": 305649233.0, + "step": 11996 + }, + { + "epoch": 1.3174829782560948, + "grad_norm": 2.1829421520233154, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7511202096939087, + "num_tokens": 305675314.0, + "step": 11997 + }, + { + "epoch": 1.3175927959587086, + "grad_norm": 2.159221649169922, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6903413534164429, + "num_tokens": 305703961.0, + "step": 11998 + }, + { + "epoch": 1.3177026136613221, + "grad_norm": 2.130140781402588, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7069498300552368, + "num_tokens": 305732387.0, + "step": 11999 + }, + { + "epoch": 1.3178124313639359, + "grad_norm": 2.2258248329162598, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7115557193756104, + "num_tokens": 305759831.0, + "step": 12000 + }, + { + "epoch": 1.3179222490665494, + "grad_norm": 2.587946653366089, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7289241552352905, + "num_tokens": 305778972.0, + "step": 12001 + }, + { + "epoch": 1.3180320667691632, + "grad_norm": 2.312607526779175, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7122613787651062, + "num_tokens": 305804856.0, + "step": 12002 + }, + { + "epoch": 1.318141884471777, + "grad_norm": 2.180629253387451, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7218196988105774, + "num_tokens": 305829202.0, + "step": 12003 + }, + { + "epoch": 1.3182517021743905, + "grad_norm": 2.265167713165283, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7012983560562134, + "num_tokens": 305854626.0, + "step": 12004 + }, + { + "epoch": 1.3183615198770042, + "grad_norm": 2.335446834564209, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7258182764053345, + "num_tokens": 305879599.0, + "step": 12005 + }, + { + "epoch": 1.3184713375796178, + "grad_norm": 2.37243390083313, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7083029747009277, + "num_tokens": 305904590.0, + "step": 12006 + }, + { + "epoch": 1.3185811552822315, + "grad_norm": 2.2583725452423096, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.703598141670227, + "num_tokens": 305931985.0, + "step": 12007 + }, + { + "epoch": 1.3186909729848453, + "grad_norm": 2.3212480545043945, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7315776348114014, + "num_tokens": 305956083.0, + "step": 12008 + }, + { + "epoch": 1.3188007906874588, + "grad_norm": 2.353520393371582, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.729225754737854, + "num_tokens": 305980111.0, + "step": 12009 + }, + { + "epoch": 1.3189106083900723, + "grad_norm": 2.379962205886841, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7258490920066833, + "num_tokens": 306004238.0, + "step": 12010 + }, + { + "epoch": 1.319020426092686, + "grad_norm": 2.527355432510376, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7398165464401245, + "num_tokens": 306024222.0, + "step": 12011 + }, + { + "epoch": 1.3191302437952999, + "grad_norm": 2.4989590644836426, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.722484827041626, + "num_tokens": 306047423.0, + "step": 12012 + }, + { + "epoch": 1.3192400614979134, + "grad_norm": 2.0977425575256348, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7162325382232666, + "num_tokens": 306074145.0, + "step": 12013 + }, + { + "epoch": 1.3193498792005272, + "grad_norm": 2.250681161880493, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7204217910766602, + "num_tokens": 306100423.0, + "step": 12014 + }, + { + "epoch": 1.3194596969031407, + "grad_norm": 2.468019723892212, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.735758364200592, + "num_tokens": 306122132.0, + "step": 12015 + }, + { + "epoch": 1.3195695146057544, + "grad_norm": 2.495509386062622, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.735145092010498, + "num_tokens": 306143960.0, + "step": 12016 + }, + { + "epoch": 1.3196793323083682, + "grad_norm": 2.394219398498535, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7345492839813232, + "num_tokens": 306166662.0, + "step": 12017 + }, + { + "epoch": 1.3197891500109817, + "grad_norm": 2.3738503456115723, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.726249635219574, + "num_tokens": 306189720.0, + "step": 12018 + }, + { + "epoch": 1.3198989677135955, + "grad_norm": 2.6573657989501953, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7363178730010986, + "num_tokens": 306209820.0, + "step": 12019 + }, + { + "epoch": 1.320008785416209, + "grad_norm": 2.0703094005584717, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.73973548412323, + "num_tokens": 306237803.0, + "step": 12020 + }, + { + "epoch": 1.3201186031188228, + "grad_norm": 2.2269246578216553, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7272925972938538, + "num_tokens": 306263879.0, + "step": 12021 + }, + { + "epoch": 1.3202284208214365, + "grad_norm": 2.1854071617126465, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7340008616447449, + "num_tokens": 306290423.0, + "step": 12022 + }, + { + "epoch": 1.32033823852405, + "grad_norm": 2.496015787124634, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7322132587432861, + "num_tokens": 306312156.0, + "step": 12023 + }, + { + "epoch": 1.3204480562266636, + "grad_norm": 2.2620339393615723, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.706300675868988, + "num_tokens": 306338891.0, + "step": 12024 + }, + { + "epoch": 1.3205578739292774, + "grad_norm": 2.193861246109009, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.7033189535140991, + "num_tokens": 306366568.0, + "step": 12025 + }, + { + "epoch": 1.3206676916318911, + "grad_norm": 2.1585533618927, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7270358800888062, + "num_tokens": 306393183.0, + "step": 12026 + }, + { + "epoch": 1.3207775093345047, + "grad_norm": 2.1250600814819336, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7060103416442871, + "num_tokens": 306421827.0, + "step": 12027 + }, + { + "epoch": 1.3208873270371184, + "grad_norm": 2.1865522861480713, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7332490682601929, + "num_tokens": 306448456.0, + "step": 12028 + }, + { + "epoch": 1.320997144739732, + "grad_norm": 2.4887728691101074, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7396776080131531, + "num_tokens": 306470017.0, + "step": 12029 + }, + { + "epoch": 1.3211069624423457, + "grad_norm": 2.479724645614624, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7096202373504639, + "num_tokens": 306492394.0, + "step": 12030 + }, + { + "epoch": 1.3212167801449595, + "grad_norm": 2.184703826904297, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7204763293266296, + "num_tokens": 306518577.0, + "step": 12031 + }, + { + "epoch": 1.321326597847573, + "grad_norm": 2.5635814666748047, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7232338190078735, + "num_tokens": 306539188.0, + "step": 12032 + }, + { + "epoch": 1.3214364155501868, + "grad_norm": 2.2296175956726074, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7338553667068481, + "num_tokens": 306564006.0, + "step": 12033 + }, + { + "epoch": 1.3215462332528003, + "grad_norm": 2.129937171936035, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7125861048698425, + "num_tokens": 306591366.0, + "step": 12034 + }, + { + "epoch": 1.321656050955414, + "grad_norm": 2.3724591732025146, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7357502579689026, + "num_tokens": 306615673.0, + "step": 12035 + }, + { + "epoch": 1.3217658686580278, + "grad_norm": 2.16103458404541, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7139222621917725, + "num_tokens": 306643319.0, + "step": 12036 + }, + { + "epoch": 1.3218756863606413, + "grad_norm": 2.2946579456329346, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7164061069488525, + "num_tokens": 306667388.0, + "step": 12037 + }, + { + "epoch": 1.3219855040632549, + "grad_norm": 2.2597758769989014, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7375196218490601, + "num_tokens": 306692601.0, + "step": 12038 + }, + { + "epoch": 1.3220953217658686, + "grad_norm": 2.142991781234741, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7060307860374451, + "num_tokens": 306721796.0, + "step": 12039 + }, + { + "epoch": 1.3222051394684824, + "grad_norm": 2.2644455432891846, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7099372148513794, + "num_tokens": 306745918.0, + "step": 12040 + }, + { + "epoch": 1.322314957171096, + "grad_norm": 2.0786142349243164, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7082699537277222, + "num_tokens": 306775744.0, + "step": 12041 + }, + { + "epoch": 1.3224247748737097, + "grad_norm": 2.034247636795044, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.704286515712738, + "num_tokens": 306805447.0, + "step": 12042 + }, + { + "epoch": 1.3225345925763232, + "grad_norm": 2.3062920570373535, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7207112908363342, + "num_tokens": 306830192.0, + "step": 12043 + }, + { + "epoch": 1.322644410278937, + "grad_norm": 2.411764144897461, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7331726551055908, + "num_tokens": 306852805.0, + "step": 12044 + }, + { + "epoch": 1.3227542279815507, + "grad_norm": 2.4848294258117676, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.73381108045578, + "num_tokens": 306873366.0, + "step": 12045 + }, + { + "epoch": 1.3228640456841643, + "grad_norm": 2.189797878265381, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.714798092842102, + "num_tokens": 306901372.0, + "step": 12046 + }, + { + "epoch": 1.322973863386778, + "grad_norm": 2.296570062637329, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6971133947372437, + "num_tokens": 306929141.0, + "step": 12047 + }, + { + "epoch": 1.3230836810893916, + "grad_norm": 2.271113395690918, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7345073223114014, + "num_tokens": 306953577.0, + "step": 12048 + }, + { + "epoch": 1.3231934987920053, + "grad_norm": 2.1917741298675537, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7191178798675537, + "num_tokens": 306980787.0, + "step": 12049 + }, + { + "epoch": 1.3233033164946189, + "grad_norm": 1.977135419845581, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7362859845161438, + "num_tokens": 307010681.0, + "step": 12050 + }, + { + "epoch": 1.3234131341972326, + "grad_norm": 2.2381629943847656, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7235857248306274, + "num_tokens": 307037326.0, + "step": 12051 + }, + { + "epoch": 1.3235229518998461, + "grad_norm": 2.2593576908111572, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7322965860366821, + "num_tokens": 307062075.0, + "step": 12052 + }, + { + "epoch": 1.32363276960246, + "grad_norm": 2.080613136291504, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7232562303543091, + "num_tokens": 307091938.0, + "step": 12053 + }, + { + "epoch": 1.3237425873050737, + "grad_norm": 2.106487274169922, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7186278700828552, + "num_tokens": 307123014.0, + "step": 12054 + }, + { + "epoch": 1.3238524050076872, + "grad_norm": 1.9911655187606812, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7198241949081421, + "num_tokens": 307152044.0, + "step": 12055 + }, + { + "epoch": 1.323962222710301, + "grad_norm": 2.3364734649658203, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7291808128356934, + "num_tokens": 307174215.0, + "step": 12056 + }, + { + "epoch": 1.3240720404129145, + "grad_norm": 2.1944448947906494, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7311685085296631, + "num_tokens": 307199298.0, + "step": 12057 + }, + { + "epoch": 1.3241818581155282, + "grad_norm": 2.4940831661224365, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.7340607047080994, + "num_tokens": 307221121.0, + "step": 12058 + }, + { + "epoch": 1.324291675818142, + "grad_norm": 2.5349411964416504, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.725764274597168, + "num_tokens": 307242619.0, + "step": 12059 + }, + { + "epoch": 1.3244014935207555, + "grad_norm": 2.1597747802734375, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7249056100845337, + "num_tokens": 307269493.0, + "step": 12060 + }, + { + "epoch": 1.324511311223369, + "grad_norm": 2.339501142501831, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7184975743293762, + "num_tokens": 307294750.0, + "step": 12061 + }, + { + "epoch": 1.3246211289259828, + "grad_norm": 2.5640010833740234, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7294511795043945, + "num_tokens": 307315639.0, + "step": 12062 + }, + { + "epoch": 1.3247309466285966, + "grad_norm": 2.2156689167022705, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.697015643119812, + "num_tokens": 307345725.0, + "step": 12063 + }, + { + "epoch": 1.3248407643312101, + "grad_norm": 2.454900026321411, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.7069438695907593, + "num_tokens": 307368567.0, + "step": 12064 + }, + { + "epoch": 1.3249505820338239, + "grad_norm": 1.8663396835327148, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7012971639633179, + "num_tokens": 307404998.0, + "step": 12065 + }, + { + "epoch": 1.3250603997364374, + "grad_norm": 2.432227849960327, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7147638201713562, + "num_tokens": 307426519.0, + "step": 12066 + }, + { + "epoch": 1.3251702174390512, + "grad_norm": 2.5364937782287598, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7200095653533936, + "num_tokens": 307447526.0, + "step": 12067 + }, + { + "epoch": 1.325280035141665, + "grad_norm": 2.2651774883270264, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7218866348266602, + "num_tokens": 307472804.0, + "step": 12068 + }, + { + "epoch": 1.3253898528442785, + "grad_norm": 2.3138415813446045, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.7033423185348511, + "num_tokens": 307498366.0, + "step": 12069 + }, + { + "epoch": 1.3254996705468922, + "grad_norm": 2.0617659091949463, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7117100954055786, + "num_tokens": 307526469.0, + "step": 12070 + }, + { + "epoch": 1.3256094882495058, + "grad_norm": 2.378260612487793, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7184740304946899, + "num_tokens": 307549229.0, + "step": 12071 + }, + { + "epoch": 1.3257193059521195, + "grad_norm": 2.2045557498931885, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.724868893623352, + "num_tokens": 307575946.0, + "step": 12072 + }, + { + "epoch": 1.3258291236547333, + "grad_norm": 2.0660128593444824, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7125818729400635, + "num_tokens": 307606255.0, + "step": 12073 + }, + { + "epoch": 1.3259389413573468, + "grad_norm": 2.3109045028686523, + "learning_rate": 1e-06, + "loss": 0.781, + "mean_token_accuracy": 0.745026171207428, + "num_tokens": 307628349.0, + "step": 12074 + }, + { + "epoch": 1.3260487590599603, + "grad_norm": 2.421043634414673, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7090286612510681, + "num_tokens": 307651110.0, + "step": 12075 + }, + { + "epoch": 1.326158576762574, + "grad_norm": 2.0512468814849854, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7394656538963318, + "num_tokens": 307680051.0, + "step": 12076 + }, + { + "epoch": 1.3262683944651878, + "grad_norm": 2.16226863861084, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7062041163444519, + "num_tokens": 307709672.0, + "step": 12077 + }, + { + "epoch": 1.3263782121678014, + "grad_norm": 2.0770959854125977, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7276955246925354, + "num_tokens": 307738665.0, + "step": 12078 + }, + { + "epoch": 1.3264880298704151, + "grad_norm": 2.1902177333831787, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.705097496509552, + "num_tokens": 307766604.0, + "step": 12079 + }, + { + "epoch": 1.3265978475730287, + "grad_norm": 2.4065184593200684, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7171366214752197, + "num_tokens": 307788886.0, + "step": 12080 + }, + { + "epoch": 1.3267076652756424, + "grad_norm": 2.3275158405303955, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7289544343948364, + "num_tokens": 307813744.0, + "step": 12081 + }, + { + "epoch": 1.3268174829782562, + "grad_norm": 2.210448741912842, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7455271482467651, + "num_tokens": 307838143.0, + "step": 12082 + }, + { + "epoch": 1.3269273006808697, + "grad_norm": 2.3934099674224854, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7305940985679626, + "num_tokens": 307860897.0, + "step": 12083 + }, + { + "epoch": 1.3270371183834835, + "grad_norm": 2.4898431301116943, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7123512625694275, + "num_tokens": 307883670.0, + "step": 12084 + }, + { + "epoch": 1.327146936086097, + "grad_norm": 2.169501304626465, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7193771600723267, + "num_tokens": 307911388.0, + "step": 12085 + }, + { + "epoch": 1.3272567537887108, + "grad_norm": 2.238379716873169, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.729217529296875, + "num_tokens": 307935568.0, + "step": 12086 + }, + { + "epoch": 1.3273665714913245, + "grad_norm": 2.3274919986724854, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7435170412063599, + "num_tokens": 307956910.0, + "step": 12087 + }, + { + "epoch": 1.327476389193938, + "grad_norm": 2.3113183975219727, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7205924391746521, + "num_tokens": 307983219.0, + "step": 12088 + }, + { + "epoch": 1.3275862068965516, + "grad_norm": 2.2006936073303223, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7315762042999268, + "num_tokens": 308009464.0, + "step": 12089 + }, + { + "epoch": 1.3276960245991654, + "grad_norm": 2.3574068546295166, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7310845851898193, + "num_tokens": 308031757.0, + "step": 12090 + }, + { + "epoch": 1.3278058423017791, + "grad_norm": 2.141127109527588, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.722793698310852, + "num_tokens": 308058513.0, + "step": 12091 + }, + { + "epoch": 1.3279156600043927, + "grad_norm": 2.3666412830352783, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7273465394973755, + "num_tokens": 308081877.0, + "step": 12092 + }, + { + "epoch": 1.3280254777070064, + "grad_norm": 2.0650522708892822, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7246216535568237, + "num_tokens": 308112152.0, + "step": 12093 + }, + { + "epoch": 1.32813529540962, + "grad_norm": 2.125863790512085, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7213248014450073, + "num_tokens": 308138406.0, + "step": 12094 + }, + { + "epoch": 1.3282451131122337, + "grad_norm": 2.288069009780884, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7070816159248352, + "num_tokens": 308165747.0, + "step": 12095 + }, + { + "epoch": 1.3283549308148475, + "grad_norm": 2.2349464893341064, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7104307413101196, + "num_tokens": 308193853.0, + "step": 12096 + }, + { + "epoch": 1.328464748517461, + "grad_norm": 2.1163790225982666, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.6903406381607056, + "num_tokens": 308223766.0, + "step": 12097 + }, + { + "epoch": 1.3285745662200747, + "grad_norm": 2.436361074447632, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7219091653823853, + "num_tokens": 308246509.0, + "step": 12098 + }, + { + "epoch": 1.3286843839226883, + "grad_norm": 2.2862415313720703, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7374049425125122, + "num_tokens": 308270330.0, + "step": 12099 + }, + { + "epoch": 1.328794201625302, + "grad_norm": 2.5760035514831543, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.719510555267334, + "num_tokens": 308290085.0, + "step": 12100 + }, + { + "epoch": 1.3289040193279158, + "grad_norm": 2.277357339859009, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7134612202644348, + "num_tokens": 308315123.0, + "step": 12101 + }, + { + "epoch": 1.3290138370305293, + "grad_norm": 2.220163345336914, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7145564556121826, + "num_tokens": 308340106.0, + "step": 12102 + }, + { + "epoch": 1.3291236547331429, + "grad_norm": 2.0158698558807373, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.711154043674469, + "num_tokens": 308370472.0, + "step": 12103 + }, + { + "epoch": 1.3292334724357566, + "grad_norm": 2.513939142227173, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7085006237030029, + "num_tokens": 308393006.0, + "step": 12104 + }, + { + "epoch": 1.3293432901383704, + "grad_norm": 2.174100637435913, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7247060537338257, + "num_tokens": 308420235.0, + "step": 12105 + }, + { + "epoch": 1.329453107840984, + "grad_norm": 2.3330583572387695, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7178587913513184, + "num_tokens": 308444554.0, + "step": 12106 + }, + { + "epoch": 1.3295629255435977, + "grad_norm": 2.071211576461792, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7176026105880737, + "num_tokens": 308474071.0, + "step": 12107 + }, + { + "epoch": 1.3296727432462112, + "grad_norm": 2.3863351345062256, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.717820942401886, + "num_tokens": 308497983.0, + "step": 12108 + }, + { + "epoch": 1.329782560948825, + "grad_norm": 2.2406504154205322, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7187584638595581, + "num_tokens": 308522824.0, + "step": 12109 + }, + { + "epoch": 1.3298923786514387, + "grad_norm": 2.6096601486206055, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7282700538635254, + "num_tokens": 308541937.0, + "step": 12110 + }, + { + "epoch": 1.3300021963540523, + "grad_norm": 2.5262420177459717, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7218490242958069, + "num_tokens": 308562719.0, + "step": 12111 + }, + { + "epoch": 1.330112014056666, + "grad_norm": 2.340595006942749, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7260445356369019, + "num_tokens": 308585999.0, + "step": 12112 + }, + { + "epoch": 1.3302218317592795, + "grad_norm": 2.379556179046631, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7140098810195923, + "num_tokens": 308610320.0, + "step": 12113 + }, + { + "epoch": 1.3303316494618933, + "grad_norm": 2.507173538208008, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7019913196563721, + "num_tokens": 308633793.0, + "step": 12114 + }, + { + "epoch": 1.3304414671645068, + "grad_norm": 2.303972005844116, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7260630130767822, + "num_tokens": 308657848.0, + "step": 12115 + }, + { + "epoch": 1.3305512848671206, + "grad_norm": 2.3758273124694824, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7046977877616882, + "num_tokens": 308683602.0, + "step": 12116 + }, + { + "epoch": 1.3306611025697341, + "grad_norm": 2.210885524749756, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7068474292755127, + "num_tokens": 308709122.0, + "step": 12117 + }, + { + "epoch": 1.330770920272348, + "grad_norm": 2.2663447856903076, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7162430286407471, + "num_tokens": 308736582.0, + "step": 12118 + }, + { + "epoch": 1.3308807379749616, + "grad_norm": 2.5908312797546387, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7397201061248779, + "num_tokens": 308755788.0, + "step": 12119 + }, + { + "epoch": 1.3309905556775752, + "grad_norm": 2.349698781967163, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.6962203979492188, + "num_tokens": 308781350.0, + "step": 12120 + }, + { + "epoch": 1.331100373380189, + "grad_norm": 2.014085531234741, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7077187895774841, + "num_tokens": 308811489.0, + "step": 12121 + }, + { + "epoch": 1.3312101910828025, + "grad_norm": 2.348019599914551, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7187308669090271, + "num_tokens": 308834299.0, + "step": 12122 + }, + { + "epoch": 1.3313200087854162, + "grad_norm": 2.197530746459961, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7308613061904907, + "num_tokens": 308859312.0, + "step": 12123 + }, + { + "epoch": 1.33142982648803, + "grad_norm": 2.0511927604675293, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7162350416183472, + "num_tokens": 308887429.0, + "step": 12124 + }, + { + "epoch": 1.3315396441906435, + "grad_norm": 2.1679625511169434, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7017582058906555, + "num_tokens": 308916462.0, + "step": 12125 + }, + { + "epoch": 1.331649461893257, + "grad_norm": 2.5106804370880127, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7204153537750244, + "num_tokens": 308937976.0, + "step": 12126 + }, + { + "epoch": 1.3317592795958708, + "grad_norm": 1.8333933353424072, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7061208486557007, + "num_tokens": 308971065.0, + "step": 12127 + }, + { + "epoch": 1.3318690972984846, + "grad_norm": 1.9490585327148438, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7218903303146362, + "num_tokens": 309002512.0, + "step": 12128 + }, + { + "epoch": 1.331978915001098, + "grad_norm": 2.1284360885620117, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7147843837738037, + "num_tokens": 309029907.0, + "step": 12129 + }, + { + "epoch": 1.3320887327037119, + "grad_norm": 2.2581355571746826, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7142232656478882, + "num_tokens": 309057078.0, + "step": 12130 + }, + { + "epoch": 1.3321985504063254, + "grad_norm": 2.1290175914764404, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7148793339729309, + "num_tokens": 309086124.0, + "step": 12131 + }, + { + "epoch": 1.3323083681089392, + "grad_norm": 1.9578100442886353, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7124622464179993, + "num_tokens": 309119900.0, + "step": 12132 + }, + { + "epoch": 1.332418185811553, + "grad_norm": 2.2324635982513428, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7087264060974121, + "num_tokens": 309147601.0, + "step": 12133 + }, + { + "epoch": 1.3325280035141664, + "grad_norm": 2.0480713844299316, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.7043495178222656, + "num_tokens": 309178573.0, + "step": 12134 + }, + { + "epoch": 1.3326378212167802, + "grad_norm": 2.386693239212036, + "learning_rate": 1e-06, + "loss": 1.0356, + "mean_token_accuracy": 0.692463755607605, + "num_tokens": 309203824.0, + "step": 12135 + }, + { + "epoch": 1.3327476389193937, + "grad_norm": 2.083059310913086, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7074981927871704, + "num_tokens": 309234309.0, + "step": 12136 + }, + { + "epoch": 1.3328574566220075, + "grad_norm": 2.1585185527801514, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7128880023956299, + "num_tokens": 309262943.0, + "step": 12137 + }, + { + "epoch": 1.3329672743246213, + "grad_norm": 2.3905723094940186, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.736929178237915, + "num_tokens": 309286093.0, + "step": 12138 + }, + { + "epoch": 1.3330770920272348, + "grad_norm": 2.444532871246338, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7223652005195618, + "num_tokens": 309308474.0, + "step": 12139 + }, + { + "epoch": 1.3331869097298483, + "grad_norm": 2.3266050815582275, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7382748126983643, + "num_tokens": 309339200.0, + "step": 12140 + }, + { + "epoch": 1.333296727432462, + "grad_norm": 2.0201711654663086, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7106060981750488, + "num_tokens": 309368803.0, + "step": 12141 + }, + { + "epoch": 1.3334065451350758, + "grad_norm": 2.1178245544433594, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7250483632087708, + "num_tokens": 309396581.0, + "step": 12142 + }, + { + "epoch": 1.3335163628376894, + "grad_norm": 2.4251675605773926, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7027856707572937, + "num_tokens": 309419041.0, + "step": 12143 + }, + { + "epoch": 1.3336261805403031, + "grad_norm": 2.1926403045654297, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.7472631335258484, + "num_tokens": 309443731.0, + "step": 12144 + }, + { + "epoch": 1.3337359982429167, + "grad_norm": 2.1379435062408447, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7232450246810913, + "num_tokens": 309471257.0, + "step": 12145 + }, + { + "epoch": 1.3338458159455304, + "grad_norm": 2.177903413772583, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.7042399048805237, + "num_tokens": 309498553.0, + "step": 12146 + }, + { + "epoch": 1.3339556336481442, + "grad_norm": 2.132061719894409, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7115763425827026, + "num_tokens": 309525748.0, + "step": 12147 + }, + { + "epoch": 1.3340654513507577, + "grad_norm": 2.454291343688965, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.735934317111969, + "num_tokens": 309548227.0, + "step": 12148 + }, + { + "epoch": 1.3341752690533715, + "grad_norm": 2.0731048583984375, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6906936168670654, + "num_tokens": 309576338.0, + "step": 12149 + }, + { + "epoch": 1.334285086755985, + "grad_norm": 2.112971067428589, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.6928836703300476, + "num_tokens": 309605112.0, + "step": 12150 + }, + { + "epoch": 1.3343949044585988, + "grad_norm": 2.247781276702881, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7250711917877197, + "num_tokens": 309629986.0, + "step": 12151 + }, + { + "epoch": 1.3345047221612125, + "grad_norm": 2.07204008102417, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.7073920369148254, + "num_tokens": 309657763.0, + "step": 12152 + }, + { + "epoch": 1.334614539863826, + "grad_norm": 2.520331859588623, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7400496602058411, + "num_tokens": 309678207.0, + "step": 12153 + }, + { + "epoch": 1.3347243575664396, + "grad_norm": 2.3897480964660645, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7168918251991272, + "num_tokens": 309703238.0, + "step": 12154 + }, + { + "epoch": 1.3348341752690533, + "grad_norm": 2.6615521907806396, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7378556728363037, + "num_tokens": 309722006.0, + "step": 12155 + }, + { + "epoch": 1.334943992971667, + "grad_norm": 2.5745108127593994, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.711963415145874, + "num_tokens": 309742729.0, + "step": 12156 + }, + { + "epoch": 1.3350538106742806, + "grad_norm": 2.1585891246795654, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7361841797828674, + "num_tokens": 309768396.0, + "step": 12157 + }, + { + "epoch": 1.3351636283768944, + "grad_norm": 2.3596951961517334, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7300715446472168, + "num_tokens": 309790359.0, + "step": 12158 + }, + { + "epoch": 1.335273446079508, + "grad_norm": 2.2658345699310303, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6919542551040649, + "num_tokens": 309815403.0, + "step": 12159 + }, + { + "epoch": 1.3353832637821217, + "grad_norm": 2.491664171218872, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7067654132843018, + "num_tokens": 309837253.0, + "step": 12160 + }, + { + "epoch": 1.3354930814847354, + "grad_norm": 2.227848529815674, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7204874753952026, + "num_tokens": 309862521.0, + "step": 12161 + }, + { + "epoch": 1.335602899187349, + "grad_norm": 2.1765432357788086, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7085721492767334, + "num_tokens": 309892294.0, + "step": 12162 + }, + { + "epoch": 1.3357127168899627, + "grad_norm": 2.2697536945343018, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7179429531097412, + "num_tokens": 309918522.0, + "step": 12163 + }, + { + "epoch": 1.3358225345925763, + "grad_norm": 2.2543349266052246, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7124167680740356, + "num_tokens": 309944919.0, + "step": 12164 + }, + { + "epoch": 1.33593235229519, + "grad_norm": 2.2545883655548096, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7172027826309204, + "num_tokens": 309970714.0, + "step": 12165 + }, + { + "epoch": 1.3360421699978036, + "grad_norm": 1.9875504970550537, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7208893299102783, + "num_tokens": 310003029.0, + "step": 12166 + }, + { + "epoch": 1.3361519877004173, + "grad_norm": 1.9629451036453247, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.6973723769187927, + "num_tokens": 310034880.0, + "step": 12167 + }, + { + "epoch": 1.3362618054030309, + "grad_norm": 2.0152716636657715, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7128868103027344, + "num_tokens": 310066024.0, + "step": 12168 + }, + { + "epoch": 1.3363716231056446, + "grad_norm": 2.5696842670440674, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7271037101745605, + "num_tokens": 310087396.0, + "step": 12169 + }, + { + "epoch": 1.3364814408082584, + "grad_norm": 2.0755834579467773, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6915521621704102, + "num_tokens": 310119312.0, + "step": 12170 + }, + { + "epoch": 1.336591258510872, + "grad_norm": 1.993411898612976, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6973003149032593, + "num_tokens": 310151304.0, + "step": 12171 + }, + { + "epoch": 1.3367010762134857, + "grad_norm": 2.660496711730957, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7359699010848999, + "num_tokens": 310171833.0, + "step": 12172 + }, + { + "epoch": 1.3368108939160992, + "grad_norm": 1.979772925376892, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.6978509426116943, + "num_tokens": 310204610.0, + "step": 12173 + }, + { + "epoch": 1.336920711618713, + "grad_norm": 2.086085796356201, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7124034762382507, + "num_tokens": 310233779.0, + "step": 12174 + }, + { + "epoch": 1.3370305293213267, + "grad_norm": 2.6218008995056152, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7350631356239319, + "num_tokens": 310252680.0, + "step": 12175 + }, + { + "epoch": 1.3371403470239402, + "grad_norm": 2.2804341316223145, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7245980501174927, + "num_tokens": 310276566.0, + "step": 12176 + }, + { + "epoch": 1.337250164726554, + "grad_norm": 2.0281965732574463, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7410275936126709, + "num_tokens": 310302183.0, + "step": 12177 + }, + { + "epoch": 1.3373599824291675, + "grad_norm": 2.3305938243865967, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7282805442810059, + "num_tokens": 310326754.0, + "step": 12178 + }, + { + "epoch": 1.3374698001317813, + "grad_norm": 2.215299606323242, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7120682597160339, + "num_tokens": 310350976.0, + "step": 12179 + }, + { + "epoch": 1.3375796178343948, + "grad_norm": 1.8999121189117432, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7055723667144775, + "num_tokens": 310384815.0, + "step": 12180 + }, + { + "epoch": 1.3376894355370086, + "grad_norm": 2.099752902984619, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7343603372573853, + "num_tokens": 310412412.0, + "step": 12181 + }, + { + "epoch": 1.3377992532396221, + "grad_norm": 2.3065218925476074, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7080498933792114, + "num_tokens": 310436521.0, + "step": 12182 + }, + { + "epoch": 1.3379090709422359, + "grad_norm": 2.435189962387085, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7323253750801086, + "num_tokens": 310458762.0, + "step": 12183 + }, + { + "epoch": 1.3380188886448496, + "grad_norm": 2.2244627475738525, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.718142569065094, + "num_tokens": 310482509.0, + "step": 12184 + }, + { + "epoch": 1.3381287063474632, + "grad_norm": 2.1445047855377197, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7080389857292175, + "num_tokens": 310510568.0, + "step": 12185 + }, + { + "epoch": 1.338238524050077, + "grad_norm": 2.5160932540893555, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7194653749465942, + "num_tokens": 310531459.0, + "step": 12186 + }, + { + "epoch": 1.3383483417526905, + "grad_norm": 2.1994614601135254, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7298440933227539, + "num_tokens": 310557664.0, + "step": 12187 + }, + { + "epoch": 1.3384581594553042, + "grad_norm": 2.062894821166992, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7229909896850586, + "num_tokens": 310584675.0, + "step": 12188 + }, + { + "epoch": 1.338567977157918, + "grad_norm": 2.1988186836242676, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7558627128601074, + "num_tokens": 310609805.0, + "step": 12189 + }, + { + "epoch": 1.3386777948605315, + "grad_norm": 2.2414355278015137, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7194441556930542, + "num_tokens": 310635551.0, + "step": 12190 + }, + { + "epoch": 1.338787612563145, + "grad_norm": 2.5437090396881104, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7221379280090332, + "num_tokens": 310655831.0, + "step": 12191 + }, + { + "epoch": 1.3388974302657588, + "grad_norm": 2.193603754043579, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.721329927444458, + "num_tokens": 310682456.0, + "step": 12192 + }, + { + "epoch": 1.3390072479683726, + "grad_norm": 1.9561762809753418, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7261579632759094, + "num_tokens": 310717101.0, + "step": 12193 + }, + { + "epoch": 1.339117065670986, + "grad_norm": 2.4143731594085693, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7266204953193665, + "num_tokens": 310739652.0, + "step": 12194 + }, + { + "epoch": 1.3392268833735999, + "grad_norm": 2.242208957672119, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7222748398780823, + "num_tokens": 310765457.0, + "step": 12195 + }, + { + "epoch": 1.3393367010762134, + "grad_norm": 2.36287260055542, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7051304578781128, + "num_tokens": 310789155.0, + "step": 12196 + }, + { + "epoch": 1.3394465187788271, + "grad_norm": 2.3726675510406494, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7335278987884521, + "num_tokens": 310813540.0, + "step": 12197 + }, + { + "epoch": 1.339556336481441, + "grad_norm": 2.635114908218384, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7411810159683228, + "num_tokens": 310832213.0, + "step": 12198 + }, + { + "epoch": 1.3396661541840544, + "grad_norm": 1.959391713142395, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6881828904151917, + "num_tokens": 310865277.0, + "step": 12199 + }, + { + "epoch": 1.3397759718866682, + "grad_norm": 2.2700562477111816, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7130794525146484, + "num_tokens": 310890764.0, + "step": 12200 + }, + { + "epoch": 1.3398857895892817, + "grad_norm": 2.122189521789551, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7155090570449829, + "num_tokens": 310919144.0, + "step": 12201 + }, + { + "epoch": 1.3399956072918955, + "grad_norm": 2.044320583343506, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7067203521728516, + "num_tokens": 310951304.0, + "step": 12202 + }, + { + "epoch": 1.3401054249945092, + "grad_norm": 2.6464569568634033, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7536729574203491, + "num_tokens": 310971038.0, + "step": 12203 + }, + { + "epoch": 1.3402152426971228, + "grad_norm": 2.3398404121398926, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7187848687171936, + "num_tokens": 310996398.0, + "step": 12204 + }, + { + "epoch": 1.3403250603997363, + "grad_norm": 2.297471761703491, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7140320539474487, + "num_tokens": 311021316.0, + "step": 12205 + }, + { + "epoch": 1.34043487810235, + "grad_norm": 2.3875458240509033, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7199604511260986, + "num_tokens": 311044605.0, + "step": 12206 + }, + { + "epoch": 1.3405446958049638, + "grad_norm": 2.2684054374694824, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7077288627624512, + "num_tokens": 311071143.0, + "step": 12207 + }, + { + "epoch": 1.3406545135075774, + "grad_norm": 2.2214772701263428, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7456731200218201, + "num_tokens": 311098632.0, + "step": 12208 + }, + { + "epoch": 1.3407643312101911, + "grad_norm": 2.399153709411621, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.7499769926071167, + "num_tokens": 311121213.0, + "step": 12209 + }, + { + "epoch": 1.3408741489128047, + "grad_norm": 2.4510498046875, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7230030298233032, + "num_tokens": 311143619.0, + "step": 12210 + }, + { + "epoch": 1.3409839666154184, + "grad_norm": 2.059511661529541, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7157861590385437, + "num_tokens": 311171149.0, + "step": 12211 + }, + { + "epoch": 1.3410937843180322, + "grad_norm": 2.3626925945281982, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.730370819568634, + "num_tokens": 311194351.0, + "step": 12212 + }, + { + "epoch": 1.3412036020206457, + "grad_norm": 2.333997964859009, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7366796731948853, + "num_tokens": 311217358.0, + "step": 12213 + }, + { + "epoch": 1.3413134197232595, + "grad_norm": 2.3022992610931396, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7463060021400452, + "num_tokens": 311241550.0, + "step": 12214 + }, + { + "epoch": 1.341423237425873, + "grad_norm": 1.9779027700424194, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.6983960866928101, + "num_tokens": 311271919.0, + "step": 12215 + }, + { + "epoch": 1.3415330551284868, + "grad_norm": 2.2291364669799805, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7175769805908203, + "num_tokens": 311296714.0, + "step": 12216 + }, + { + "epoch": 1.3416428728311005, + "grad_norm": 2.101517915725708, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.7050104141235352, + "num_tokens": 311325659.0, + "step": 12217 + }, + { + "epoch": 1.341752690533714, + "grad_norm": 2.806649923324585, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7204881310462952, + "num_tokens": 311343140.0, + "step": 12218 + }, + { + "epoch": 1.3418625082363276, + "grad_norm": 2.100752115249634, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7258029580116272, + "num_tokens": 311371434.0, + "step": 12219 + }, + { + "epoch": 1.3419723259389413, + "grad_norm": 2.3998055458068848, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7095226049423218, + "num_tokens": 311394586.0, + "step": 12220 + }, + { + "epoch": 1.342082143641555, + "grad_norm": 2.4197769165039062, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7226073741912842, + "num_tokens": 311417544.0, + "step": 12221 + }, + { + "epoch": 1.3421919613441686, + "grad_norm": 1.948434829711914, + "learning_rate": 1e-06, + "loss": 1.0254, + "mean_token_accuracy": 0.6929434537887573, + "num_tokens": 311450354.0, + "step": 12222 + }, + { + "epoch": 1.3423017790467824, + "grad_norm": 2.25776743888855, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7423171997070312, + "num_tokens": 311474611.0, + "step": 12223 + }, + { + "epoch": 1.342411596749396, + "grad_norm": 2.120574951171875, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7296598553657532, + "num_tokens": 311499949.0, + "step": 12224 + }, + { + "epoch": 1.3425214144520097, + "grad_norm": 2.1948630809783936, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.741007924079895, + "num_tokens": 311525527.0, + "step": 12225 + }, + { + "epoch": 1.3426312321546234, + "grad_norm": 2.513572931289673, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7182983160018921, + "num_tokens": 311548712.0, + "step": 12226 + }, + { + "epoch": 1.342741049857237, + "grad_norm": 2.3141047954559326, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7179353833198547, + "num_tokens": 311573809.0, + "step": 12227 + }, + { + "epoch": 1.3428508675598507, + "grad_norm": 2.68292498588562, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7167447209358215, + "num_tokens": 311592147.0, + "step": 12228 + }, + { + "epoch": 1.3429606852624643, + "grad_norm": 2.2014613151550293, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7232638597488403, + "num_tokens": 311621387.0, + "step": 12229 + }, + { + "epoch": 1.343070502965078, + "grad_norm": 2.404003381729126, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7237092852592468, + "num_tokens": 311644565.0, + "step": 12230 + }, + { + "epoch": 1.3431803206676916, + "grad_norm": 2.3473246097564697, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7288758754730225, + "num_tokens": 311668300.0, + "step": 12231 + }, + { + "epoch": 1.3432901383703053, + "grad_norm": 2.4008655548095703, + "learning_rate": 1e-06, + "loss": 0.8266, + "mean_token_accuracy": 0.7321297526359558, + "num_tokens": 311690527.0, + "step": 12232 + }, + { + "epoch": 1.3433999560729188, + "grad_norm": 2.1096513271331787, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7112652659416199, + "num_tokens": 311720555.0, + "step": 12233 + }, + { + "epoch": 1.3435097737755326, + "grad_norm": 2.0641138553619385, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7002846002578735, + "num_tokens": 311748857.0, + "step": 12234 + }, + { + "epoch": 1.3436195914781464, + "grad_norm": 1.9855252504348755, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7122981548309326, + "num_tokens": 311779296.0, + "step": 12235 + }, + { + "epoch": 1.34372940918076, + "grad_norm": 2.2111635208129883, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7121541500091553, + "num_tokens": 311805637.0, + "step": 12236 + }, + { + "epoch": 1.3438392268833736, + "grad_norm": 2.3659377098083496, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7284777164459229, + "num_tokens": 311828930.0, + "step": 12237 + }, + { + "epoch": 1.3439490445859872, + "grad_norm": 2.27555775642395, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.6969246864318848, + "num_tokens": 311856588.0, + "step": 12238 + }, + { + "epoch": 1.344058862288601, + "grad_norm": 1.9339759349822998, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7111155986785889, + "num_tokens": 311890479.0, + "step": 12239 + }, + { + "epoch": 1.3441686799912147, + "grad_norm": 2.16477632522583, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7166444063186646, + "num_tokens": 311918319.0, + "step": 12240 + }, + { + "epoch": 1.3442784976938282, + "grad_norm": 2.1692581176757812, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6904982328414917, + "num_tokens": 311945471.0, + "step": 12241 + }, + { + "epoch": 1.3443883153964418, + "grad_norm": 2.2608280181884766, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7091004848480225, + "num_tokens": 311970062.0, + "step": 12242 + }, + { + "epoch": 1.3444981330990555, + "grad_norm": 2.5988857746124268, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.7504042387008667, + "num_tokens": 311989329.0, + "step": 12243 + }, + { + "epoch": 1.3446079508016693, + "grad_norm": 1.9835364818572998, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7009357213973999, + "num_tokens": 312022300.0, + "step": 12244 + }, + { + "epoch": 1.3447177685042828, + "grad_norm": 2.2311363220214844, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7158378958702087, + "num_tokens": 312048875.0, + "step": 12245 + }, + { + "epoch": 1.3448275862068966, + "grad_norm": 2.421755075454712, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7407115697860718, + "num_tokens": 312069530.0, + "step": 12246 + }, + { + "epoch": 1.34493740390951, + "grad_norm": 2.347801685333252, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7147630453109741, + "num_tokens": 312094664.0, + "step": 12247 + }, + { + "epoch": 1.3450472216121239, + "grad_norm": 2.373007297515869, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7001497149467468, + "num_tokens": 312120389.0, + "step": 12248 + }, + { + "epoch": 1.3451570393147376, + "grad_norm": 2.0935304164886475, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.6867437958717346, + "num_tokens": 312151055.0, + "step": 12249 + }, + { + "epoch": 1.3452668570173512, + "grad_norm": 2.1386940479278564, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7039282917976379, + "num_tokens": 312182010.0, + "step": 12250 + }, + { + "epoch": 1.345376674719965, + "grad_norm": 2.6329598426818848, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7368548512458801, + "num_tokens": 312202779.0, + "step": 12251 + }, + { + "epoch": 1.3454864924225785, + "grad_norm": 2.3973124027252197, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7186493873596191, + "num_tokens": 312227600.0, + "step": 12252 + }, + { + "epoch": 1.3455963101251922, + "grad_norm": 2.162360191345215, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7219882011413574, + "num_tokens": 312254268.0, + "step": 12253 + }, + { + "epoch": 1.345706127827806, + "grad_norm": 2.37778902053833, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7142826914787292, + "num_tokens": 312276446.0, + "step": 12254 + }, + { + "epoch": 1.3458159455304195, + "grad_norm": 2.4292056560516357, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7429652214050293, + "num_tokens": 312297384.0, + "step": 12255 + }, + { + "epoch": 1.345925763233033, + "grad_norm": 2.374661684036255, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7312362194061279, + "num_tokens": 312320633.0, + "step": 12256 + }, + { + "epoch": 1.3460355809356468, + "grad_norm": 2.3342113494873047, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7162141799926758, + "num_tokens": 312344146.0, + "step": 12257 + }, + { + "epoch": 1.3461453986382605, + "grad_norm": 2.2153146266937256, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7158607840538025, + "num_tokens": 312372798.0, + "step": 12258 + }, + { + "epoch": 1.346255216340874, + "grad_norm": 2.40714693069458, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.7496539354324341, + "num_tokens": 312394537.0, + "step": 12259 + }, + { + "epoch": 1.3463650340434878, + "grad_norm": 2.3678977489471436, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7384551763534546, + "num_tokens": 312418345.0, + "step": 12260 + }, + { + "epoch": 1.3464748517461014, + "grad_norm": 2.623732566833496, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7288321256637573, + "num_tokens": 312439149.0, + "step": 12261 + }, + { + "epoch": 1.3465846694487151, + "grad_norm": 2.2125461101531982, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7304426431655884, + "num_tokens": 312465328.0, + "step": 12262 + }, + { + "epoch": 1.346694487151329, + "grad_norm": 2.042131185531616, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7074569463729858, + "num_tokens": 312495368.0, + "step": 12263 + }, + { + "epoch": 1.3468043048539424, + "grad_norm": 2.243912935256958, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7029733657836914, + "num_tokens": 312523493.0, + "step": 12264 + }, + { + "epoch": 1.3469141225565562, + "grad_norm": 2.1398284435272217, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7267999649047852, + "num_tokens": 312551889.0, + "step": 12265 + }, + { + "epoch": 1.3470239402591697, + "grad_norm": 2.337125062942505, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7173650860786438, + "num_tokens": 312575249.0, + "step": 12266 + }, + { + "epoch": 1.3471337579617835, + "grad_norm": 2.044963836669922, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.6934632658958435, + "num_tokens": 312604863.0, + "step": 12267 + }, + { + "epoch": 1.3472435756643972, + "grad_norm": 2.1886816024780273, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.6895322799682617, + "num_tokens": 312632899.0, + "step": 12268 + }, + { + "epoch": 1.3473533933670108, + "grad_norm": 2.5122017860412598, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.6988905072212219, + "num_tokens": 312653887.0, + "step": 12269 + }, + { + "epoch": 1.3474632110696243, + "grad_norm": 2.208111524581909, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7017073631286621, + "num_tokens": 312680105.0, + "step": 12270 + }, + { + "epoch": 1.347573028772238, + "grad_norm": 1.9029794931411743, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7356780767440796, + "num_tokens": 312712370.0, + "step": 12271 + }, + { + "epoch": 1.3476828464748518, + "grad_norm": 2.3201045989990234, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7193637490272522, + "num_tokens": 312737983.0, + "step": 12272 + }, + { + "epoch": 1.3477926641774653, + "grad_norm": 2.2370622158050537, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.6993816494941711, + "num_tokens": 312763755.0, + "step": 12273 + }, + { + "epoch": 1.347902481880079, + "grad_norm": 2.3510007858276367, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.703089714050293, + "num_tokens": 312787798.0, + "step": 12274 + }, + { + "epoch": 1.3480122995826926, + "grad_norm": 2.0888500213623047, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7089186906814575, + "num_tokens": 312818067.0, + "step": 12275 + }, + { + "epoch": 1.3481221172853064, + "grad_norm": 1.9979661703109741, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7298145890235901, + "num_tokens": 312847061.0, + "step": 12276 + }, + { + "epoch": 1.3482319349879202, + "grad_norm": 2.2956173419952393, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7203254103660583, + "num_tokens": 312871604.0, + "step": 12277 + }, + { + "epoch": 1.3483417526905337, + "grad_norm": 2.1655445098876953, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7025195956230164, + "num_tokens": 312901517.0, + "step": 12278 + }, + { + "epoch": 1.3484515703931474, + "grad_norm": 1.959700584411621, + "learning_rate": 1e-06, + "loss": 1.0329, + "mean_token_accuracy": 0.6923637390136719, + "num_tokens": 312937752.0, + "step": 12279 + }, + { + "epoch": 1.348561388095761, + "grad_norm": 2.056025981903076, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.6980012655258179, + "num_tokens": 312968590.0, + "step": 12280 + }, + { + "epoch": 1.3486712057983747, + "grad_norm": 2.466235637664795, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7293436527252197, + "num_tokens": 312989253.0, + "step": 12281 + }, + { + "epoch": 1.3487810235009885, + "grad_norm": 2.2034263610839844, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7198973894119263, + "num_tokens": 313014634.0, + "step": 12282 + }, + { + "epoch": 1.348890841203602, + "grad_norm": 2.122231960296631, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7367455959320068, + "num_tokens": 313043572.0, + "step": 12283 + }, + { + "epoch": 1.3490006589062156, + "grad_norm": 2.318881034851074, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.721998393535614, + "num_tokens": 313066785.0, + "step": 12284 + }, + { + "epoch": 1.3491104766088293, + "grad_norm": 2.0951411724090576, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7017044425010681, + "num_tokens": 313095095.0, + "step": 12285 + }, + { + "epoch": 1.349220294311443, + "grad_norm": 2.321991205215454, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7041563987731934, + "num_tokens": 313120175.0, + "step": 12286 + }, + { + "epoch": 1.3493301120140566, + "grad_norm": 2.400974750518799, + "learning_rate": 1e-06, + "loss": 0.8287, + "mean_token_accuracy": 0.735871434211731, + "num_tokens": 313142312.0, + "step": 12287 + }, + { + "epoch": 1.3494399297166704, + "grad_norm": 2.425914764404297, + "learning_rate": 1e-06, + "loss": 0.8128, + "mean_token_accuracy": 0.7400103807449341, + "num_tokens": 313163130.0, + "step": 12288 + }, + { + "epoch": 1.349549747419284, + "grad_norm": 2.21539568901062, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7382198572158813, + "num_tokens": 313187526.0, + "step": 12289 + }, + { + "epoch": 1.3496595651218977, + "grad_norm": 2.1804072856903076, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7083587050437927, + "num_tokens": 313213390.0, + "step": 12290 + }, + { + "epoch": 1.3497693828245114, + "grad_norm": 2.336543560028076, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7221832871437073, + "num_tokens": 313237160.0, + "step": 12291 + }, + { + "epoch": 1.349879200527125, + "grad_norm": 2.0744524002075195, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7219991683959961, + "num_tokens": 313265234.0, + "step": 12292 + }, + { + "epoch": 1.3499890182297387, + "grad_norm": 2.4291460514068604, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7113046050071716, + "num_tokens": 313288927.0, + "step": 12293 + }, + { + "epoch": 1.3500988359323522, + "grad_norm": 2.4333655834198, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7155296802520752, + "num_tokens": 313311170.0, + "step": 12294 + }, + { + "epoch": 1.350208653634966, + "grad_norm": 2.3881475925445557, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7190716862678528, + "num_tokens": 313332696.0, + "step": 12295 + }, + { + "epoch": 1.3503184713375795, + "grad_norm": 2.2543423175811768, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7296839952468872, + "num_tokens": 313359346.0, + "step": 12296 + }, + { + "epoch": 1.3504282890401933, + "grad_norm": 2.242112398147583, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7189513444900513, + "num_tokens": 313387387.0, + "step": 12297 + }, + { + "epoch": 1.3505381067428068, + "grad_norm": 1.9483826160430908, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7226609587669373, + "num_tokens": 313419973.0, + "step": 12298 + }, + { + "epoch": 1.3506479244454206, + "grad_norm": 2.1317431926727295, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7049237489700317, + "num_tokens": 313446697.0, + "step": 12299 + }, + { + "epoch": 1.3507577421480343, + "grad_norm": 2.291067123413086, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7332814335823059, + "num_tokens": 313470786.0, + "step": 12300 + }, + { + "epoch": 1.3508675598506479, + "grad_norm": 2.179852247238159, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.6915508508682251, + "num_tokens": 313499376.0, + "step": 12301 + }, + { + "epoch": 1.3509773775532616, + "grad_norm": 2.5485799312591553, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7241397500038147, + "num_tokens": 313518967.0, + "step": 12302 + }, + { + "epoch": 1.3510871952558752, + "grad_norm": 2.1781578063964844, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7182637453079224, + "num_tokens": 313546166.0, + "step": 12303 + }, + { + "epoch": 1.351197012958489, + "grad_norm": 2.316225290298462, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7099034786224365, + "num_tokens": 313571083.0, + "step": 12304 + }, + { + "epoch": 1.3513068306611027, + "grad_norm": 2.3655753135681152, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7017993927001953, + "num_tokens": 313595052.0, + "step": 12305 + }, + { + "epoch": 1.3514166483637162, + "grad_norm": 2.7880871295928955, + "learning_rate": 1e-06, + "loss": 0.757, + "mean_token_accuracy": 0.752966046333313, + "num_tokens": 313611961.0, + "step": 12306 + }, + { + "epoch": 1.3515264660663298, + "grad_norm": 2.1538822650909424, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6876887083053589, + "num_tokens": 313638710.0, + "step": 12307 + }, + { + "epoch": 1.3516362837689435, + "grad_norm": 2.528224468231201, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7331724762916565, + "num_tokens": 313658883.0, + "step": 12308 + }, + { + "epoch": 1.3517461014715573, + "grad_norm": 2.2384235858917236, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7142794132232666, + "num_tokens": 313685634.0, + "step": 12309 + }, + { + "epoch": 1.3518559191741708, + "grad_norm": 2.482241153717041, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.71542888879776, + "num_tokens": 313707817.0, + "step": 12310 + }, + { + "epoch": 1.3519657368767846, + "grad_norm": 2.316831588745117, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7303203344345093, + "num_tokens": 313732847.0, + "step": 12311 + }, + { + "epoch": 1.352075554579398, + "grad_norm": 2.292673110961914, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7228732109069824, + "num_tokens": 313756603.0, + "step": 12312 + }, + { + "epoch": 1.3521853722820119, + "grad_norm": 2.243776321411133, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7177756428718567, + "num_tokens": 313781904.0, + "step": 12313 + }, + { + "epoch": 1.3522951899846256, + "grad_norm": 2.0731260776519775, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7299678325653076, + "num_tokens": 313810301.0, + "step": 12314 + }, + { + "epoch": 1.3524050076872391, + "grad_norm": 2.5294103622436523, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6986753940582275, + "num_tokens": 313836307.0, + "step": 12315 + }, + { + "epoch": 1.352514825389853, + "grad_norm": 2.2528011798858643, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7365871667861938, + "num_tokens": 313863913.0, + "step": 12316 + }, + { + "epoch": 1.3526246430924664, + "grad_norm": 2.2575860023498535, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.722232460975647, + "num_tokens": 313889955.0, + "step": 12317 + }, + { + "epoch": 1.3527344607950802, + "grad_norm": 2.3412585258483887, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.719997763633728, + "num_tokens": 313914949.0, + "step": 12318 + }, + { + "epoch": 1.352844278497694, + "grad_norm": 2.1993510723114014, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7120066285133362, + "num_tokens": 313940445.0, + "step": 12319 + }, + { + "epoch": 1.3529540962003075, + "grad_norm": 2.1537463665008545, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7057222127914429, + "num_tokens": 313968219.0, + "step": 12320 + }, + { + "epoch": 1.353063913902921, + "grad_norm": 2.3024303913116455, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7235460877418518, + "num_tokens": 313992144.0, + "step": 12321 + }, + { + "epoch": 1.3531737316055348, + "grad_norm": 2.073867082595825, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.722048282623291, + "num_tokens": 314022253.0, + "step": 12322 + }, + { + "epoch": 1.3532835493081485, + "grad_norm": 2.407444715499878, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7385647296905518, + "num_tokens": 314045038.0, + "step": 12323 + }, + { + "epoch": 1.353393367010762, + "grad_norm": 2.3115358352661133, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7317171096801758, + "num_tokens": 314067992.0, + "step": 12324 + }, + { + "epoch": 1.3535031847133758, + "grad_norm": 2.190960168838501, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7103749513626099, + "num_tokens": 314093099.0, + "step": 12325 + }, + { + "epoch": 1.3536130024159894, + "grad_norm": 2.491978883743286, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7228304147720337, + "num_tokens": 314116171.0, + "step": 12326 + }, + { + "epoch": 1.3537228201186031, + "grad_norm": 2.2377569675445557, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7256622314453125, + "num_tokens": 314140347.0, + "step": 12327 + }, + { + "epoch": 1.3538326378212169, + "grad_norm": 2.3041129112243652, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7155351042747498, + "num_tokens": 314164572.0, + "step": 12328 + }, + { + "epoch": 1.3539424555238304, + "grad_norm": 2.218846559524536, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7217632532119751, + "num_tokens": 314190711.0, + "step": 12329 + }, + { + "epoch": 1.3540522732264442, + "grad_norm": 2.2285287380218506, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.729154109954834, + "num_tokens": 314216236.0, + "step": 12330 + }, + { + "epoch": 1.3541620909290577, + "grad_norm": 2.1621694564819336, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7139748334884644, + "num_tokens": 314244341.0, + "step": 12331 + }, + { + "epoch": 1.3542719086316715, + "grad_norm": 2.381870985031128, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7190009951591492, + "num_tokens": 314268358.0, + "step": 12332 + }, + { + "epoch": 1.3543817263342852, + "grad_norm": 2.3438215255737305, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7410157918930054, + "num_tokens": 314290694.0, + "step": 12333 + }, + { + "epoch": 1.3544915440368988, + "grad_norm": 2.1992452144622803, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7179820537567139, + "num_tokens": 314319332.0, + "step": 12334 + }, + { + "epoch": 1.3546013617395123, + "grad_norm": 2.0394539833068848, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7040985226631165, + "num_tokens": 314350427.0, + "step": 12335 + }, + { + "epoch": 1.354711179442126, + "grad_norm": 2.766237497329712, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7332181930541992, + "num_tokens": 314368307.0, + "step": 12336 + }, + { + "epoch": 1.3548209971447398, + "grad_norm": 2.288799285888672, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.708897054195404, + "num_tokens": 314394736.0, + "step": 12337 + }, + { + "epoch": 1.3549308148473533, + "grad_norm": 2.5390398502349854, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7192062139511108, + "num_tokens": 314415177.0, + "step": 12338 + }, + { + "epoch": 1.355040632549967, + "grad_norm": 2.0965416431427, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.726446807384491, + "num_tokens": 314444515.0, + "step": 12339 + }, + { + "epoch": 1.3551504502525806, + "grad_norm": 2.3585829734802246, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7178264856338501, + "num_tokens": 314469786.0, + "step": 12340 + }, + { + "epoch": 1.3552602679551944, + "grad_norm": 2.1342597007751465, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7022720575332642, + "num_tokens": 314496504.0, + "step": 12341 + }, + { + "epoch": 1.3553700856578081, + "grad_norm": 2.58243727684021, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7353229522705078, + "num_tokens": 314515902.0, + "step": 12342 + }, + { + "epoch": 1.3554799033604217, + "grad_norm": 2.1903295516967773, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7312624454498291, + "num_tokens": 314541990.0, + "step": 12343 + }, + { + "epoch": 1.3555897210630354, + "grad_norm": 2.067561149597168, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7147796154022217, + "num_tokens": 314572054.0, + "step": 12344 + }, + { + "epoch": 1.355699538765649, + "grad_norm": 2.4128308296203613, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7330796718597412, + "num_tokens": 314594132.0, + "step": 12345 + }, + { + "epoch": 1.3558093564682627, + "grad_norm": 2.4060769081115723, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7243673205375671, + "num_tokens": 314616560.0, + "step": 12346 + }, + { + "epoch": 1.3559191741708763, + "grad_norm": 2.5769472122192383, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.741417646408081, + "num_tokens": 314635240.0, + "step": 12347 + }, + { + "epoch": 1.35602899187349, + "grad_norm": 2.3329501152038574, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7266616821289062, + "num_tokens": 314659065.0, + "step": 12348 + }, + { + "epoch": 1.3561388095761036, + "grad_norm": 2.264622926712036, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7150678634643555, + "num_tokens": 314684261.0, + "step": 12349 + }, + { + "epoch": 1.3562486272787173, + "grad_norm": 2.2334542274475098, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7332198619842529, + "num_tokens": 314710795.0, + "step": 12350 + }, + { + "epoch": 1.356358444981331, + "grad_norm": 2.141052484512329, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7128114700317383, + "num_tokens": 314738122.0, + "step": 12351 + }, + { + "epoch": 1.3564682626839446, + "grad_norm": 2.404256582260132, + "learning_rate": 1e-06, + "loss": 0.7465, + "mean_token_accuracy": 0.7671425938606262, + "num_tokens": 314758628.0, + "step": 12352 + }, + { + "epoch": 1.3565780803865584, + "grad_norm": 2.092322587966919, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.696100652217865, + "num_tokens": 314787911.0, + "step": 12353 + }, + { + "epoch": 1.356687898089172, + "grad_norm": 2.0712480545043945, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7176132202148438, + "num_tokens": 314816650.0, + "step": 12354 + }, + { + "epoch": 1.3567977157917857, + "grad_norm": 2.1581602096557617, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7352141737937927, + "num_tokens": 314843906.0, + "step": 12355 + }, + { + "epoch": 1.3569075334943994, + "grad_norm": 2.5472354888916016, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7216963171958923, + "num_tokens": 314864260.0, + "step": 12356 + }, + { + "epoch": 1.357017351197013, + "grad_norm": 2.4295077323913574, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7215992212295532, + "num_tokens": 314888103.0, + "step": 12357 + }, + { + "epoch": 1.3571271688996267, + "grad_norm": 2.1509358882904053, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.728359043598175, + "num_tokens": 314916167.0, + "step": 12358 + }, + { + "epoch": 1.3572369866022402, + "grad_norm": 2.5571205615997314, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7407057285308838, + "num_tokens": 314937282.0, + "step": 12359 + }, + { + "epoch": 1.357346804304854, + "grad_norm": 2.046499490737915, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.6968607306480408, + "num_tokens": 314965771.0, + "step": 12360 + }, + { + "epoch": 1.3574566220074675, + "grad_norm": 2.3300580978393555, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7163633108139038, + "num_tokens": 314989466.0, + "step": 12361 + }, + { + "epoch": 1.3575664397100813, + "grad_norm": 2.3645732402801514, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.7067452669143677, + "num_tokens": 315015412.0, + "step": 12362 + }, + { + "epoch": 1.3576762574126948, + "grad_norm": 2.402448892593384, + "learning_rate": 1e-06, + "loss": 0.8045, + "mean_token_accuracy": 0.747651994228363, + "num_tokens": 315037102.0, + "step": 12363 + }, + { + "epoch": 1.3577860751153086, + "grad_norm": 2.411078929901123, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7222893238067627, + "num_tokens": 315059842.0, + "step": 12364 + }, + { + "epoch": 1.3578958928179223, + "grad_norm": 2.212337017059326, + "learning_rate": 1e-06, + "loss": 0.7802, + "mean_token_accuracy": 0.7531795501708984, + "num_tokens": 315084958.0, + "step": 12365 + }, + { + "epoch": 1.3580057105205359, + "grad_norm": 2.5668632984161377, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7149141430854797, + "num_tokens": 315106068.0, + "step": 12366 + }, + { + "epoch": 1.3581155282231496, + "grad_norm": 2.2457244396209717, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.721408486366272, + "num_tokens": 315133026.0, + "step": 12367 + }, + { + "epoch": 1.3582253459257632, + "grad_norm": 2.7620584964752197, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7425538301467896, + "num_tokens": 315153440.0, + "step": 12368 + }, + { + "epoch": 1.358335163628377, + "grad_norm": 2.587815523147583, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7069675326347351, + "num_tokens": 315175295.0, + "step": 12369 + }, + { + "epoch": 1.3584449813309907, + "grad_norm": 1.9286483526229858, + "learning_rate": 1e-06, + "loss": 1.0137, + "mean_token_accuracy": 0.6912403106689453, + "num_tokens": 315209380.0, + "step": 12370 + }, + { + "epoch": 1.3585547990336042, + "grad_norm": 2.1066534519195557, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7146741151809692, + "num_tokens": 315238216.0, + "step": 12371 + }, + { + "epoch": 1.3586646167362177, + "grad_norm": 2.814915180206299, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7165639400482178, + "num_tokens": 315258354.0, + "step": 12372 + }, + { + "epoch": 1.3587744344388315, + "grad_norm": 2.260715961456299, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.6965246200561523, + "num_tokens": 315284597.0, + "step": 12373 + }, + { + "epoch": 1.3588842521414453, + "grad_norm": 2.1592283248901367, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7036393284797668, + "num_tokens": 315309770.0, + "step": 12374 + }, + { + "epoch": 1.3589940698440588, + "grad_norm": 2.315894842147827, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7357560396194458, + "num_tokens": 315332970.0, + "step": 12375 + }, + { + "epoch": 1.3591038875466726, + "grad_norm": 2.259023666381836, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7278851866722107, + "num_tokens": 315359266.0, + "step": 12376 + }, + { + "epoch": 1.359213705249286, + "grad_norm": 2.2365026473999023, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7271720767021179, + "num_tokens": 315382744.0, + "step": 12377 + }, + { + "epoch": 1.3593235229518998, + "grad_norm": 2.404287338256836, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7300006151199341, + "num_tokens": 315406259.0, + "step": 12378 + }, + { + "epoch": 1.3594333406545136, + "grad_norm": 2.303919792175293, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7453252077102661, + "num_tokens": 315429734.0, + "step": 12379 + }, + { + "epoch": 1.3595431583571271, + "grad_norm": 2.5243782997131348, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7151347398757935, + "num_tokens": 315451089.0, + "step": 12380 + }, + { + "epoch": 1.359652976059741, + "grad_norm": 2.1183905601501465, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7275832295417786, + "num_tokens": 315479445.0, + "step": 12381 + }, + { + "epoch": 1.3597627937623544, + "grad_norm": 2.6056864261627197, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7155072689056396, + "num_tokens": 315499057.0, + "step": 12382 + }, + { + "epoch": 1.3598726114649682, + "grad_norm": 2.5266613960266113, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.6986805200576782, + "num_tokens": 315522789.0, + "step": 12383 + }, + { + "epoch": 1.359982429167582, + "grad_norm": 2.4375064373016357, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7096459865570068, + "num_tokens": 315546294.0, + "step": 12384 + }, + { + "epoch": 1.3600922468701955, + "grad_norm": 2.3765664100646973, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7055168151855469, + "num_tokens": 315571267.0, + "step": 12385 + }, + { + "epoch": 1.360202064572809, + "grad_norm": 2.1493966579437256, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6951082944869995, + "num_tokens": 315601844.0, + "step": 12386 + }, + { + "epoch": 1.3603118822754228, + "grad_norm": 2.355201244354248, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7307403683662415, + "num_tokens": 315625720.0, + "step": 12387 + }, + { + "epoch": 1.3604216999780365, + "grad_norm": 2.101120948791504, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7108256816864014, + "num_tokens": 315655327.0, + "step": 12388 + }, + { + "epoch": 1.36053151768065, + "grad_norm": 2.32942271232605, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7401950359344482, + "num_tokens": 315678488.0, + "step": 12389 + }, + { + "epoch": 1.3606413353832638, + "grad_norm": 1.9776744842529297, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7458242177963257, + "num_tokens": 315709712.0, + "step": 12390 + }, + { + "epoch": 1.3607511530858774, + "grad_norm": 2.3575432300567627, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7166925072669983, + "num_tokens": 315732120.0, + "step": 12391 + }, + { + "epoch": 1.360860970788491, + "grad_norm": 2.1865122318267822, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.6956804990768433, + "num_tokens": 315758514.0, + "step": 12392 + }, + { + "epoch": 1.3609707884911049, + "grad_norm": 2.2365574836730957, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7050196528434753, + "num_tokens": 315785282.0, + "step": 12393 + }, + { + "epoch": 1.3610806061937184, + "grad_norm": 2.4589295387268066, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7292454242706299, + "num_tokens": 315807325.0, + "step": 12394 + }, + { + "epoch": 1.3611904238963322, + "grad_norm": 2.2503552436828613, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7405089735984802, + "num_tokens": 315832391.0, + "step": 12395 + }, + { + "epoch": 1.3613002415989457, + "grad_norm": 2.3599741458892822, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7229164242744446, + "num_tokens": 315856325.0, + "step": 12396 + }, + { + "epoch": 1.3614100593015594, + "grad_norm": 2.169097900390625, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.69698166847229, + "num_tokens": 315885118.0, + "step": 12397 + }, + { + "epoch": 1.3615198770041732, + "grad_norm": 2.086688280105591, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7067270278930664, + "num_tokens": 315915740.0, + "step": 12398 + }, + { + "epoch": 1.3616296947067867, + "grad_norm": 2.171201229095459, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6950349807739258, + "num_tokens": 315945303.0, + "step": 12399 + }, + { + "epoch": 1.3617395124094003, + "grad_norm": 2.181948184967041, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7126195430755615, + "num_tokens": 315972872.0, + "step": 12400 + }, + { + "epoch": 1.361849330112014, + "grad_norm": 2.4823660850524902, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7074328064918518, + "num_tokens": 315994525.0, + "step": 12401 + }, + { + "epoch": 1.3619591478146278, + "grad_norm": 2.292830467224121, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7250536680221558, + "num_tokens": 316017834.0, + "step": 12402 + }, + { + "epoch": 1.3620689655172413, + "grad_norm": 2.4503285884857178, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.705311119556427, + "num_tokens": 316040501.0, + "step": 12403 + }, + { + "epoch": 1.362178783219855, + "grad_norm": 2.232745409011841, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7219319343566895, + "num_tokens": 316067519.0, + "step": 12404 + }, + { + "epoch": 1.3622886009224686, + "grad_norm": 2.196369171142578, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6942906379699707, + "num_tokens": 316096167.0, + "step": 12405 + }, + { + "epoch": 1.3623984186250824, + "grad_norm": 2.2420852184295654, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7409474849700928, + "num_tokens": 316120688.0, + "step": 12406 + }, + { + "epoch": 1.3625082363276961, + "grad_norm": 2.4635791778564453, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7168042063713074, + "num_tokens": 316142998.0, + "step": 12407 + }, + { + "epoch": 1.3626180540303097, + "grad_norm": 2.472896099090576, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7193108797073364, + "num_tokens": 316165774.0, + "step": 12408 + }, + { + "epoch": 1.3627278717329234, + "grad_norm": 2.162644386291504, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7277056574821472, + "num_tokens": 316193168.0, + "step": 12409 + }, + { + "epoch": 1.362837689435537, + "grad_norm": 2.3090105056762695, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7229428291320801, + "num_tokens": 316221141.0, + "step": 12410 + }, + { + "epoch": 1.3629475071381507, + "grad_norm": 2.3242228031158447, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.737308144569397, + "num_tokens": 316244930.0, + "step": 12411 + }, + { + "epoch": 1.3630573248407643, + "grad_norm": 2.131497383117676, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7099006772041321, + "num_tokens": 316271866.0, + "step": 12412 + }, + { + "epoch": 1.363167142543378, + "grad_norm": 2.245347261428833, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7288761734962463, + "num_tokens": 316295606.0, + "step": 12413 + }, + { + "epoch": 1.3632769602459915, + "grad_norm": 2.3658406734466553, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7345259189605713, + "num_tokens": 316317739.0, + "step": 12414 + }, + { + "epoch": 1.3633867779486053, + "grad_norm": 2.3116350173950195, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7104777097702026, + "num_tokens": 316342786.0, + "step": 12415 + }, + { + "epoch": 1.363496595651219, + "grad_norm": 2.2465009689331055, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.7003788948059082, + "num_tokens": 316370571.0, + "step": 12416 + }, + { + "epoch": 1.3636064133538326, + "grad_norm": 2.2179574966430664, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7171747088432312, + "num_tokens": 316395764.0, + "step": 12417 + }, + { + "epoch": 1.3637162310564463, + "grad_norm": 1.9672025442123413, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6812982559204102, + "num_tokens": 316429743.0, + "step": 12418 + }, + { + "epoch": 1.3638260487590599, + "grad_norm": 2.060849666595459, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7118438482284546, + "num_tokens": 316460577.0, + "step": 12419 + }, + { + "epoch": 1.3639358664616736, + "grad_norm": 2.2337613105773926, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7283534407615662, + "num_tokens": 316486942.0, + "step": 12420 + }, + { + "epoch": 1.3640456841642874, + "grad_norm": 2.1664373874664307, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7224475145339966, + "num_tokens": 316514273.0, + "step": 12421 + }, + { + "epoch": 1.364155501866901, + "grad_norm": 2.0299768447875977, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7052420377731323, + "num_tokens": 316545380.0, + "step": 12422 + }, + { + "epoch": 1.3642653195695145, + "grad_norm": 2.398587703704834, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7218958139419556, + "num_tokens": 316568245.0, + "step": 12423 + }, + { + "epoch": 1.3643751372721282, + "grad_norm": 2.152158498764038, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7155675292015076, + "num_tokens": 316595308.0, + "step": 12424 + }, + { + "epoch": 1.364484954974742, + "grad_norm": 2.7299513816833496, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7242960333824158, + "num_tokens": 316613619.0, + "step": 12425 + }, + { + "epoch": 1.3645947726773555, + "grad_norm": 2.333913803100586, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.697850227355957, + "num_tokens": 316640657.0, + "step": 12426 + }, + { + "epoch": 1.3647045903799693, + "grad_norm": 2.55867075920105, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7101376056671143, + "num_tokens": 316663840.0, + "step": 12427 + }, + { + "epoch": 1.3648144080825828, + "grad_norm": 2.1800930500030518, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6946049928665161, + "num_tokens": 316694019.0, + "step": 12428 + }, + { + "epoch": 1.3649242257851966, + "grad_norm": 2.511807441711426, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7281047701835632, + "num_tokens": 316716108.0, + "step": 12429 + }, + { + "epoch": 1.3650340434878103, + "grad_norm": 2.0142595767974854, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7445542216300964, + "num_tokens": 316743476.0, + "step": 12430 + }, + { + "epoch": 1.3651438611904239, + "grad_norm": 2.013012170791626, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7070791721343994, + "num_tokens": 316773899.0, + "step": 12431 + }, + { + "epoch": 1.3652536788930376, + "grad_norm": 2.2254858016967773, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.711733341217041, + "num_tokens": 316798109.0, + "step": 12432 + }, + { + "epoch": 1.3653634965956511, + "grad_norm": 2.230212926864624, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7271812558174133, + "num_tokens": 316822293.0, + "step": 12433 + }, + { + "epoch": 1.365473314298265, + "grad_norm": 2.2871999740600586, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7174340486526489, + "num_tokens": 316846251.0, + "step": 12434 + }, + { + "epoch": 1.3655831320008787, + "grad_norm": 2.2635130882263184, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7054927945137024, + "num_tokens": 316871730.0, + "step": 12435 + }, + { + "epoch": 1.3656929497034922, + "grad_norm": 2.2333922386169434, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7165558338165283, + "num_tokens": 316895892.0, + "step": 12436 + }, + { + "epoch": 1.3658027674061057, + "grad_norm": 2.2414979934692383, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.6888773441314697, + "num_tokens": 316923198.0, + "step": 12437 + }, + { + "epoch": 1.3659125851087195, + "grad_norm": 2.3192639350891113, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7377556562423706, + "num_tokens": 316947242.0, + "step": 12438 + }, + { + "epoch": 1.3660224028113332, + "grad_norm": 2.522412061691284, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7221542596817017, + "num_tokens": 316968837.0, + "step": 12439 + }, + { + "epoch": 1.3661322205139468, + "grad_norm": 2.3721675872802734, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7091233730316162, + "num_tokens": 316994200.0, + "step": 12440 + }, + { + "epoch": 1.3662420382165605, + "grad_norm": 2.395324945449829, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7158663272857666, + "num_tokens": 317018812.0, + "step": 12441 + }, + { + "epoch": 1.366351855919174, + "grad_norm": 2.1240804195404053, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.6994596719741821, + "num_tokens": 317048129.0, + "step": 12442 + }, + { + "epoch": 1.3664616736217878, + "grad_norm": 2.3613297939300537, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7367063164710999, + "num_tokens": 317070834.0, + "step": 12443 + }, + { + "epoch": 1.3665714913244016, + "grad_norm": 2.5638322830200195, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.722952127456665, + "num_tokens": 317090754.0, + "step": 12444 + }, + { + "epoch": 1.3666813090270151, + "grad_norm": 2.370002508163452, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7362868785858154, + "num_tokens": 317112788.0, + "step": 12445 + }, + { + "epoch": 1.3667911267296289, + "grad_norm": 2.259390115737915, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7232591509819031, + "num_tokens": 317137944.0, + "step": 12446 + }, + { + "epoch": 1.3669009444322424, + "grad_norm": 2.7093706130981445, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7098388075828552, + "num_tokens": 317157417.0, + "step": 12447 + }, + { + "epoch": 1.3670107621348562, + "grad_norm": 2.5879764556884766, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.745944619178772, + "num_tokens": 317176609.0, + "step": 12448 + }, + { + "epoch": 1.36712057983747, + "grad_norm": 2.289285182952881, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7298663854598999, + "num_tokens": 317199732.0, + "step": 12449 + }, + { + "epoch": 1.3672303975400835, + "grad_norm": 2.33915376663208, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.732300341129303, + "num_tokens": 317221144.0, + "step": 12450 + }, + { + "epoch": 1.367340215242697, + "grad_norm": 2.2643959522247314, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7252224087715149, + "num_tokens": 317246110.0, + "step": 12451 + }, + { + "epoch": 1.3674500329453108, + "grad_norm": 2.473177433013916, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7244135141372681, + "num_tokens": 317268257.0, + "step": 12452 + }, + { + "epoch": 1.3675598506479245, + "grad_norm": 2.172295093536377, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7053876519203186, + "num_tokens": 317295867.0, + "step": 12453 + }, + { + "epoch": 1.367669668350538, + "grad_norm": 2.0540473461151123, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7230604290962219, + "num_tokens": 317324988.0, + "step": 12454 + }, + { + "epoch": 1.3677794860531518, + "grad_norm": 2.3783586025238037, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7187497019767761, + "num_tokens": 317348356.0, + "step": 12455 + }, + { + "epoch": 1.3678893037557653, + "grad_norm": 2.070296287536621, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.6968838572502136, + "num_tokens": 317380938.0, + "step": 12456 + }, + { + "epoch": 1.367999121458379, + "grad_norm": 2.104292869567871, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7139865756034851, + "num_tokens": 317411168.0, + "step": 12457 + }, + { + "epoch": 1.3681089391609929, + "grad_norm": 2.0760183334350586, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.729529857635498, + "num_tokens": 317440731.0, + "step": 12458 + }, + { + "epoch": 1.3682187568636064, + "grad_norm": 2.292931318283081, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7079780697822571, + "num_tokens": 317464243.0, + "step": 12459 + }, + { + "epoch": 1.3683285745662201, + "grad_norm": 2.3533003330230713, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7356730103492737, + "num_tokens": 317488034.0, + "step": 12460 + }, + { + "epoch": 1.3684383922688337, + "grad_norm": 2.1584508419036865, + "learning_rate": 1e-06, + "loss": 1.0546, + "mean_token_accuracy": 0.6797969341278076, + "num_tokens": 317515961.0, + "step": 12461 + }, + { + "epoch": 1.3685482099714474, + "grad_norm": 2.4432098865509033, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7143715023994446, + "num_tokens": 317539282.0, + "step": 12462 + }, + { + "epoch": 1.3686580276740612, + "grad_norm": 2.109644889831543, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7236989736557007, + "num_tokens": 317565942.0, + "step": 12463 + }, + { + "epoch": 1.3687678453766747, + "grad_norm": 2.1567575931549072, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.721152663230896, + "num_tokens": 317592894.0, + "step": 12464 + }, + { + "epoch": 1.3688776630792883, + "grad_norm": 2.290032386779785, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7114313840866089, + "num_tokens": 317618568.0, + "step": 12465 + }, + { + "epoch": 1.368987480781902, + "grad_norm": 2.703619956970215, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7287728786468506, + "num_tokens": 317637285.0, + "step": 12466 + }, + { + "epoch": 1.3690972984845158, + "grad_norm": 2.265409469604492, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.726884663105011, + "num_tokens": 317660475.0, + "step": 12467 + }, + { + "epoch": 1.3692071161871293, + "grad_norm": 2.2924344539642334, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7016794085502625, + "num_tokens": 317686519.0, + "step": 12468 + }, + { + "epoch": 1.369316933889743, + "grad_norm": 2.3489136695861816, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7340501546859741, + "num_tokens": 317709991.0, + "step": 12469 + }, + { + "epoch": 1.3694267515923566, + "grad_norm": 2.020052433013916, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7154357433319092, + "num_tokens": 317744056.0, + "step": 12470 + }, + { + "epoch": 1.3695365692949704, + "grad_norm": 2.165264844894409, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7145352363586426, + "num_tokens": 317771619.0, + "step": 12471 + }, + { + "epoch": 1.3696463869975841, + "grad_norm": 2.1490085124969482, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.72109055519104, + "num_tokens": 317796983.0, + "step": 12472 + }, + { + "epoch": 1.3697562047001977, + "grad_norm": 2.1686251163482666, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7218942642211914, + "num_tokens": 317823960.0, + "step": 12473 + }, + { + "epoch": 1.3698660224028114, + "grad_norm": 2.330965280532837, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7086395025253296, + "num_tokens": 317849364.0, + "step": 12474 + }, + { + "epoch": 1.369975840105425, + "grad_norm": 2.082305669784546, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7171896696090698, + "num_tokens": 317876199.0, + "step": 12475 + }, + { + "epoch": 1.3700856578080387, + "grad_norm": 2.3843295574188232, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7305233478546143, + "num_tokens": 317899229.0, + "step": 12476 + }, + { + "epoch": 1.3701954755106522, + "grad_norm": 2.3007843494415283, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.717355489730835, + "num_tokens": 317923940.0, + "step": 12477 + }, + { + "epoch": 1.370305293213266, + "grad_norm": 2.0413849353790283, + "learning_rate": 1e-06, + "loss": 0.8029, + "mean_token_accuracy": 0.7378700971603394, + "num_tokens": 317952678.0, + "step": 12478 + }, + { + "epoch": 1.3704151109158795, + "grad_norm": 2.2681069374084473, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7167032361030579, + "num_tokens": 317976935.0, + "step": 12479 + }, + { + "epoch": 1.3705249286184933, + "grad_norm": 2.305401086807251, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7373311519622803, + "num_tokens": 318000072.0, + "step": 12480 + }, + { + "epoch": 1.370634746321107, + "grad_norm": 2.313911199569702, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7225363254547119, + "num_tokens": 318022828.0, + "step": 12481 + }, + { + "epoch": 1.3707445640237206, + "grad_norm": 2.166095495223999, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7131592035293579, + "num_tokens": 318050091.0, + "step": 12482 + }, + { + "epoch": 1.3708543817263343, + "grad_norm": 2.2291321754455566, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7180642485618591, + "num_tokens": 318077345.0, + "step": 12483 + }, + { + "epoch": 1.3709641994289479, + "grad_norm": 2.0886709690093994, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7319645881652832, + "num_tokens": 318106344.0, + "step": 12484 + }, + { + "epoch": 1.3710740171315616, + "grad_norm": 2.318847894668579, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7124934792518616, + "num_tokens": 318130921.0, + "step": 12485 + }, + { + "epoch": 1.3711838348341754, + "grad_norm": 2.2132251262664795, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7429727911949158, + "num_tokens": 318156550.0, + "step": 12486 + }, + { + "epoch": 1.371293652536789, + "grad_norm": 2.118128538131714, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7274754047393799, + "num_tokens": 318183407.0, + "step": 12487 + }, + { + "epoch": 1.3714034702394025, + "grad_norm": 2.1929948329925537, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7131058573722839, + "num_tokens": 318210546.0, + "step": 12488 + }, + { + "epoch": 1.3715132879420162, + "grad_norm": 2.258305072784424, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7332223057746887, + "num_tokens": 318234800.0, + "step": 12489 + }, + { + "epoch": 1.37162310564463, + "grad_norm": 2.47839617729187, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7237681150436401, + "num_tokens": 318256923.0, + "step": 12490 + }, + { + "epoch": 1.3717329233472435, + "grad_norm": 2.3206775188446045, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7326997518539429, + "num_tokens": 318280894.0, + "step": 12491 + }, + { + "epoch": 1.3718427410498573, + "grad_norm": 2.218385696411133, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7399270534515381, + "num_tokens": 318305901.0, + "step": 12492 + }, + { + "epoch": 1.3719525587524708, + "grad_norm": 2.430798053741455, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7323608994483948, + "num_tokens": 318329533.0, + "step": 12493 + }, + { + "epoch": 1.3720623764550846, + "grad_norm": 2.4179494380950928, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7053250074386597, + "num_tokens": 318353109.0, + "step": 12494 + }, + { + "epoch": 1.3721721941576983, + "grad_norm": 2.4059202671051025, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7417037487030029, + "num_tokens": 318375012.0, + "step": 12495 + }, + { + "epoch": 1.3722820118603118, + "grad_norm": 2.308682918548584, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7202033996582031, + "num_tokens": 318398748.0, + "step": 12496 + }, + { + "epoch": 1.3723918295629256, + "grad_norm": 2.1457955837249756, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7141582369804382, + "num_tokens": 318428524.0, + "step": 12497 + }, + { + "epoch": 1.3725016472655391, + "grad_norm": 2.1315062046051025, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7270587682723999, + "num_tokens": 318454300.0, + "step": 12498 + }, + { + "epoch": 1.372611464968153, + "grad_norm": 2.2352025508880615, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.7333719730377197, + "num_tokens": 318478777.0, + "step": 12499 + }, + { + "epoch": 1.3727212826707667, + "grad_norm": 2.195732593536377, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7194240093231201, + "num_tokens": 318508290.0, + "step": 12500 + }, + { + "epoch": 1.3728311003733802, + "grad_norm": 2.7188093662261963, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7239522337913513, + "num_tokens": 318526672.0, + "step": 12501 + }, + { + "epoch": 1.3729409180759937, + "grad_norm": 2.4170212745666504, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7200438976287842, + "num_tokens": 318547882.0, + "step": 12502 + }, + { + "epoch": 1.3730507357786075, + "grad_norm": 2.423562526702881, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7200024127960205, + "num_tokens": 318569905.0, + "step": 12503 + }, + { + "epoch": 1.3731605534812212, + "grad_norm": 2.1109166145324707, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7119724750518799, + "num_tokens": 318600145.0, + "step": 12504 + }, + { + "epoch": 1.3732703711838348, + "grad_norm": 2.2254674434661865, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7075833082199097, + "num_tokens": 318627560.0, + "step": 12505 + }, + { + "epoch": 1.3733801888864485, + "grad_norm": 1.9954197406768799, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7220334410667419, + "num_tokens": 318659137.0, + "step": 12506 + }, + { + "epoch": 1.373490006589062, + "grad_norm": 2.430300712585449, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7188172936439514, + "num_tokens": 318681862.0, + "step": 12507 + }, + { + "epoch": 1.3735998242916758, + "grad_norm": 2.115239143371582, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7147035598754883, + "num_tokens": 318708723.0, + "step": 12508 + }, + { + "epoch": 1.3737096419942896, + "grad_norm": 2.197519302368164, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7414659857749939, + "num_tokens": 318734903.0, + "step": 12509 + }, + { + "epoch": 1.3738194596969031, + "grad_norm": 2.362520933151245, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7127147316932678, + "num_tokens": 318759103.0, + "step": 12510 + }, + { + "epoch": 1.3739292773995169, + "grad_norm": 2.4799299240112305, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7197282314300537, + "num_tokens": 318780984.0, + "step": 12511 + }, + { + "epoch": 1.3740390951021304, + "grad_norm": 2.0709471702575684, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.7048054933547974, + "num_tokens": 318810736.0, + "step": 12512 + }, + { + "epoch": 1.3741489128047442, + "grad_norm": 2.2456321716308594, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6854540109634399, + "num_tokens": 318837794.0, + "step": 12513 + }, + { + "epoch": 1.374258730507358, + "grad_norm": 2.071695327758789, + "learning_rate": 1e-06, + "loss": 1.0466, + "mean_token_accuracy": 0.6836249828338623, + "num_tokens": 318869211.0, + "step": 12514 + }, + { + "epoch": 1.3743685482099715, + "grad_norm": 2.2800939083099365, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7206294536590576, + "num_tokens": 318891762.0, + "step": 12515 + }, + { + "epoch": 1.374478365912585, + "grad_norm": 2.320561647415161, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7211756706237793, + "num_tokens": 318916488.0, + "step": 12516 + }, + { + "epoch": 1.3745881836151987, + "grad_norm": 2.1000092029571533, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7006862163543701, + "num_tokens": 318946060.0, + "step": 12517 + }, + { + "epoch": 1.3746980013178125, + "grad_norm": 2.496598720550537, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7252137064933777, + "num_tokens": 318967093.0, + "step": 12518 + }, + { + "epoch": 1.374807819020426, + "grad_norm": 2.1619510650634766, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7241287231445312, + "num_tokens": 318996331.0, + "step": 12519 + }, + { + "epoch": 1.3749176367230398, + "grad_norm": 2.023817539215088, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7181369662284851, + "num_tokens": 319026644.0, + "step": 12520 + }, + { + "epoch": 1.3750274544256533, + "grad_norm": 2.3304381370544434, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7340651750564575, + "num_tokens": 319049395.0, + "step": 12521 + }, + { + "epoch": 1.375137272128267, + "grad_norm": 2.328042507171631, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.726083517074585, + "num_tokens": 319071972.0, + "step": 12522 + }, + { + "epoch": 1.3752470898308808, + "grad_norm": 2.1588356494903564, + "learning_rate": 1e-06, + "loss": 1.0176, + "mean_token_accuracy": 0.7016753554344177, + "num_tokens": 319098532.0, + "step": 12523 + }, + { + "epoch": 1.3753569075334944, + "grad_norm": 2.4505131244659424, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.722342312335968, + "num_tokens": 319121750.0, + "step": 12524 + }, + { + "epoch": 1.3754667252361081, + "grad_norm": 2.235212802886963, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7448462247848511, + "num_tokens": 319146986.0, + "step": 12525 + }, + { + "epoch": 1.3755765429387217, + "grad_norm": 2.2380924224853516, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7049192190170288, + "num_tokens": 319172453.0, + "step": 12526 + }, + { + "epoch": 1.3756863606413354, + "grad_norm": 1.9975254535675049, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7538790702819824, + "num_tokens": 319201395.0, + "step": 12527 + }, + { + "epoch": 1.3757961783439492, + "grad_norm": 2.083832025527954, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7095862030982971, + "num_tokens": 319229803.0, + "step": 12528 + }, + { + "epoch": 1.3759059960465627, + "grad_norm": 2.4840946197509766, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7223145961761475, + "num_tokens": 319253560.0, + "step": 12529 + }, + { + "epoch": 1.3760158137491763, + "grad_norm": 2.352355480194092, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7377090454101562, + "num_tokens": 319276046.0, + "step": 12530 + }, + { + "epoch": 1.37612563145179, + "grad_norm": 2.2766897678375244, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7295087575912476, + "num_tokens": 319301681.0, + "step": 12531 + }, + { + "epoch": 1.3762354491544038, + "grad_norm": 2.340000867843628, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7395757436752319, + "num_tokens": 319325234.0, + "step": 12532 + }, + { + "epoch": 1.3763452668570173, + "grad_norm": 2.0529844760894775, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7022936344146729, + "num_tokens": 319356402.0, + "step": 12533 + }, + { + "epoch": 1.376455084559631, + "grad_norm": 2.0479507446289062, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7287932634353638, + "num_tokens": 319384702.0, + "step": 12534 + }, + { + "epoch": 1.3765649022622446, + "grad_norm": 2.318525552749634, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7336957454681396, + "num_tokens": 319407519.0, + "step": 12535 + }, + { + "epoch": 1.3766747199648584, + "grad_norm": 2.325011730194092, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7233647108078003, + "num_tokens": 319431223.0, + "step": 12536 + }, + { + "epoch": 1.376784537667472, + "grad_norm": 2.2405335903167725, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7109869718551636, + "num_tokens": 319457421.0, + "step": 12537 + }, + { + "epoch": 1.3768943553700856, + "grad_norm": 2.0612266063690186, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.709667980670929, + "num_tokens": 319485702.0, + "step": 12538 + }, + { + "epoch": 1.3770041730726994, + "grad_norm": 2.199864387512207, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.705662190914154, + "num_tokens": 319512739.0, + "step": 12539 + }, + { + "epoch": 1.377113990775313, + "grad_norm": 2.1916327476501465, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.7044723629951477, + "num_tokens": 319541022.0, + "step": 12540 + }, + { + "epoch": 1.3772238084779267, + "grad_norm": 2.3547677993774414, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7424509525299072, + "num_tokens": 319564674.0, + "step": 12541 + }, + { + "epoch": 1.3773336261805402, + "grad_norm": 2.134021520614624, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7685818672180176, + "num_tokens": 319589607.0, + "step": 12542 + }, + { + "epoch": 1.377443443883154, + "grad_norm": 2.1965160369873047, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7189021110534668, + "num_tokens": 319617643.0, + "step": 12543 + }, + { + "epoch": 1.3775532615857675, + "grad_norm": 2.3080830574035645, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7418613433837891, + "num_tokens": 319640143.0, + "step": 12544 + }, + { + "epoch": 1.3776630792883813, + "grad_norm": 2.4531726837158203, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7072697877883911, + "num_tokens": 319662287.0, + "step": 12545 + }, + { + "epoch": 1.377772896990995, + "grad_norm": 2.194567918777466, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7268732190132141, + "num_tokens": 319689291.0, + "step": 12546 + }, + { + "epoch": 1.3778827146936086, + "grad_norm": 2.0538618564605713, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.6979585886001587, + "num_tokens": 319720896.0, + "step": 12547 + }, + { + "epoch": 1.3779925323962223, + "grad_norm": 2.197021484375, + "learning_rate": 1e-06, + "loss": 0.7814, + "mean_token_accuracy": 0.7531206607818604, + "num_tokens": 319745602.0, + "step": 12548 + }, + { + "epoch": 1.3781023500988359, + "grad_norm": 2.432328939437866, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7238227725028992, + "num_tokens": 319766043.0, + "step": 12549 + }, + { + "epoch": 1.3782121678014496, + "grad_norm": 2.271491527557373, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.731049656867981, + "num_tokens": 319789744.0, + "step": 12550 + }, + { + "epoch": 1.3783219855040634, + "grad_norm": 2.423349380493164, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7086670398712158, + "num_tokens": 319815098.0, + "step": 12551 + }, + { + "epoch": 1.378431803206677, + "grad_norm": 2.403477668762207, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.7529714107513428, + "num_tokens": 319839095.0, + "step": 12552 + }, + { + "epoch": 1.3785416209092904, + "grad_norm": 2.48173189163208, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7114249467849731, + "num_tokens": 319863581.0, + "step": 12553 + }, + { + "epoch": 1.3786514386119042, + "grad_norm": 2.2637360095977783, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.749602198600769, + "num_tokens": 319886404.0, + "step": 12554 + }, + { + "epoch": 1.378761256314518, + "grad_norm": 2.8780155181884766, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7372949719429016, + "num_tokens": 319904125.0, + "step": 12555 + }, + { + "epoch": 1.3788710740171315, + "grad_norm": 2.464292287826538, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7170701026916504, + "num_tokens": 319927141.0, + "step": 12556 + }, + { + "epoch": 1.3789808917197452, + "grad_norm": 2.27174711227417, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.734953761100769, + "num_tokens": 319952102.0, + "step": 12557 + }, + { + "epoch": 1.3790907094223588, + "grad_norm": 2.388681650161743, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7100120782852173, + "num_tokens": 319976351.0, + "step": 12558 + }, + { + "epoch": 1.3792005271249725, + "grad_norm": 2.2364461421966553, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7411767840385437, + "num_tokens": 320001381.0, + "step": 12559 + }, + { + "epoch": 1.3793103448275863, + "grad_norm": 2.3539018630981445, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7234774231910706, + "num_tokens": 320026184.0, + "step": 12560 + }, + { + "epoch": 1.3794201625301998, + "grad_norm": 2.4710333347320557, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7323490381240845, + "num_tokens": 320047407.0, + "step": 12561 + }, + { + "epoch": 1.3795299802328136, + "grad_norm": 2.3154289722442627, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7235958576202393, + "num_tokens": 320071841.0, + "step": 12562 + }, + { + "epoch": 1.3796397979354271, + "grad_norm": 2.1659469604492188, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.6999151706695557, + "num_tokens": 320101441.0, + "step": 12563 + }, + { + "epoch": 1.3797496156380409, + "grad_norm": 2.100379467010498, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7008100748062134, + "num_tokens": 320131160.0, + "step": 12564 + }, + { + "epoch": 1.3798594333406546, + "grad_norm": 2.0039539337158203, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7513388395309448, + "num_tokens": 320160620.0, + "step": 12565 + }, + { + "epoch": 1.3799692510432682, + "grad_norm": 2.1117186546325684, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7108250856399536, + "num_tokens": 320189540.0, + "step": 12566 + }, + { + "epoch": 1.3800790687458817, + "grad_norm": 2.4942657947540283, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7384268641471863, + "num_tokens": 320210676.0, + "step": 12567 + }, + { + "epoch": 1.3801888864484955, + "grad_norm": 2.1786255836486816, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.6994887590408325, + "num_tokens": 320238567.0, + "step": 12568 + }, + { + "epoch": 1.3802987041511092, + "grad_norm": 2.240703582763672, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7050950527191162, + "num_tokens": 320263815.0, + "step": 12569 + }, + { + "epoch": 1.3804085218537228, + "grad_norm": 1.960399866104126, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7397168874740601, + "num_tokens": 320296519.0, + "step": 12570 + }, + { + "epoch": 1.3805183395563365, + "grad_norm": 2.30001163482666, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7349465489387512, + "num_tokens": 320320117.0, + "step": 12571 + }, + { + "epoch": 1.38062815725895, + "grad_norm": 2.0997204780578613, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7269940972328186, + "num_tokens": 320348738.0, + "step": 12572 + }, + { + "epoch": 1.3807379749615638, + "grad_norm": 2.182278871536255, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7171956300735474, + "num_tokens": 320376831.0, + "step": 12573 + }, + { + "epoch": 1.3808477926641776, + "grad_norm": 2.2788546085357666, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7176568508148193, + "num_tokens": 320402832.0, + "step": 12574 + }, + { + "epoch": 1.380957610366791, + "grad_norm": 2.31535005569458, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7224607467651367, + "num_tokens": 320426521.0, + "step": 12575 + }, + { + "epoch": 1.3810674280694049, + "grad_norm": 2.2719147205352783, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7223474383354187, + "num_tokens": 320452566.0, + "step": 12576 + }, + { + "epoch": 1.3811772457720184, + "grad_norm": 2.1398377418518066, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7314394116401672, + "num_tokens": 320480563.0, + "step": 12577 + }, + { + "epoch": 1.3812870634746321, + "grad_norm": 2.315556287765503, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7173446416854858, + "num_tokens": 320506115.0, + "step": 12578 + }, + { + "epoch": 1.381396881177246, + "grad_norm": 2.289313316345215, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7055873870849609, + "num_tokens": 320532092.0, + "step": 12579 + }, + { + "epoch": 1.3815066988798594, + "grad_norm": 2.16001558303833, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.705559253692627, + "num_tokens": 320561403.0, + "step": 12580 + }, + { + "epoch": 1.381616516582473, + "grad_norm": 2.5832576751708984, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7355654239654541, + "num_tokens": 320582244.0, + "step": 12581 + }, + { + "epoch": 1.3817263342850867, + "grad_norm": 2.151912212371826, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7005912065505981, + "num_tokens": 320613116.0, + "step": 12582 + }, + { + "epoch": 1.3818361519877005, + "grad_norm": 2.1433422565460205, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7359818816184998, + "num_tokens": 320641961.0, + "step": 12583 + }, + { + "epoch": 1.381945969690314, + "grad_norm": 2.092116594314575, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7195987701416016, + "num_tokens": 320670467.0, + "step": 12584 + }, + { + "epoch": 1.3820557873929278, + "grad_norm": 2.3761959075927734, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.718242883682251, + "num_tokens": 320693505.0, + "step": 12585 + }, + { + "epoch": 1.3821656050955413, + "grad_norm": 2.2454116344451904, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7256642580032349, + "num_tokens": 320719018.0, + "step": 12586 + }, + { + "epoch": 1.382275422798155, + "grad_norm": 2.185528516769409, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7073574662208557, + "num_tokens": 320747300.0, + "step": 12587 + }, + { + "epoch": 1.3823852405007688, + "grad_norm": 2.5054876804351807, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7133254408836365, + "num_tokens": 320771383.0, + "step": 12588 + }, + { + "epoch": 1.3824950582033824, + "grad_norm": 2.388345718383789, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7304067015647888, + "num_tokens": 320794218.0, + "step": 12589 + }, + { + "epoch": 1.3826048759059961, + "grad_norm": 2.0985794067382812, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7222739458084106, + "num_tokens": 320822193.0, + "step": 12590 + }, + { + "epoch": 1.3827146936086097, + "grad_norm": 2.328083038330078, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7065073847770691, + "num_tokens": 320846801.0, + "step": 12591 + }, + { + "epoch": 1.3828245113112234, + "grad_norm": 1.8850849866867065, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7234885692596436, + "num_tokens": 320876438.0, + "step": 12592 + }, + { + "epoch": 1.382934329013837, + "grad_norm": 2.1782753467559814, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7481048107147217, + "num_tokens": 320900676.0, + "step": 12593 + }, + { + "epoch": 1.3830441467164507, + "grad_norm": 2.658794641494751, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7160521149635315, + "num_tokens": 320920541.0, + "step": 12594 + }, + { + "epoch": 1.3831539644190642, + "grad_norm": 2.638051986694336, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7241792678833008, + "num_tokens": 320940733.0, + "step": 12595 + }, + { + "epoch": 1.383263782121678, + "grad_norm": 2.6139285564422607, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7360402941703796, + "num_tokens": 320959591.0, + "step": 12596 + }, + { + "epoch": 1.3833735998242918, + "grad_norm": 2.1820876598358154, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7034933567047119, + "num_tokens": 320989051.0, + "step": 12597 + }, + { + "epoch": 1.3834834175269053, + "grad_norm": 2.2020158767700195, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7241514325141907, + "num_tokens": 321014115.0, + "step": 12598 + }, + { + "epoch": 1.383593235229519, + "grad_norm": 2.221771717071533, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7079814672470093, + "num_tokens": 321039121.0, + "step": 12599 + }, + { + "epoch": 1.3837030529321326, + "grad_norm": 2.3531172275543213, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7185958027839661, + "num_tokens": 321062641.0, + "step": 12600 + }, + { + "epoch": 1.3838128706347463, + "grad_norm": 2.020296812057495, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.735121488571167, + "num_tokens": 321093125.0, + "step": 12601 + }, + { + "epoch": 1.38392268833736, + "grad_norm": 2.413555860519409, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.6977276802062988, + "num_tokens": 321115950.0, + "step": 12602 + }, + { + "epoch": 1.3840325060399736, + "grad_norm": 2.1516916751861572, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7042101621627808, + "num_tokens": 321143572.0, + "step": 12603 + }, + { + "epoch": 1.3841423237425874, + "grad_norm": 2.1728830337524414, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7291102409362793, + "num_tokens": 321170151.0, + "step": 12604 + }, + { + "epoch": 1.384252141445201, + "grad_norm": 2.1587073802948, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.7549009919166565, + "num_tokens": 321195151.0, + "step": 12605 + }, + { + "epoch": 1.3843619591478147, + "grad_norm": 2.4533421993255615, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7409741878509521, + "num_tokens": 321216011.0, + "step": 12606 + }, + { + "epoch": 1.3844717768504282, + "grad_norm": 2.0571820735931396, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7241313457489014, + "num_tokens": 321245975.0, + "step": 12607 + }, + { + "epoch": 1.384581594553042, + "grad_norm": 2.3059043884277344, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6926978230476379, + "num_tokens": 321272696.0, + "step": 12608 + }, + { + "epoch": 1.3846914122556555, + "grad_norm": 2.0089664459228516, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7069835066795349, + "num_tokens": 321305002.0, + "step": 12609 + }, + { + "epoch": 1.3848012299582693, + "grad_norm": 2.1133525371551514, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.685683012008667, + "num_tokens": 321334673.0, + "step": 12610 + }, + { + "epoch": 1.384911047660883, + "grad_norm": 2.023484945297241, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.716665506362915, + "num_tokens": 321363662.0, + "step": 12611 + }, + { + "epoch": 1.3850208653634966, + "grad_norm": 2.483760118484497, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7412663102149963, + "num_tokens": 321384538.0, + "step": 12612 + }, + { + "epoch": 1.3851306830661103, + "grad_norm": 2.214988946914673, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.7070375680923462, + "num_tokens": 321413682.0, + "step": 12613 + }, + { + "epoch": 1.3852405007687238, + "grad_norm": 2.375481605529785, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7189512252807617, + "num_tokens": 321437219.0, + "step": 12614 + }, + { + "epoch": 1.3853503184713376, + "grad_norm": 2.048297166824341, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.6988284587860107, + "num_tokens": 321469221.0, + "step": 12615 + }, + { + "epoch": 1.3854601361739514, + "grad_norm": 2.2909979820251465, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.714730978012085, + "num_tokens": 321494829.0, + "step": 12616 + }, + { + "epoch": 1.385569953876565, + "grad_norm": 2.437300205230713, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7032941579818726, + "num_tokens": 321518860.0, + "step": 12617 + }, + { + "epoch": 1.3856797715791784, + "grad_norm": 2.1434290409088135, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7094733119010925, + "num_tokens": 321547400.0, + "step": 12618 + }, + { + "epoch": 1.3857895892817922, + "grad_norm": 2.356553792953491, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7238311767578125, + "num_tokens": 321570136.0, + "step": 12619 + }, + { + "epoch": 1.385899406984406, + "grad_norm": 2.2350521087646484, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7167239189147949, + "num_tokens": 321595403.0, + "step": 12620 + }, + { + "epoch": 1.3860092246870195, + "grad_norm": 1.9187264442443848, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7059831619262695, + "num_tokens": 321628049.0, + "step": 12621 + }, + { + "epoch": 1.3861190423896332, + "grad_norm": 2.4161274433135986, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7092193961143494, + "num_tokens": 321652049.0, + "step": 12622 + }, + { + "epoch": 1.3862288600922468, + "grad_norm": 2.278316020965576, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7292153835296631, + "num_tokens": 321679159.0, + "step": 12623 + }, + { + "epoch": 1.3863386777948605, + "grad_norm": 2.236729383468628, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7012260556221008, + "num_tokens": 321705755.0, + "step": 12624 + }, + { + "epoch": 1.3864484954974743, + "grad_norm": 2.179377794265747, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7113697528839111, + "num_tokens": 321733342.0, + "step": 12625 + }, + { + "epoch": 1.3865583132000878, + "grad_norm": 2.4242537021636963, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7361251711845398, + "num_tokens": 321753671.0, + "step": 12626 + }, + { + "epoch": 1.3866681309027016, + "grad_norm": 2.176342010498047, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6974474191665649, + "num_tokens": 321783735.0, + "step": 12627 + }, + { + "epoch": 1.3867779486053151, + "grad_norm": 2.4539051055908203, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7488553524017334, + "num_tokens": 321805979.0, + "step": 12628 + }, + { + "epoch": 1.3868877663079289, + "grad_norm": 2.251654624938965, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.7354862689971924, + "num_tokens": 321830291.0, + "step": 12629 + }, + { + "epoch": 1.3869975840105426, + "grad_norm": 2.081449031829834, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.732454776763916, + "num_tokens": 321856687.0, + "step": 12630 + }, + { + "epoch": 1.3871074017131562, + "grad_norm": 2.120861291885376, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7286605834960938, + "num_tokens": 321884500.0, + "step": 12631 + }, + { + "epoch": 1.3872172194157697, + "grad_norm": 2.0466020107269287, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7171881198883057, + "num_tokens": 321913728.0, + "step": 12632 + }, + { + "epoch": 1.3873270371183835, + "grad_norm": 2.092132329940796, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7001709342002869, + "num_tokens": 321944892.0, + "step": 12633 + }, + { + "epoch": 1.3874368548209972, + "grad_norm": 2.264925956726074, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7254688739776611, + "num_tokens": 321969926.0, + "step": 12634 + }, + { + "epoch": 1.3875466725236107, + "grad_norm": 2.4590513706207275, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7118188142776489, + "num_tokens": 321992638.0, + "step": 12635 + }, + { + "epoch": 1.3876564902262245, + "grad_norm": 2.2603282928466797, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.681095540523529, + "num_tokens": 322020092.0, + "step": 12636 + }, + { + "epoch": 1.387766307928838, + "grad_norm": 2.646095037460327, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7204270362854004, + "num_tokens": 322040657.0, + "step": 12637 + }, + { + "epoch": 1.3878761256314518, + "grad_norm": 2.185697317123413, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.712246298789978, + "num_tokens": 322070418.0, + "step": 12638 + }, + { + "epoch": 1.3879859433340656, + "grad_norm": 2.242814302444458, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7277761697769165, + "num_tokens": 322094377.0, + "step": 12639 + }, + { + "epoch": 1.388095761036679, + "grad_norm": 2.1291751861572266, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7159789204597473, + "num_tokens": 322122650.0, + "step": 12640 + }, + { + "epoch": 1.3882055787392928, + "grad_norm": 2.4449496269226074, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7472341060638428, + "num_tokens": 322143590.0, + "step": 12641 + }, + { + "epoch": 1.3883153964419064, + "grad_norm": 2.403139114379883, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7427546977996826, + "num_tokens": 322165546.0, + "step": 12642 + }, + { + "epoch": 1.3884252141445201, + "grad_norm": 2.3968420028686523, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7187386155128479, + "num_tokens": 322188448.0, + "step": 12643 + }, + { + "epoch": 1.388535031847134, + "grad_norm": 2.6695468425750732, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7236455678939819, + "num_tokens": 322208784.0, + "step": 12644 + }, + { + "epoch": 1.3886448495497474, + "grad_norm": 2.1618239879608154, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7367634773254395, + "num_tokens": 322236836.0, + "step": 12645 + }, + { + "epoch": 1.388754667252361, + "grad_norm": 2.2816832065582275, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7125450372695923, + "num_tokens": 322263073.0, + "step": 12646 + }, + { + "epoch": 1.3888644849549747, + "grad_norm": 2.425328493118286, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7222490906715393, + "num_tokens": 322285089.0, + "step": 12647 + }, + { + "epoch": 1.3889743026575885, + "grad_norm": 2.3370168209075928, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7247610092163086, + "num_tokens": 322309092.0, + "step": 12648 + }, + { + "epoch": 1.389084120360202, + "grad_norm": 2.6302154064178467, + "learning_rate": 1e-06, + "loss": 0.8069, + "mean_token_accuracy": 0.7437158823013306, + "num_tokens": 322329074.0, + "step": 12649 + }, + { + "epoch": 1.3891939380628158, + "grad_norm": 2.185279607772827, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6971431970596313, + "num_tokens": 322357371.0, + "step": 12650 + }, + { + "epoch": 1.3893037557654293, + "grad_norm": 2.0176780223846436, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7160152196884155, + "num_tokens": 322389255.0, + "step": 12651 + }, + { + "epoch": 1.389413573468043, + "grad_norm": 2.2313058376312256, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7171953916549683, + "num_tokens": 322417484.0, + "step": 12652 + }, + { + "epoch": 1.3895233911706568, + "grad_norm": 2.029956817626953, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7059563398361206, + "num_tokens": 322447383.0, + "step": 12653 + }, + { + "epoch": 1.3896332088732704, + "grad_norm": 2.324777841567993, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7152152061462402, + "num_tokens": 322473142.0, + "step": 12654 + }, + { + "epoch": 1.389743026575884, + "grad_norm": 2.009625196456909, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.704675555229187, + "num_tokens": 322505323.0, + "step": 12655 + }, + { + "epoch": 1.3898528442784976, + "grad_norm": 2.6244049072265625, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7186484932899475, + "num_tokens": 322525106.0, + "step": 12656 + }, + { + "epoch": 1.3899626619811114, + "grad_norm": 2.063190460205078, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7252400517463684, + "num_tokens": 322552224.0, + "step": 12657 + }, + { + "epoch": 1.390072479683725, + "grad_norm": 2.1193249225616455, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7084696292877197, + "num_tokens": 322583253.0, + "step": 12658 + }, + { + "epoch": 1.3901822973863387, + "grad_norm": 2.335479974746704, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7243363857269287, + "num_tokens": 322605880.0, + "step": 12659 + }, + { + "epoch": 1.3902921150889522, + "grad_norm": 2.3498547077178955, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.7456963062286377, + "num_tokens": 322628978.0, + "step": 12660 + }, + { + "epoch": 1.390401932791566, + "grad_norm": 2.443096399307251, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7323333621025085, + "num_tokens": 322652387.0, + "step": 12661 + }, + { + "epoch": 1.3905117504941797, + "grad_norm": 2.2457892894744873, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7434098720550537, + "num_tokens": 322675404.0, + "step": 12662 + }, + { + "epoch": 1.3906215681967933, + "grad_norm": 2.3700013160705566, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.733309268951416, + "num_tokens": 322699618.0, + "step": 12663 + }, + { + "epoch": 1.390731385899407, + "grad_norm": 2.0823402404785156, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7097582817077637, + "num_tokens": 322729272.0, + "step": 12664 + }, + { + "epoch": 1.3908412036020206, + "grad_norm": 2.499972105026245, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7144064903259277, + "num_tokens": 322751566.0, + "step": 12665 + }, + { + "epoch": 1.3909510213046343, + "grad_norm": 2.326411247253418, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7285394668579102, + "num_tokens": 322774343.0, + "step": 12666 + }, + { + "epoch": 1.391060839007248, + "grad_norm": 2.1491541862487793, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7267286777496338, + "num_tokens": 322803274.0, + "step": 12667 + }, + { + "epoch": 1.3911706567098616, + "grad_norm": 2.1607248783111572, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7398554682731628, + "num_tokens": 322828630.0, + "step": 12668 + }, + { + "epoch": 1.3912804744124752, + "grad_norm": 2.2049667835235596, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.6985850930213928, + "num_tokens": 322854819.0, + "step": 12669 + }, + { + "epoch": 1.391390292115089, + "grad_norm": 1.833442211151123, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.702099084854126, + "num_tokens": 322887593.0, + "step": 12670 + }, + { + "epoch": 1.3915001098177027, + "grad_norm": 2.193241596221924, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7291767597198486, + "num_tokens": 322914451.0, + "step": 12671 + }, + { + "epoch": 1.3916099275203162, + "grad_norm": 2.4276657104492188, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7214435935020447, + "num_tokens": 322938966.0, + "step": 12672 + }, + { + "epoch": 1.39171974522293, + "grad_norm": 2.1104736328125, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7151092886924744, + "num_tokens": 322965559.0, + "step": 12673 + }, + { + "epoch": 1.3918295629255435, + "grad_norm": 2.412457227706909, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7427473068237305, + "num_tokens": 322985833.0, + "step": 12674 + }, + { + "epoch": 1.3919393806281573, + "grad_norm": 2.1579761505126953, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7334998846054077, + "num_tokens": 323010883.0, + "step": 12675 + }, + { + "epoch": 1.392049198330771, + "grad_norm": 2.1794731616973877, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7381875514984131, + "num_tokens": 323036853.0, + "step": 12676 + }, + { + "epoch": 1.3921590160333845, + "grad_norm": 2.283059597015381, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7220617532730103, + "num_tokens": 323062129.0, + "step": 12677 + }, + { + "epoch": 1.3922688337359983, + "grad_norm": 2.4673829078674316, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.715529203414917, + "num_tokens": 323083709.0, + "step": 12678 + }, + { + "epoch": 1.3923786514386118, + "grad_norm": 2.364819288253784, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7136719226837158, + "num_tokens": 323107041.0, + "step": 12679 + }, + { + "epoch": 1.3924884691412256, + "grad_norm": 2.258695125579834, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7408673763275146, + "num_tokens": 323129809.0, + "step": 12680 + }, + { + "epoch": 1.3925982868438394, + "grad_norm": 2.218583822250366, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.730889618396759, + "num_tokens": 323155678.0, + "step": 12681 + }, + { + "epoch": 1.3927081045464529, + "grad_norm": 2.3316688537597656, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7392183542251587, + "num_tokens": 323179085.0, + "step": 12682 + }, + { + "epoch": 1.3928179222490664, + "grad_norm": 2.4343008995056152, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7367148399353027, + "num_tokens": 323199896.0, + "step": 12683 + }, + { + "epoch": 1.3929277399516802, + "grad_norm": 2.2247588634490967, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7289398908615112, + "num_tokens": 323225014.0, + "step": 12684 + }, + { + "epoch": 1.393037557654294, + "grad_norm": 2.300189971923828, + "learning_rate": 1e-06, + "loss": 0.8213, + "mean_token_accuracy": 0.7524008750915527, + "num_tokens": 323246999.0, + "step": 12685 + }, + { + "epoch": 1.3931473753569075, + "grad_norm": 2.072906732559204, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.706868052482605, + "num_tokens": 323277798.0, + "step": 12686 + }, + { + "epoch": 1.3932571930595212, + "grad_norm": 2.289588212966919, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7272989749908447, + "num_tokens": 323304302.0, + "step": 12687 + }, + { + "epoch": 1.3933670107621348, + "grad_norm": 2.7493343353271484, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7396561503410339, + "num_tokens": 323322964.0, + "step": 12688 + }, + { + "epoch": 1.3934768284647485, + "grad_norm": 2.1635565757751465, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7124819159507751, + "num_tokens": 323351762.0, + "step": 12689 + }, + { + "epoch": 1.3935866461673623, + "grad_norm": 2.502875804901123, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7359803915023804, + "num_tokens": 323373771.0, + "step": 12690 + }, + { + "epoch": 1.3936964638699758, + "grad_norm": 2.2881252765655518, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7131828665733337, + "num_tokens": 323400431.0, + "step": 12691 + }, + { + "epoch": 1.3938062815725896, + "grad_norm": 2.305042028427124, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7310726642608643, + "num_tokens": 323423648.0, + "step": 12692 + }, + { + "epoch": 1.393916099275203, + "grad_norm": 2.257031202316284, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7394871711730957, + "num_tokens": 323448857.0, + "step": 12693 + }, + { + "epoch": 1.3940259169778169, + "grad_norm": 2.1208913326263428, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7321755886077881, + "num_tokens": 323475914.0, + "step": 12694 + }, + { + "epoch": 1.3941357346804306, + "grad_norm": 2.187119245529175, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7248871326446533, + "num_tokens": 323501169.0, + "step": 12695 + }, + { + "epoch": 1.3942455523830442, + "grad_norm": 2.173389196395874, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.720258355140686, + "num_tokens": 323527567.0, + "step": 12696 + }, + { + "epoch": 1.3943553700856577, + "grad_norm": 2.6127943992614746, + "learning_rate": 1e-06, + "loss": 0.8176, + "mean_token_accuracy": 0.744565486907959, + "num_tokens": 323547976.0, + "step": 12697 + }, + { + "epoch": 1.3944651877882714, + "grad_norm": 2.168689012527466, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7163922190666199, + "num_tokens": 323574938.0, + "step": 12698 + }, + { + "epoch": 1.3945750054908852, + "grad_norm": 2.729671001434326, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7254641056060791, + "num_tokens": 323593576.0, + "step": 12699 + }, + { + "epoch": 1.3946848231934987, + "grad_norm": 2.341080904006958, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7242188453674316, + "num_tokens": 323616768.0, + "step": 12700 + }, + { + "epoch": 1.3947946408961125, + "grad_norm": 2.719980239868164, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7346270084381104, + "num_tokens": 323636394.0, + "step": 12701 + }, + { + "epoch": 1.394904458598726, + "grad_norm": 2.809539318084717, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7336730360984802, + "num_tokens": 323655150.0, + "step": 12702 + }, + { + "epoch": 1.3950142763013398, + "grad_norm": 1.995561122894287, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.6865760087966919, + "num_tokens": 323687781.0, + "step": 12703 + }, + { + "epoch": 1.3951240940039535, + "grad_norm": 2.131913185119629, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7078955173492432, + "num_tokens": 323717711.0, + "step": 12704 + }, + { + "epoch": 1.395233911706567, + "grad_norm": 2.2647364139556885, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7452126741409302, + "num_tokens": 323741875.0, + "step": 12705 + }, + { + "epoch": 1.3953437294091808, + "grad_norm": 2.101620674133301, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7295824289321899, + "num_tokens": 323767432.0, + "step": 12706 + }, + { + "epoch": 1.3954535471117944, + "grad_norm": 2.0849504470825195, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7008447647094727, + "num_tokens": 323795823.0, + "step": 12707 + }, + { + "epoch": 1.3955633648144081, + "grad_norm": 2.3905770778656006, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7196836471557617, + "num_tokens": 323821327.0, + "step": 12708 + }, + { + "epoch": 1.3956731825170219, + "grad_norm": 2.1311686038970947, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7039501070976257, + "num_tokens": 323849813.0, + "step": 12709 + }, + { + "epoch": 1.3957830002196354, + "grad_norm": 2.1545915603637695, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7331962585449219, + "num_tokens": 323877226.0, + "step": 12710 + }, + { + "epoch": 1.395892817922249, + "grad_norm": 2.3058698177337646, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7205929160118103, + "num_tokens": 323902632.0, + "step": 12711 + }, + { + "epoch": 1.3960026356248627, + "grad_norm": 2.1259188652038574, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.6950556635856628, + "num_tokens": 323932121.0, + "step": 12712 + }, + { + "epoch": 1.3961124533274765, + "grad_norm": 2.007227659225464, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7035701274871826, + "num_tokens": 323961714.0, + "step": 12713 + }, + { + "epoch": 1.39622227103009, + "grad_norm": 2.107956647872925, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7162481546401978, + "num_tokens": 323992315.0, + "step": 12714 + }, + { + "epoch": 1.3963320887327038, + "grad_norm": 2.0503103733062744, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7018613815307617, + "num_tokens": 324024126.0, + "step": 12715 + }, + { + "epoch": 1.3964419064353173, + "grad_norm": 2.572305917739868, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7398525476455688, + "num_tokens": 324043957.0, + "step": 12716 + }, + { + "epoch": 1.396551724137931, + "grad_norm": 2.3209798336029053, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7159978151321411, + "num_tokens": 324069787.0, + "step": 12717 + }, + { + "epoch": 1.3966615418405448, + "grad_norm": 2.1819264888763428, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.699002742767334, + "num_tokens": 324098754.0, + "step": 12718 + }, + { + "epoch": 1.3967713595431583, + "grad_norm": 2.1303820610046387, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7037023901939392, + "num_tokens": 324126763.0, + "step": 12719 + }, + { + "epoch": 1.396881177245772, + "grad_norm": 2.1947379112243652, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7169091701507568, + "num_tokens": 324152718.0, + "step": 12720 + }, + { + "epoch": 1.3969909949483856, + "grad_norm": 1.9564313888549805, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7242356538772583, + "num_tokens": 324184058.0, + "step": 12721 + }, + { + "epoch": 1.3971008126509994, + "grad_norm": 2.2594690322875977, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7410916090011597, + "num_tokens": 324210160.0, + "step": 12722 + }, + { + "epoch": 1.397210630353613, + "grad_norm": 2.2612786293029785, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7230855226516724, + "num_tokens": 324235478.0, + "step": 12723 + }, + { + "epoch": 1.3973204480562267, + "grad_norm": 2.0173890590667725, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7183817625045776, + "num_tokens": 324264687.0, + "step": 12724 + }, + { + "epoch": 1.3974302657588402, + "grad_norm": 2.1413180828094482, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7148908376693726, + "num_tokens": 324292793.0, + "step": 12725 + }, + { + "epoch": 1.397540083461454, + "grad_norm": 2.0921287536621094, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.711410403251648, + "num_tokens": 324323213.0, + "step": 12726 + }, + { + "epoch": 1.3976499011640677, + "grad_norm": 2.3485870361328125, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.713471531867981, + "num_tokens": 324349584.0, + "step": 12727 + }, + { + "epoch": 1.3977597188666813, + "grad_norm": 2.273963689804077, + "learning_rate": 1e-06, + "loss": 1.034, + "mean_token_accuracy": 0.6853178143501282, + "num_tokens": 324374489.0, + "step": 12728 + }, + { + "epoch": 1.397869536569295, + "grad_norm": 2.0760254859924316, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.6983786225318909, + "num_tokens": 324403536.0, + "step": 12729 + }, + { + "epoch": 1.3979793542719086, + "grad_norm": 2.08571720123291, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7064834833145142, + "num_tokens": 324434477.0, + "step": 12730 + }, + { + "epoch": 1.3980891719745223, + "grad_norm": 2.631730318069458, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.734430193901062, + "num_tokens": 324454028.0, + "step": 12731 + }, + { + "epoch": 1.398198989677136, + "grad_norm": 2.4300317764282227, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7431018352508545, + "num_tokens": 324475210.0, + "step": 12732 + }, + { + "epoch": 1.3983088073797496, + "grad_norm": 2.236797332763672, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7165167927742004, + "num_tokens": 324501260.0, + "step": 12733 + }, + { + "epoch": 1.3984186250823631, + "grad_norm": 2.153743028640747, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7393051385879517, + "num_tokens": 324527033.0, + "step": 12734 + }, + { + "epoch": 1.398528442784977, + "grad_norm": 2.4552996158599854, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7052251696586609, + "num_tokens": 324548834.0, + "step": 12735 + }, + { + "epoch": 1.3986382604875907, + "grad_norm": 2.1632683277130127, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6917072534561157, + "num_tokens": 324576662.0, + "step": 12736 + }, + { + "epoch": 1.3987480781902042, + "grad_norm": 2.270496129989624, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7135385274887085, + "num_tokens": 324601958.0, + "step": 12737 + }, + { + "epoch": 1.398857895892818, + "grad_norm": 2.2095491886138916, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7160087823867798, + "num_tokens": 324627900.0, + "step": 12738 + }, + { + "epoch": 1.3989677135954315, + "grad_norm": 2.086552381515503, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7088875770568848, + "num_tokens": 324655574.0, + "step": 12739 + }, + { + "epoch": 1.3990775312980452, + "grad_norm": 2.5116684436798096, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7221258878707886, + "num_tokens": 324677940.0, + "step": 12740 + }, + { + "epoch": 1.399187349000659, + "grad_norm": 2.388502359390259, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7086194753646851, + "num_tokens": 324701973.0, + "step": 12741 + }, + { + "epoch": 1.3992971667032725, + "grad_norm": 2.2477095127105713, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7046898603439331, + "num_tokens": 324727121.0, + "step": 12742 + }, + { + "epoch": 1.3994069844058863, + "grad_norm": 2.3803317546844482, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7209481000900269, + "num_tokens": 324751177.0, + "step": 12743 + }, + { + "epoch": 1.3995168021084998, + "grad_norm": 2.2801313400268555, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.718002200126648, + "num_tokens": 324776672.0, + "step": 12744 + }, + { + "epoch": 1.3996266198111136, + "grad_norm": 2.3065390586853027, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7212464809417725, + "num_tokens": 324801963.0, + "step": 12745 + }, + { + "epoch": 1.3997364375137273, + "grad_norm": 2.4186737537384033, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7073884606361389, + "num_tokens": 324824665.0, + "step": 12746 + }, + { + "epoch": 1.3998462552163409, + "grad_norm": 2.0525078773498535, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7149901390075684, + "num_tokens": 324854112.0, + "step": 12747 + }, + { + "epoch": 1.3999560729189544, + "grad_norm": 2.1507725715637207, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.684666633605957, + "num_tokens": 324881865.0, + "step": 12748 + }, + { + "epoch": 1.4000658906215682, + "grad_norm": 2.323317527770996, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7239383459091187, + "num_tokens": 324906213.0, + "step": 12749 + }, + { + "epoch": 1.400175708324182, + "grad_norm": 2.3271749019622803, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7226303815841675, + "num_tokens": 324928760.0, + "step": 12750 + }, + { + "epoch": 1.4002855260267955, + "grad_norm": 2.30790376663208, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.715976893901825, + "num_tokens": 324952349.0, + "step": 12751 + }, + { + "epoch": 1.4003953437294092, + "grad_norm": 2.2758023738861084, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7390118837356567, + "num_tokens": 324975849.0, + "step": 12752 + }, + { + "epoch": 1.4005051614320227, + "grad_norm": 2.0579230785369873, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7332884073257446, + "num_tokens": 325004107.0, + "step": 12753 + }, + { + "epoch": 1.4006149791346365, + "grad_norm": 2.122702121734619, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7126141786575317, + "num_tokens": 325029711.0, + "step": 12754 + }, + { + "epoch": 1.4007247968372503, + "grad_norm": 2.302610158920288, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7440881133079529, + "num_tokens": 325052680.0, + "step": 12755 + }, + { + "epoch": 1.4008346145398638, + "grad_norm": 2.2269880771636963, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.7004772424697876, + "num_tokens": 325081131.0, + "step": 12756 + }, + { + "epoch": 1.4009444322424776, + "grad_norm": 2.666464328765869, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7183212041854858, + "num_tokens": 325101101.0, + "step": 12757 + }, + { + "epoch": 1.401054249945091, + "grad_norm": 2.2552623748779297, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.696132242679596, + "num_tokens": 325128088.0, + "step": 12758 + }, + { + "epoch": 1.4011640676477048, + "grad_norm": 2.2811169624328613, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7220483422279358, + "num_tokens": 325152823.0, + "step": 12759 + }, + { + "epoch": 1.4012738853503186, + "grad_norm": 2.2024612426757812, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7468396425247192, + "num_tokens": 325178009.0, + "step": 12760 + }, + { + "epoch": 1.4013837030529321, + "grad_norm": 2.1599676609039307, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7052206993103027, + "num_tokens": 325206472.0, + "step": 12761 + }, + { + "epoch": 1.4014935207555457, + "grad_norm": 2.3793890476226807, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7168432474136353, + "num_tokens": 325228250.0, + "step": 12762 + }, + { + "epoch": 1.4016033384581594, + "grad_norm": 2.1947569847106934, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.7069816589355469, + "num_tokens": 325255413.0, + "step": 12763 + }, + { + "epoch": 1.4017131561607732, + "grad_norm": 2.4449737071990967, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7134811282157898, + "num_tokens": 325277824.0, + "step": 12764 + }, + { + "epoch": 1.4018229738633867, + "grad_norm": 2.158971071243286, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7224909067153931, + "num_tokens": 325306539.0, + "step": 12765 + }, + { + "epoch": 1.4019327915660005, + "grad_norm": 2.5304229259490967, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7253267765045166, + "num_tokens": 325327043.0, + "step": 12766 + }, + { + "epoch": 1.402042609268614, + "grad_norm": 2.31026029586792, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7440189123153687, + "num_tokens": 325348822.0, + "step": 12767 + }, + { + "epoch": 1.4021524269712278, + "grad_norm": 2.3172571659088135, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7227615118026733, + "num_tokens": 325372143.0, + "step": 12768 + }, + { + "epoch": 1.4022622446738415, + "grad_norm": 2.1586010456085205, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.72201007604599, + "num_tokens": 325397657.0, + "step": 12769 + }, + { + "epoch": 1.402372062376455, + "grad_norm": 2.268298625946045, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7236759662628174, + "num_tokens": 325421905.0, + "step": 12770 + }, + { + "epoch": 1.4024818800790688, + "grad_norm": 2.2098536491394043, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7026485800743103, + "num_tokens": 325448619.0, + "step": 12771 + }, + { + "epoch": 1.4025916977816824, + "grad_norm": 2.439422845840454, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7369215488433838, + "num_tokens": 325471096.0, + "step": 12772 + }, + { + "epoch": 1.4027015154842961, + "grad_norm": 2.3640782833099365, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7313969135284424, + "num_tokens": 325494992.0, + "step": 12773 + }, + { + "epoch": 1.4028113331869096, + "grad_norm": 1.9967503547668457, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7332018613815308, + "num_tokens": 325524032.0, + "step": 12774 + }, + { + "epoch": 1.4029211508895234, + "grad_norm": 2.23252272605896, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7301156520843506, + "num_tokens": 325548740.0, + "step": 12775 + }, + { + "epoch": 1.403030968592137, + "grad_norm": 2.2937653064727783, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.709868848323822, + "num_tokens": 325573758.0, + "step": 12776 + }, + { + "epoch": 1.4031407862947507, + "grad_norm": 2.238375425338745, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7424347996711731, + "num_tokens": 325598074.0, + "step": 12777 + }, + { + "epoch": 1.4032506039973645, + "grad_norm": 2.171430826187134, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7171624898910522, + "num_tokens": 325624666.0, + "step": 12778 + }, + { + "epoch": 1.403360421699978, + "grad_norm": 2.0210630893707275, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.7041195631027222, + "num_tokens": 325655569.0, + "step": 12779 + }, + { + "epoch": 1.4034702394025917, + "grad_norm": 1.9270308017730713, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6879154443740845, + "num_tokens": 325688902.0, + "step": 12780 + }, + { + "epoch": 1.4035800571052053, + "grad_norm": 1.9828627109527588, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7188572287559509, + "num_tokens": 325719489.0, + "step": 12781 + }, + { + "epoch": 1.403689874807819, + "grad_norm": 1.9965944290161133, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.6975115537643433, + "num_tokens": 325749366.0, + "step": 12782 + }, + { + "epoch": 1.4037996925104328, + "grad_norm": 2.2912960052490234, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7198653221130371, + "num_tokens": 325775186.0, + "step": 12783 + }, + { + "epoch": 1.4039095102130463, + "grad_norm": 2.1844091415405273, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.6990140676498413, + "num_tokens": 325803472.0, + "step": 12784 + }, + { + "epoch": 1.40401932791566, + "grad_norm": 2.1005401611328125, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7031431794166565, + "num_tokens": 325831464.0, + "step": 12785 + }, + { + "epoch": 1.4041291456182736, + "grad_norm": 2.0133297443389893, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.6994641423225403, + "num_tokens": 325862696.0, + "step": 12786 + }, + { + "epoch": 1.4042389633208874, + "grad_norm": 2.3875741958618164, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7239507436752319, + "num_tokens": 325884797.0, + "step": 12787 + }, + { + "epoch": 1.404348781023501, + "grad_norm": 2.4205119609832764, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7390470504760742, + "num_tokens": 325906809.0, + "step": 12788 + }, + { + "epoch": 1.4044585987261147, + "grad_norm": 2.059997081756592, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7186656594276428, + "num_tokens": 325934104.0, + "step": 12789 + }, + { + "epoch": 1.4045684164287282, + "grad_norm": 2.1308484077453613, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7291789054870605, + "num_tokens": 325963329.0, + "step": 12790 + }, + { + "epoch": 1.404678234131342, + "grad_norm": 2.3177297115325928, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7131367921829224, + "num_tokens": 325986822.0, + "step": 12791 + }, + { + "epoch": 1.4047880518339557, + "grad_norm": 2.1330840587615967, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7134060859680176, + "num_tokens": 326015227.0, + "step": 12792 + }, + { + "epoch": 1.4048978695365693, + "grad_norm": 2.483678102493286, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7320884466171265, + "num_tokens": 326036958.0, + "step": 12793 + }, + { + "epoch": 1.405007687239183, + "grad_norm": 2.337811231613159, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7064577341079712, + "num_tokens": 326060413.0, + "step": 12794 + }, + { + "epoch": 1.4051175049417965, + "grad_norm": 2.2695558071136475, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7160595059394836, + "num_tokens": 326086851.0, + "step": 12795 + }, + { + "epoch": 1.4052273226444103, + "grad_norm": 2.2462313175201416, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7232810258865356, + "num_tokens": 326112371.0, + "step": 12796 + }, + { + "epoch": 1.405337140347024, + "grad_norm": 2.55057954788208, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7275725603103638, + "num_tokens": 326132501.0, + "step": 12797 + }, + { + "epoch": 1.4054469580496376, + "grad_norm": 2.526224136352539, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7198452353477478, + "num_tokens": 326153503.0, + "step": 12798 + }, + { + "epoch": 1.4055567757522511, + "grad_norm": 2.2189548015594482, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6992942094802856, + "num_tokens": 326182154.0, + "step": 12799 + }, + { + "epoch": 1.4056665934548649, + "grad_norm": 2.443234920501709, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7278326153755188, + "num_tokens": 326204097.0, + "step": 12800 + }, + { + "epoch": 1.4057764111574786, + "grad_norm": 2.3204379081726074, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7150386571884155, + "num_tokens": 326229626.0, + "step": 12801 + }, + { + "epoch": 1.4058862288600922, + "grad_norm": 2.167431116104126, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7072317004203796, + "num_tokens": 326257973.0, + "step": 12802 + }, + { + "epoch": 1.405996046562706, + "grad_norm": 2.464921712875366, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.712499737739563, + "num_tokens": 326280216.0, + "step": 12803 + }, + { + "epoch": 1.4061058642653195, + "grad_norm": 2.1717731952667236, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7151368856430054, + "num_tokens": 326308142.0, + "step": 12804 + }, + { + "epoch": 1.4062156819679332, + "grad_norm": 1.9699649810791016, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.726706326007843, + "num_tokens": 326336731.0, + "step": 12805 + }, + { + "epoch": 1.406325499670547, + "grad_norm": 2.4694576263427734, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7317230701446533, + "num_tokens": 326358348.0, + "step": 12806 + }, + { + "epoch": 1.4064353173731605, + "grad_norm": 2.571154832839966, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7396341562271118, + "num_tokens": 326378647.0, + "step": 12807 + }, + { + "epoch": 1.4065451350757743, + "grad_norm": 2.086686849594116, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7189939022064209, + "num_tokens": 326405529.0, + "step": 12808 + }, + { + "epoch": 1.4066549527783878, + "grad_norm": 2.302825450897217, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7212758660316467, + "num_tokens": 326429938.0, + "step": 12809 + }, + { + "epoch": 1.4067647704810016, + "grad_norm": 2.6178336143493652, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7380157709121704, + "num_tokens": 326449830.0, + "step": 12810 + }, + { + "epoch": 1.4068745881836153, + "grad_norm": 2.2287776470184326, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.6941448450088501, + "num_tokens": 326477280.0, + "step": 12811 + }, + { + "epoch": 1.4069844058862289, + "grad_norm": 2.5483882427215576, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7236904501914978, + "num_tokens": 326500016.0, + "step": 12812 + }, + { + "epoch": 1.4070942235888424, + "grad_norm": 2.35080623626709, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7205466032028198, + "num_tokens": 326523269.0, + "step": 12813 + }, + { + "epoch": 1.4072040412914562, + "grad_norm": 2.3369407653808594, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7092536687850952, + "num_tokens": 326550372.0, + "step": 12814 + }, + { + "epoch": 1.40731385899407, + "grad_norm": 2.4145374298095703, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7239179015159607, + "num_tokens": 326574095.0, + "step": 12815 + }, + { + "epoch": 1.4074236766966834, + "grad_norm": 2.17858624458313, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7157957553863525, + "num_tokens": 326602205.0, + "step": 12816 + }, + { + "epoch": 1.4075334943992972, + "grad_norm": 2.404782295227051, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7173341512680054, + "num_tokens": 326625962.0, + "step": 12817 + }, + { + "epoch": 1.4076433121019107, + "grad_norm": 2.1059677600860596, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7216458320617676, + "num_tokens": 326653600.0, + "step": 12818 + }, + { + "epoch": 1.4077531298045245, + "grad_norm": 2.0484306812286377, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7042399048805237, + "num_tokens": 326684648.0, + "step": 12819 + }, + { + "epoch": 1.4078629475071383, + "grad_norm": 2.406909942626953, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7227983474731445, + "num_tokens": 326706388.0, + "step": 12820 + }, + { + "epoch": 1.4079727652097518, + "grad_norm": 2.3213744163513184, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7375351190567017, + "num_tokens": 326728102.0, + "step": 12821 + }, + { + "epoch": 1.4080825829123655, + "grad_norm": 2.180208206176758, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7345601320266724, + "num_tokens": 326753547.0, + "step": 12822 + }, + { + "epoch": 1.408192400614979, + "grad_norm": 2.2354133129119873, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7444371581077576, + "num_tokens": 326776611.0, + "step": 12823 + }, + { + "epoch": 1.4083022183175928, + "grad_norm": 2.518098831176758, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7421703934669495, + "num_tokens": 326798075.0, + "step": 12824 + }, + { + "epoch": 1.4084120360202066, + "grad_norm": 2.7669715881347656, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7329632043838501, + "num_tokens": 326815997.0, + "step": 12825 + }, + { + "epoch": 1.4085218537228201, + "grad_norm": 2.2282745838165283, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7214492559432983, + "num_tokens": 326841090.0, + "step": 12826 + }, + { + "epoch": 1.4086316714254337, + "grad_norm": 2.4351518154144287, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7114487290382385, + "num_tokens": 326863740.0, + "step": 12827 + }, + { + "epoch": 1.4087414891280474, + "grad_norm": 2.129856824874878, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.716551661491394, + "num_tokens": 326891994.0, + "step": 12828 + }, + { + "epoch": 1.4088513068306612, + "grad_norm": 1.9567179679870605, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7085216045379639, + "num_tokens": 326926000.0, + "step": 12829 + }, + { + "epoch": 1.4089611245332747, + "grad_norm": 2.3715920448303223, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7388553619384766, + "num_tokens": 326949305.0, + "step": 12830 + }, + { + "epoch": 1.4090709422358885, + "grad_norm": 2.24013352394104, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7071681618690491, + "num_tokens": 326975404.0, + "step": 12831 + }, + { + "epoch": 1.409180759938502, + "grad_norm": 2.3044004440307617, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7186890840530396, + "num_tokens": 326999373.0, + "step": 12832 + }, + { + "epoch": 1.4092905776411158, + "grad_norm": 2.0463626384735107, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7128183841705322, + "num_tokens": 327028770.0, + "step": 12833 + }, + { + "epoch": 1.4094003953437295, + "grad_norm": 2.554492712020874, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7305367588996887, + "num_tokens": 327048875.0, + "step": 12834 + }, + { + "epoch": 1.409510213046343, + "grad_norm": 2.302525281906128, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7352960109710693, + "num_tokens": 327072638.0, + "step": 12835 + }, + { + "epoch": 1.4096200307489568, + "grad_norm": 2.159147262573242, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7300684452056885, + "num_tokens": 327098591.0, + "step": 12836 + }, + { + "epoch": 1.4097298484515703, + "grad_norm": 2.466357707977295, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7195584774017334, + "num_tokens": 327120499.0, + "step": 12837 + }, + { + "epoch": 1.409839666154184, + "grad_norm": 2.0425195693969727, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7189255952835083, + "num_tokens": 327150501.0, + "step": 12838 + }, + { + "epoch": 1.4099494838567976, + "grad_norm": 2.195253849029541, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7471213936805725, + "num_tokens": 327175277.0, + "step": 12839 + }, + { + "epoch": 1.4100593015594114, + "grad_norm": 2.060546875, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.7011334896087646, + "num_tokens": 327205718.0, + "step": 12840 + }, + { + "epoch": 1.410169119262025, + "grad_norm": 2.282578945159912, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7299704551696777, + "num_tokens": 327230047.0, + "step": 12841 + }, + { + "epoch": 1.4102789369646387, + "grad_norm": 2.048135995864868, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7035348415374756, + "num_tokens": 327261942.0, + "step": 12842 + }, + { + "epoch": 1.4103887546672524, + "grad_norm": 2.124107599258423, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6992985010147095, + "num_tokens": 327289180.0, + "step": 12843 + }, + { + "epoch": 1.410498572369866, + "grad_norm": 2.2269043922424316, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7050129771232605, + "num_tokens": 327315785.0, + "step": 12844 + }, + { + "epoch": 1.4106083900724797, + "grad_norm": 2.0346336364746094, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.7029763460159302, + "num_tokens": 327345250.0, + "step": 12845 + }, + { + "epoch": 1.4107182077750933, + "grad_norm": 2.281426429748535, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7110344171524048, + "num_tokens": 327372029.0, + "step": 12846 + }, + { + "epoch": 1.410828025477707, + "grad_norm": 2.036172389984131, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.710543155670166, + "num_tokens": 327402015.0, + "step": 12847 + }, + { + "epoch": 1.4109378431803208, + "grad_norm": 2.3092472553253174, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7042871713638306, + "num_tokens": 327427674.0, + "step": 12848 + }, + { + "epoch": 1.4110476608829343, + "grad_norm": 2.165372848510742, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7142090797424316, + "num_tokens": 327456897.0, + "step": 12849 + }, + { + "epoch": 1.4111574785855479, + "grad_norm": 2.134000062942505, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.6970594525337219, + "num_tokens": 327486381.0, + "step": 12850 + }, + { + "epoch": 1.4112672962881616, + "grad_norm": 2.332693099975586, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7217419147491455, + "num_tokens": 327510841.0, + "step": 12851 + }, + { + "epoch": 1.4113771139907754, + "grad_norm": 2.5142788887023926, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7319473028182983, + "num_tokens": 327535130.0, + "step": 12852 + }, + { + "epoch": 1.411486931693389, + "grad_norm": 2.216322898864746, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.713568925857544, + "num_tokens": 327563474.0, + "step": 12853 + }, + { + "epoch": 1.4115967493960027, + "grad_norm": 2.520118236541748, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7388933897018433, + "num_tokens": 327584241.0, + "step": 12854 + }, + { + "epoch": 1.4117065670986162, + "grad_norm": 2.1089296340942383, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7313089370727539, + "num_tokens": 327612360.0, + "step": 12855 + }, + { + "epoch": 1.41181638480123, + "grad_norm": 1.9754496812820435, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7149076461791992, + "num_tokens": 327645387.0, + "step": 12856 + }, + { + "epoch": 1.4119262025038437, + "grad_norm": 1.916573166847229, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7302382588386536, + "num_tokens": 327679021.0, + "step": 12857 + }, + { + "epoch": 1.4120360202064572, + "grad_norm": 2.431445598602295, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7372719049453735, + "num_tokens": 327700201.0, + "step": 12858 + }, + { + "epoch": 1.412145837909071, + "grad_norm": 2.332951068878174, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7212638854980469, + "num_tokens": 327725182.0, + "step": 12859 + }, + { + "epoch": 1.4122556556116845, + "grad_norm": 2.0892040729522705, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.689059853553772, + "num_tokens": 327757093.0, + "step": 12860 + }, + { + "epoch": 1.4123654733142983, + "grad_norm": 2.200615167617798, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7179586887359619, + "num_tokens": 327784551.0, + "step": 12861 + }, + { + "epoch": 1.412475291016912, + "grad_norm": 2.278083562850952, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.7092317938804626, + "num_tokens": 327809360.0, + "step": 12862 + }, + { + "epoch": 1.4125851087195256, + "grad_norm": 2.327367067337036, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7181563377380371, + "num_tokens": 327832999.0, + "step": 12863 + }, + { + "epoch": 1.4126949264221391, + "grad_norm": 2.174043893814087, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7188341617584229, + "num_tokens": 327859983.0, + "step": 12864 + }, + { + "epoch": 1.4128047441247529, + "grad_norm": 2.4133353233337402, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7228737473487854, + "num_tokens": 327882614.0, + "step": 12865 + }, + { + "epoch": 1.4129145618273666, + "grad_norm": 2.2132408618927, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7232599854469299, + "num_tokens": 327907535.0, + "step": 12866 + }, + { + "epoch": 1.4130243795299802, + "grad_norm": 2.3586223125457764, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7322816848754883, + "num_tokens": 327932254.0, + "step": 12867 + }, + { + "epoch": 1.413134197232594, + "grad_norm": 2.2114129066467285, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7305167317390442, + "num_tokens": 327960135.0, + "step": 12868 + }, + { + "epoch": 1.4132440149352075, + "grad_norm": 1.808712363243103, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.717130184173584, + "num_tokens": 327999068.0, + "step": 12869 + }, + { + "epoch": 1.4133538326378212, + "grad_norm": 1.9720643758773804, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.687873363494873, + "num_tokens": 328033673.0, + "step": 12870 + }, + { + "epoch": 1.413463650340435, + "grad_norm": 2.2797558307647705, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7094331383705139, + "num_tokens": 328059189.0, + "step": 12871 + }, + { + "epoch": 1.4135734680430485, + "grad_norm": 2.5553793907165527, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7160732746124268, + "num_tokens": 328079441.0, + "step": 12872 + }, + { + "epoch": 1.4136832857456623, + "grad_norm": 2.4864516258239746, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7261294722557068, + "num_tokens": 328103580.0, + "step": 12873 + }, + { + "epoch": 1.4137931034482758, + "grad_norm": 2.0820486545562744, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.7452515363693237, + "num_tokens": 328131074.0, + "step": 12874 + }, + { + "epoch": 1.4139029211508896, + "grad_norm": 2.388328790664673, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7277836799621582, + "num_tokens": 328153427.0, + "step": 12875 + }, + { + "epoch": 1.4140127388535033, + "grad_norm": 2.3534646034240723, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7317987084388733, + "num_tokens": 328175847.0, + "step": 12876 + }, + { + "epoch": 1.4141225565561168, + "grad_norm": 1.9819079637527466, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7289849519729614, + "num_tokens": 328205380.0, + "step": 12877 + }, + { + "epoch": 1.4142323742587304, + "grad_norm": 2.392481803894043, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7273467779159546, + "num_tokens": 328229466.0, + "step": 12878 + }, + { + "epoch": 1.4143421919613441, + "grad_norm": 2.3136868476867676, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7111128568649292, + "num_tokens": 328255533.0, + "step": 12879 + }, + { + "epoch": 1.414452009663958, + "grad_norm": 2.1695821285247803, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7353599071502686, + "num_tokens": 328281315.0, + "step": 12880 + }, + { + "epoch": 1.4145618273665714, + "grad_norm": 1.8877828121185303, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7011842727661133, + "num_tokens": 328315112.0, + "step": 12881 + }, + { + "epoch": 1.4146716450691852, + "grad_norm": 2.1587226390838623, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7134844064712524, + "num_tokens": 328341854.0, + "step": 12882 + }, + { + "epoch": 1.4147814627717987, + "grad_norm": 2.4188804626464844, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7200102806091309, + "num_tokens": 328365297.0, + "step": 12883 + }, + { + "epoch": 1.4148912804744125, + "grad_norm": 1.9715837240219116, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6993162631988525, + "num_tokens": 328398594.0, + "step": 12884 + }, + { + "epoch": 1.4150010981770262, + "grad_norm": 2.407341241836548, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7357283234596252, + "num_tokens": 328422199.0, + "step": 12885 + }, + { + "epoch": 1.4151109158796398, + "grad_norm": 2.094913959503174, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7038719654083252, + "num_tokens": 328453186.0, + "step": 12886 + }, + { + "epoch": 1.4152207335822535, + "grad_norm": 2.3553466796875, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7162858247756958, + "num_tokens": 328478182.0, + "step": 12887 + }, + { + "epoch": 1.415330551284867, + "grad_norm": 2.0832982063293457, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7147904634475708, + "num_tokens": 328506344.0, + "step": 12888 + }, + { + "epoch": 1.4154403689874808, + "grad_norm": 2.0872371196746826, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7567064762115479, + "num_tokens": 328534218.0, + "step": 12889 + }, + { + "epoch": 1.4155501866900946, + "grad_norm": 2.3051748275756836, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7312350273132324, + "num_tokens": 328559130.0, + "step": 12890 + }, + { + "epoch": 1.4156600043927081, + "grad_norm": 2.231623649597168, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7247848510742188, + "num_tokens": 328585628.0, + "step": 12891 + }, + { + "epoch": 1.4157698220953217, + "grad_norm": 1.84794282913208, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7264162302017212, + "num_tokens": 328619814.0, + "step": 12892 + }, + { + "epoch": 1.4158796397979354, + "grad_norm": 2.3002195358276367, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.714505672454834, + "num_tokens": 328643389.0, + "step": 12893 + }, + { + "epoch": 1.4159894575005492, + "grad_norm": 2.1408536434173584, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7145965099334717, + "num_tokens": 328670933.0, + "step": 12894 + }, + { + "epoch": 1.4160992752031627, + "grad_norm": 2.420891523361206, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7144137620925903, + "num_tokens": 328694487.0, + "step": 12895 + }, + { + "epoch": 1.4162090929057765, + "grad_norm": 2.43456768989563, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7165771722793579, + "num_tokens": 328716180.0, + "step": 12896 + }, + { + "epoch": 1.41631891060839, + "grad_norm": 2.0882532596588135, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7036384344100952, + "num_tokens": 328745842.0, + "step": 12897 + }, + { + "epoch": 1.4164287283110037, + "grad_norm": 2.5348942279815674, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7419149875640869, + "num_tokens": 328765148.0, + "step": 12898 + }, + { + "epoch": 1.4165385460136175, + "grad_norm": 2.0211129188537598, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7067286372184753, + "num_tokens": 328798496.0, + "step": 12899 + }, + { + "epoch": 1.416648363716231, + "grad_norm": 2.2751567363739014, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7023833394050598, + "num_tokens": 328823366.0, + "step": 12900 + }, + { + "epoch": 1.4167581814188448, + "grad_norm": 2.1472878456115723, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7213895916938782, + "num_tokens": 328850155.0, + "step": 12901 + }, + { + "epoch": 1.4168679991214583, + "grad_norm": 1.9966338872909546, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7067303657531738, + "num_tokens": 328883099.0, + "step": 12902 + }, + { + "epoch": 1.416977816824072, + "grad_norm": 2.144381284713745, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7071883678436279, + "num_tokens": 328911155.0, + "step": 12903 + }, + { + "epoch": 1.4170876345266856, + "grad_norm": 2.4409806728363037, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.731039822101593, + "num_tokens": 328932151.0, + "step": 12904 + }, + { + "epoch": 1.4171974522292994, + "grad_norm": 2.089550495147705, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.722519040107727, + "num_tokens": 328958992.0, + "step": 12905 + }, + { + "epoch": 1.417307269931913, + "grad_norm": 2.225886821746826, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.724111795425415, + "num_tokens": 328983817.0, + "step": 12906 + }, + { + "epoch": 1.4174170876345267, + "grad_norm": 2.1146786212921143, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7461452484130859, + "num_tokens": 329012014.0, + "step": 12907 + }, + { + "epoch": 1.4175269053371404, + "grad_norm": 2.340278148651123, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7411867380142212, + "num_tokens": 329032945.0, + "step": 12908 + }, + { + "epoch": 1.417636723039754, + "grad_norm": 2.334127187728882, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7128857374191284, + "num_tokens": 329056818.0, + "step": 12909 + }, + { + "epoch": 1.4177465407423677, + "grad_norm": 1.9318337440490723, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7166926860809326, + "num_tokens": 329091005.0, + "step": 12910 + }, + { + "epoch": 1.4178563584449813, + "grad_norm": 2.1218886375427246, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7187528610229492, + "num_tokens": 329121970.0, + "step": 12911 + }, + { + "epoch": 1.417966176147595, + "grad_norm": 2.274782180786133, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7088000774383545, + "num_tokens": 329146610.0, + "step": 12912 + }, + { + "epoch": 1.4180759938502088, + "grad_norm": 2.488150119781494, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7191243171691895, + "num_tokens": 329166866.0, + "step": 12913 + }, + { + "epoch": 1.4181858115528223, + "grad_norm": 2.246535062789917, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7112120389938354, + "num_tokens": 329196282.0, + "step": 12914 + }, + { + "epoch": 1.4182956292554358, + "grad_norm": 2.1281816959381104, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7085257172584534, + "num_tokens": 329223877.0, + "step": 12915 + }, + { + "epoch": 1.4184054469580496, + "grad_norm": 2.2934281826019287, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7236636877059937, + "num_tokens": 329248027.0, + "step": 12916 + }, + { + "epoch": 1.4185152646606634, + "grad_norm": 2.287132978439331, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7321996092796326, + "num_tokens": 329272599.0, + "step": 12917 + }, + { + "epoch": 1.418625082363277, + "grad_norm": 2.294804811477661, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7159618139266968, + "num_tokens": 329298831.0, + "step": 12918 + }, + { + "epoch": 1.4187349000658906, + "grad_norm": 2.1820929050445557, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7275371551513672, + "num_tokens": 329325332.0, + "step": 12919 + }, + { + "epoch": 1.4188447177685042, + "grad_norm": 2.271324634552002, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7390303611755371, + "num_tokens": 329350263.0, + "step": 12920 + }, + { + "epoch": 1.418954535471118, + "grad_norm": 2.0888829231262207, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7101908922195435, + "num_tokens": 329378270.0, + "step": 12921 + }, + { + "epoch": 1.4190643531737317, + "grad_norm": 2.0569913387298584, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7333012819290161, + "num_tokens": 329405271.0, + "step": 12922 + }, + { + "epoch": 1.4191741708763452, + "grad_norm": 2.2255699634552, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7240222692489624, + "num_tokens": 329429774.0, + "step": 12923 + }, + { + "epoch": 1.419283988578959, + "grad_norm": 2.2432188987731934, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.6953076124191284, + "num_tokens": 329457143.0, + "step": 12924 + }, + { + "epoch": 1.4193938062815725, + "grad_norm": 2.0987441539764404, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7376523017883301, + "num_tokens": 329483231.0, + "step": 12925 + }, + { + "epoch": 1.4195036239841863, + "grad_norm": 2.364447593688965, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7217252850532532, + "num_tokens": 329506168.0, + "step": 12926 + }, + { + "epoch": 1.4196134416868, + "grad_norm": 2.081122398376465, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.718575119972229, + "num_tokens": 329535225.0, + "step": 12927 + }, + { + "epoch": 1.4197232593894136, + "grad_norm": 2.170898675918579, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7230032682418823, + "num_tokens": 329561915.0, + "step": 12928 + }, + { + "epoch": 1.419833077092027, + "grad_norm": 2.175851821899414, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.716641902923584, + "num_tokens": 329587791.0, + "step": 12929 + }, + { + "epoch": 1.4199428947946409, + "grad_norm": 2.079883337020874, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7149420976638794, + "num_tokens": 329616274.0, + "step": 12930 + }, + { + "epoch": 1.4200527124972546, + "grad_norm": 2.180150270462036, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7300543785095215, + "num_tokens": 329641189.0, + "step": 12931 + }, + { + "epoch": 1.4201625301998682, + "grad_norm": 2.4991555213928223, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7249888777732849, + "num_tokens": 329661573.0, + "step": 12932 + }, + { + "epoch": 1.420272347902482, + "grad_norm": 2.0401060581207275, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6857051849365234, + "num_tokens": 329693417.0, + "step": 12933 + }, + { + "epoch": 1.4203821656050954, + "grad_norm": 2.2471234798431396, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7156354188919067, + "num_tokens": 329720583.0, + "step": 12934 + }, + { + "epoch": 1.4204919833077092, + "grad_norm": 2.4475724697113037, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7208289504051208, + "num_tokens": 329741951.0, + "step": 12935 + }, + { + "epoch": 1.420601801010323, + "grad_norm": 2.4251389503479004, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7339857816696167, + "num_tokens": 329763621.0, + "step": 12936 + }, + { + "epoch": 1.4207116187129365, + "grad_norm": 2.1963236331939697, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7165096998214722, + "num_tokens": 329793557.0, + "step": 12937 + }, + { + "epoch": 1.4208214364155503, + "grad_norm": 2.9567437171936035, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7138006687164307, + "num_tokens": 329811349.0, + "step": 12938 + }, + { + "epoch": 1.4209312541181638, + "grad_norm": 2.6046040058135986, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7131215929985046, + "num_tokens": 329832980.0, + "step": 12939 + }, + { + "epoch": 1.4210410718207775, + "grad_norm": 2.1542513370513916, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7048704624176025, + "num_tokens": 329859423.0, + "step": 12940 + }, + { + "epoch": 1.4211508895233913, + "grad_norm": 2.1748743057250977, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7014938592910767, + "num_tokens": 329888696.0, + "step": 12941 + }, + { + "epoch": 1.4212607072260048, + "grad_norm": 2.312260627746582, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7182984352111816, + "num_tokens": 329912875.0, + "step": 12942 + }, + { + "epoch": 1.4213705249286184, + "grad_norm": 2.181349039077759, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7276217937469482, + "num_tokens": 329939955.0, + "step": 12943 + }, + { + "epoch": 1.4214803426312321, + "grad_norm": 2.11893630027771, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7401158809661865, + "num_tokens": 329967353.0, + "step": 12944 + }, + { + "epoch": 1.4215901603338459, + "grad_norm": 2.2797353267669678, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.757302463054657, + "num_tokens": 329989463.0, + "step": 12945 + }, + { + "epoch": 1.4216999780364594, + "grad_norm": 2.4428229331970215, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6938632726669312, + "num_tokens": 330013669.0, + "step": 12946 + }, + { + "epoch": 1.4218097957390732, + "grad_norm": 2.4107799530029297, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7145684957504272, + "num_tokens": 330037755.0, + "step": 12947 + }, + { + "epoch": 1.4219196134416867, + "grad_norm": 2.430377960205078, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.735100507736206, + "num_tokens": 330059889.0, + "step": 12948 + }, + { + "epoch": 1.4220294311443005, + "grad_norm": 2.301896333694458, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7060588598251343, + "num_tokens": 330087091.0, + "step": 12949 + }, + { + "epoch": 1.4221392488469142, + "grad_norm": 2.4894838333129883, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7033481001853943, + "num_tokens": 330109806.0, + "step": 12950 + }, + { + "epoch": 1.4222490665495278, + "grad_norm": 2.186300754547119, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7235101461410522, + "num_tokens": 330134456.0, + "step": 12951 + }, + { + "epoch": 1.4223588842521415, + "grad_norm": 2.4634792804718018, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.728062093257904, + "num_tokens": 330155051.0, + "step": 12952 + }, + { + "epoch": 1.422468701954755, + "grad_norm": 2.34035587310791, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.714077353477478, + "num_tokens": 330179171.0, + "step": 12953 + }, + { + "epoch": 1.4225785196573688, + "grad_norm": 2.6501736640930176, + "learning_rate": 1e-06, + "loss": 0.805, + "mean_token_accuracy": 0.7406246662139893, + "num_tokens": 330197051.0, + "step": 12954 + }, + { + "epoch": 1.4226883373599823, + "grad_norm": 2.486499547958374, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7076399326324463, + "num_tokens": 330219525.0, + "step": 12955 + }, + { + "epoch": 1.422798155062596, + "grad_norm": 2.2586252689361572, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7133430242538452, + "num_tokens": 330247414.0, + "step": 12956 + }, + { + "epoch": 1.4229079727652096, + "grad_norm": 2.3260958194732666, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7274461388587952, + "num_tokens": 330271890.0, + "step": 12957 + }, + { + "epoch": 1.4230177904678234, + "grad_norm": 2.3069980144500732, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7073385715484619, + "num_tokens": 330297125.0, + "step": 12958 + }, + { + "epoch": 1.4231276081704372, + "grad_norm": 2.431849241256714, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7234281897544861, + "num_tokens": 330319879.0, + "step": 12959 + }, + { + "epoch": 1.4232374258730507, + "grad_norm": 2.4753243923187256, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7140750885009766, + "num_tokens": 330341879.0, + "step": 12960 + }, + { + "epoch": 1.4233472435756644, + "grad_norm": 2.1851704120635986, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7170213460922241, + "num_tokens": 330368622.0, + "step": 12961 + }, + { + "epoch": 1.423457061278278, + "grad_norm": 2.3091306686401367, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7279436588287354, + "num_tokens": 330393185.0, + "step": 12962 + }, + { + "epoch": 1.4235668789808917, + "grad_norm": 2.3754985332489014, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7328991293907166, + "num_tokens": 330415557.0, + "step": 12963 + }, + { + "epoch": 1.4236766966835055, + "grad_norm": 2.2070465087890625, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7316960096359253, + "num_tokens": 330440334.0, + "step": 12964 + }, + { + "epoch": 1.423786514386119, + "grad_norm": 2.4274673461914062, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7125893831253052, + "num_tokens": 330464337.0, + "step": 12965 + }, + { + "epoch": 1.4238963320887328, + "grad_norm": 2.4904303550720215, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.7430001497268677, + "num_tokens": 330485866.0, + "step": 12966 + }, + { + "epoch": 1.4240061497913463, + "grad_norm": 2.3025667667388916, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7235907316207886, + "num_tokens": 330510853.0, + "step": 12967 + }, + { + "epoch": 1.42411596749396, + "grad_norm": 2.161280632019043, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7129946351051331, + "num_tokens": 330538455.0, + "step": 12968 + }, + { + "epoch": 1.4242257851965736, + "grad_norm": 2.2724668979644775, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.710692286491394, + "num_tokens": 330562983.0, + "step": 12969 + }, + { + "epoch": 1.4243356028991874, + "grad_norm": 2.1888363361358643, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7005506753921509, + "num_tokens": 330589272.0, + "step": 12970 + }, + { + "epoch": 1.424445420601801, + "grad_norm": 2.4547066688537598, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.703787088394165, + "num_tokens": 330612009.0, + "step": 12971 + }, + { + "epoch": 1.4245552383044147, + "grad_norm": 2.299187183380127, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7176809310913086, + "num_tokens": 330635250.0, + "step": 12972 + }, + { + "epoch": 1.4246650560070284, + "grad_norm": 2.4606666564941406, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.730482816696167, + "num_tokens": 330656976.0, + "step": 12973 + }, + { + "epoch": 1.424774873709642, + "grad_norm": 1.8469314575195312, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7174145579338074, + "num_tokens": 330689974.0, + "step": 12974 + }, + { + "epoch": 1.4248846914122557, + "grad_norm": 2.4813666343688965, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7243354320526123, + "num_tokens": 330714755.0, + "step": 12975 + }, + { + "epoch": 1.4249945091148692, + "grad_norm": 2.564344644546509, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7041877508163452, + "num_tokens": 330738404.0, + "step": 12976 + }, + { + "epoch": 1.425104326817483, + "grad_norm": 2.339975595474243, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7001006603240967, + "num_tokens": 330762886.0, + "step": 12977 + }, + { + "epoch": 1.4252141445200968, + "grad_norm": 2.3571603298187256, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.711493730545044, + "num_tokens": 330786920.0, + "step": 12978 + }, + { + "epoch": 1.4253239622227103, + "grad_norm": 2.5967795848846436, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.718279242515564, + "num_tokens": 330807197.0, + "step": 12979 + }, + { + "epoch": 1.4254337799253238, + "grad_norm": 2.4967081546783447, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7273973226547241, + "num_tokens": 330829072.0, + "step": 12980 + }, + { + "epoch": 1.4255435976279376, + "grad_norm": 2.4894304275512695, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7142652869224548, + "num_tokens": 330850437.0, + "step": 12981 + }, + { + "epoch": 1.4256534153305513, + "grad_norm": 2.592005491256714, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7330458164215088, + "num_tokens": 330871106.0, + "step": 12982 + }, + { + "epoch": 1.4257632330331649, + "grad_norm": 2.2110373973846436, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7284021973609924, + "num_tokens": 330898050.0, + "step": 12983 + }, + { + "epoch": 1.4258730507357786, + "grad_norm": 2.210347890853882, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7239205837249756, + "num_tokens": 330925053.0, + "step": 12984 + }, + { + "epoch": 1.4259828684383922, + "grad_norm": 2.2216861248016357, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.697123646736145, + "num_tokens": 330952338.0, + "step": 12985 + }, + { + "epoch": 1.426092686141006, + "grad_norm": 2.078958511352539, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.709246814250946, + "num_tokens": 330983356.0, + "step": 12986 + }, + { + "epoch": 1.4262025038436197, + "grad_norm": 2.3099331855773926, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7366358637809753, + "num_tokens": 331007990.0, + "step": 12987 + }, + { + "epoch": 1.4263123215462332, + "grad_norm": 2.2381999492645264, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7227104902267456, + "num_tokens": 331033790.0, + "step": 12988 + }, + { + "epoch": 1.426422139248847, + "grad_norm": 2.554372787475586, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7295819520950317, + "num_tokens": 331054413.0, + "step": 12989 + }, + { + "epoch": 1.4265319569514605, + "grad_norm": 2.0880162715911865, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.7025126218795776, + "num_tokens": 331083873.0, + "step": 12990 + }, + { + "epoch": 1.4266417746540743, + "grad_norm": 2.3949177265167236, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.6935126781463623, + "num_tokens": 331108954.0, + "step": 12991 + }, + { + "epoch": 1.426751592356688, + "grad_norm": 2.321178674697876, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7298881411552429, + "num_tokens": 331133042.0, + "step": 12992 + }, + { + "epoch": 1.4268614100593016, + "grad_norm": 2.243954658508301, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7244426012039185, + "num_tokens": 331157622.0, + "step": 12993 + }, + { + "epoch": 1.426971227761915, + "grad_norm": 2.204883098602295, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7119847536087036, + "num_tokens": 331183038.0, + "step": 12994 + }, + { + "epoch": 1.4270810454645289, + "grad_norm": 2.695798635482788, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7399516105651855, + "num_tokens": 331200290.0, + "step": 12995 + }, + { + "epoch": 1.4271908631671426, + "grad_norm": 2.3444526195526123, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7344654202461243, + "num_tokens": 331224203.0, + "step": 12996 + }, + { + "epoch": 1.4273006808697561, + "grad_norm": 2.2974843978881836, + "learning_rate": 1e-06, + "loss": 1.0395, + "mean_token_accuracy": 0.6758418083190918, + "num_tokens": 331251786.0, + "step": 12997 + }, + { + "epoch": 1.42741049857237, + "grad_norm": 2.1728579998016357, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7195702195167542, + "num_tokens": 331281428.0, + "step": 12998 + }, + { + "epoch": 1.4275203162749834, + "grad_norm": 2.4325571060180664, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7083793878555298, + "num_tokens": 331305153.0, + "step": 12999 + }, + { + "epoch": 1.4276301339775972, + "grad_norm": 2.2691497802734375, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7295169830322266, + "num_tokens": 331329150.0, + "step": 13000 + }, + { + "epoch": 1.427739951680211, + "grad_norm": 2.0999293327331543, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7154011130332947, + "num_tokens": 331356673.0, + "step": 13001 + }, + { + "epoch": 1.4278497693828245, + "grad_norm": 2.2362992763519287, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7336292862892151, + "num_tokens": 331380753.0, + "step": 13002 + }, + { + "epoch": 1.4279595870854382, + "grad_norm": 2.277899980545044, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7292352318763733, + "num_tokens": 331404194.0, + "step": 13003 + }, + { + "epoch": 1.4280694047880518, + "grad_norm": 2.5047717094421387, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.702919065952301, + "num_tokens": 331427116.0, + "step": 13004 + }, + { + "epoch": 1.4281792224906655, + "grad_norm": 2.094998836517334, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7074265480041504, + "num_tokens": 331455624.0, + "step": 13005 + }, + { + "epoch": 1.4282890401932793, + "grad_norm": 2.3925745487213135, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7028167247772217, + "num_tokens": 331480186.0, + "step": 13006 + }, + { + "epoch": 1.4283988578958928, + "grad_norm": 2.0006723403930664, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7122089862823486, + "num_tokens": 331510489.0, + "step": 13007 + }, + { + "epoch": 1.4285086755985064, + "grad_norm": 2.3819637298583984, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7223074436187744, + "num_tokens": 331533106.0, + "step": 13008 + }, + { + "epoch": 1.4286184933011201, + "grad_norm": 2.142993927001953, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7137067914009094, + "num_tokens": 331561139.0, + "step": 13009 + }, + { + "epoch": 1.4287283110037339, + "grad_norm": 2.262369394302368, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.6999415159225464, + "num_tokens": 331587257.0, + "step": 13010 + }, + { + "epoch": 1.4288381287063474, + "grad_norm": 2.2129578590393066, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7263236045837402, + "num_tokens": 331613703.0, + "step": 13011 + }, + { + "epoch": 1.4289479464089612, + "grad_norm": 2.1914470195770264, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7144238948822021, + "num_tokens": 331639362.0, + "step": 13012 + }, + { + "epoch": 1.4290577641115747, + "grad_norm": 2.1311841011047363, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6967072486877441, + "num_tokens": 331669323.0, + "step": 13013 + }, + { + "epoch": 1.4291675818141885, + "grad_norm": 2.3281519412994385, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.707168698310852, + "num_tokens": 331694639.0, + "step": 13014 + }, + { + "epoch": 1.4292773995168022, + "grad_norm": 2.341496706008911, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7140523195266724, + "num_tokens": 331717424.0, + "step": 13015 + }, + { + "epoch": 1.4293872172194158, + "grad_norm": 2.333091974258423, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7174071073532104, + "num_tokens": 331742731.0, + "step": 13016 + }, + { + "epoch": 1.4294970349220295, + "grad_norm": 2.196166753768921, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7105817794799805, + "num_tokens": 331771465.0, + "step": 13017 + }, + { + "epoch": 1.429606852624643, + "grad_norm": 2.5444533824920654, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7326153516769409, + "num_tokens": 331792058.0, + "step": 13018 + }, + { + "epoch": 1.4297166703272568, + "grad_norm": 2.4343628883361816, + "learning_rate": 1e-06, + "loss": 0.7883, + "mean_token_accuracy": 0.7454547882080078, + "num_tokens": 331813197.0, + "step": 13019 + }, + { + "epoch": 1.4298264880298703, + "grad_norm": 2.0692427158355713, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7184664011001587, + "num_tokens": 331842188.0, + "step": 13020 + }, + { + "epoch": 1.429936305732484, + "grad_norm": 2.4837586879730225, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7166063785552979, + "num_tokens": 331865004.0, + "step": 13021 + }, + { + "epoch": 1.4300461234350976, + "grad_norm": 2.0641770362854004, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.720323920249939, + "num_tokens": 331895246.0, + "step": 13022 + }, + { + "epoch": 1.4301559411377114, + "grad_norm": 2.3497238159179688, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7033883333206177, + "num_tokens": 331921157.0, + "step": 13023 + }, + { + "epoch": 1.4302657588403251, + "grad_norm": 2.4488625526428223, + "learning_rate": 1e-06, + "loss": 0.7646, + "mean_token_accuracy": 0.7567394971847534, + "num_tokens": 331940689.0, + "step": 13024 + }, + { + "epoch": 1.4303755765429387, + "grad_norm": 2.414562702178955, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7217037677764893, + "num_tokens": 331964182.0, + "step": 13025 + }, + { + "epoch": 1.4304853942455524, + "grad_norm": 2.1346523761749268, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7114596962928772, + "num_tokens": 331992033.0, + "step": 13026 + }, + { + "epoch": 1.430595211948166, + "grad_norm": 2.2774176597595215, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7352199554443359, + "num_tokens": 332015954.0, + "step": 13027 + }, + { + "epoch": 1.4307050296507797, + "grad_norm": 2.1644740104675293, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7247799634933472, + "num_tokens": 332043195.0, + "step": 13028 + }, + { + "epoch": 1.4308148473533935, + "grad_norm": 2.0291638374328613, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6956741213798523, + "num_tokens": 332075793.0, + "step": 13029 + }, + { + "epoch": 1.430924665056007, + "grad_norm": 2.0539872646331787, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.70604008436203, + "num_tokens": 332105478.0, + "step": 13030 + }, + { + "epoch": 1.4310344827586206, + "grad_norm": 2.4202487468719482, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7196466326713562, + "num_tokens": 332128174.0, + "step": 13031 + }, + { + "epoch": 1.4311443004612343, + "grad_norm": 2.094456195831299, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7333426475524902, + "num_tokens": 332155567.0, + "step": 13032 + }, + { + "epoch": 1.431254118163848, + "grad_norm": 2.165027379989624, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.706510603427887, + "num_tokens": 332182654.0, + "step": 13033 + }, + { + "epoch": 1.4313639358664616, + "grad_norm": 2.2512319087982178, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7083319425582886, + "num_tokens": 332206518.0, + "step": 13034 + }, + { + "epoch": 1.4314737535690754, + "grad_norm": 2.4657630920410156, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7218377590179443, + "num_tokens": 332229678.0, + "step": 13035 + }, + { + "epoch": 1.431583571271689, + "grad_norm": 2.4345431327819824, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7272568941116333, + "num_tokens": 332253012.0, + "step": 13036 + }, + { + "epoch": 1.4316933889743026, + "grad_norm": 2.203583002090454, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7172138690948486, + "num_tokens": 332281769.0, + "step": 13037 + }, + { + "epoch": 1.4318032066769164, + "grad_norm": 2.086425304412842, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7373861074447632, + "num_tokens": 332310115.0, + "step": 13038 + }, + { + "epoch": 1.43191302437953, + "grad_norm": 2.21232271194458, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6987857818603516, + "num_tokens": 332336412.0, + "step": 13039 + }, + { + "epoch": 1.4320228420821437, + "grad_norm": 2.1174726486206055, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7307851910591125, + "num_tokens": 332364268.0, + "step": 13040 + }, + { + "epoch": 1.4321326597847572, + "grad_norm": 2.215026378631592, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.6951810121536255, + "num_tokens": 332392938.0, + "step": 13041 + }, + { + "epoch": 1.432242477487371, + "grad_norm": 2.198540449142456, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7207198143005371, + "num_tokens": 332420211.0, + "step": 13042 + }, + { + "epoch": 1.4323522951899847, + "grad_norm": 2.162909746170044, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7372359037399292, + "num_tokens": 332445975.0, + "step": 13043 + }, + { + "epoch": 1.4324621128925983, + "grad_norm": 2.350297689437866, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7144140005111694, + "num_tokens": 332469717.0, + "step": 13044 + }, + { + "epoch": 1.4325719305952118, + "grad_norm": 2.327340602874756, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7253115177154541, + "num_tokens": 332492071.0, + "step": 13045 + }, + { + "epoch": 1.4326817482978256, + "grad_norm": 1.7891347408294678, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7198253870010376, + "num_tokens": 332527100.0, + "step": 13046 + }, + { + "epoch": 1.4327915660004393, + "grad_norm": 2.130427598953247, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7033201456069946, + "num_tokens": 332553612.0, + "step": 13047 + }, + { + "epoch": 1.4329013837030529, + "grad_norm": 2.308161735534668, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7334160208702087, + "num_tokens": 332578530.0, + "step": 13048 + }, + { + "epoch": 1.4330112014056666, + "grad_norm": 2.5840089321136475, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.715216875076294, + "num_tokens": 332599742.0, + "step": 13049 + }, + { + "epoch": 1.4331210191082802, + "grad_norm": 1.8969316482543945, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7238783836364746, + "num_tokens": 332634208.0, + "step": 13050 + }, + { + "epoch": 1.433230836810894, + "grad_norm": 2.310452699661255, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7255141735076904, + "num_tokens": 332658845.0, + "step": 13051 + }, + { + "epoch": 1.4333406545135077, + "grad_norm": 2.2441444396972656, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7125730514526367, + "num_tokens": 332683566.0, + "step": 13052 + }, + { + "epoch": 1.4334504722161212, + "grad_norm": 2.3497555255889893, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7331886291503906, + "num_tokens": 332707762.0, + "step": 13053 + }, + { + "epoch": 1.433560289918735, + "grad_norm": 2.6273577213287354, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7376583814620972, + "num_tokens": 332725894.0, + "step": 13054 + }, + { + "epoch": 1.4336701076213485, + "grad_norm": 2.2509453296661377, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7307637929916382, + "num_tokens": 332749998.0, + "step": 13055 + }, + { + "epoch": 1.4337799253239623, + "grad_norm": 2.3448903560638428, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7089228630065918, + "num_tokens": 332774632.0, + "step": 13056 + }, + { + "epoch": 1.433889743026576, + "grad_norm": 2.2996251583099365, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7260313034057617, + "num_tokens": 332798895.0, + "step": 13057 + }, + { + "epoch": 1.4339995607291895, + "grad_norm": 2.399092435836792, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7123927474021912, + "num_tokens": 332821943.0, + "step": 13058 + }, + { + "epoch": 1.434109378431803, + "grad_norm": 2.295236587524414, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7167649865150452, + "num_tokens": 332846790.0, + "step": 13059 + }, + { + "epoch": 1.4342191961344168, + "grad_norm": 2.2589786052703857, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7106847763061523, + "num_tokens": 332875422.0, + "step": 13060 + }, + { + "epoch": 1.4343290138370306, + "grad_norm": 2.459630250930786, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7175152897834778, + "num_tokens": 332899820.0, + "step": 13061 + }, + { + "epoch": 1.4344388315396441, + "grad_norm": 2.479719877243042, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.726061224937439, + "num_tokens": 332923944.0, + "step": 13062 + }, + { + "epoch": 1.434548649242258, + "grad_norm": 2.362968683242798, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7019284963607788, + "num_tokens": 332949830.0, + "step": 13063 + }, + { + "epoch": 1.4346584669448714, + "grad_norm": 2.5282890796661377, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7341031432151794, + "num_tokens": 332968227.0, + "step": 13064 + }, + { + "epoch": 1.4347682846474852, + "grad_norm": 2.3376410007476807, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7176063656806946, + "num_tokens": 332992585.0, + "step": 13065 + }, + { + "epoch": 1.434878102350099, + "grad_norm": 2.3271656036376953, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7224690914154053, + "num_tokens": 333017435.0, + "step": 13066 + }, + { + "epoch": 1.4349879200527125, + "grad_norm": 2.3162922859191895, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7109378576278687, + "num_tokens": 333040743.0, + "step": 13067 + }, + { + "epoch": 1.4350977377553262, + "grad_norm": 2.263896942138672, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7197361588478088, + "num_tokens": 333065281.0, + "step": 13068 + }, + { + "epoch": 1.4352075554579398, + "grad_norm": 2.8725407123565674, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7473210692405701, + "num_tokens": 333083852.0, + "step": 13069 + }, + { + "epoch": 1.4353173731605535, + "grad_norm": 2.0409657955169678, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7309449911117554, + "num_tokens": 333113091.0, + "step": 13070 + }, + { + "epoch": 1.4354271908631673, + "grad_norm": 2.110302686691284, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7124955654144287, + "num_tokens": 333143418.0, + "step": 13071 + }, + { + "epoch": 1.4355370085657808, + "grad_norm": 2.4273643493652344, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7205830812454224, + "num_tokens": 333166710.0, + "step": 13072 + }, + { + "epoch": 1.4356468262683943, + "grad_norm": 2.494964122772217, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7205464243888855, + "num_tokens": 333189953.0, + "step": 13073 + }, + { + "epoch": 1.435756643971008, + "grad_norm": 2.3471498489379883, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7380334734916687, + "num_tokens": 333210621.0, + "step": 13074 + }, + { + "epoch": 1.4358664616736219, + "grad_norm": 2.209728956222534, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7086853981018066, + "num_tokens": 333237458.0, + "step": 13075 + }, + { + "epoch": 1.4359762793762354, + "grad_norm": 2.002981424331665, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7314306497573853, + "num_tokens": 333269019.0, + "step": 13076 + }, + { + "epoch": 1.4360860970788492, + "grad_norm": 2.058931827545166, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.6898976564407349, + "num_tokens": 333299619.0, + "step": 13077 + }, + { + "epoch": 1.4361959147814627, + "grad_norm": 2.601956367492676, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7302918434143066, + "num_tokens": 333320749.0, + "step": 13078 + }, + { + "epoch": 1.4363057324840764, + "grad_norm": 2.2002387046813965, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.706967294216156, + "num_tokens": 333347086.0, + "step": 13079 + }, + { + "epoch": 1.4364155501866902, + "grad_norm": 2.4916207790374756, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7218446731567383, + "num_tokens": 333369666.0, + "step": 13080 + }, + { + "epoch": 1.4365253678893037, + "grad_norm": 2.732407331466675, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7337062358856201, + "num_tokens": 333388784.0, + "step": 13081 + }, + { + "epoch": 1.4366351855919175, + "grad_norm": 2.61074161529541, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.724305272102356, + "num_tokens": 333407810.0, + "step": 13082 + }, + { + "epoch": 1.436745003294531, + "grad_norm": 2.4005672931671143, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.714144229888916, + "num_tokens": 333432952.0, + "step": 13083 + }, + { + "epoch": 1.4368548209971448, + "grad_norm": 2.171912431716919, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.708022952079773, + "num_tokens": 333460725.0, + "step": 13084 + }, + { + "epoch": 1.4369646386997583, + "grad_norm": 2.2001354694366455, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7260701060295105, + "num_tokens": 333486935.0, + "step": 13085 + }, + { + "epoch": 1.437074456402372, + "grad_norm": 2.274961471557617, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7235119342803955, + "num_tokens": 333513136.0, + "step": 13086 + }, + { + "epoch": 1.4371842741049856, + "grad_norm": 2.1612627506256104, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7090837359428406, + "num_tokens": 333538385.0, + "step": 13087 + }, + { + "epoch": 1.4372940918075994, + "grad_norm": 2.566303014755249, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7108535766601562, + "num_tokens": 333558733.0, + "step": 13088 + }, + { + "epoch": 1.4374039095102131, + "grad_norm": 2.192669153213501, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7365859150886536, + "num_tokens": 333584730.0, + "step": 13089 + }, + { + "epoch": 1.4375137272128267, + "grad_norm": 2.1615982055664062, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7206462025642395, + "num_tokens": 333613495.0, + "step": 13090 + }, + { + "epoch": 1.4376235449154404, + "grad_norm": 2.4160282611846924, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7389287948608398, + "num_tokens": 333636043.0, + "step": 13091 + }, + { + "epoch": 1.437733362618054, + "grad_norm": 2.416499376296997, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7157454490661621, + "num_tokens": 333658697.0, + "step": 13092 + }, + { + "epoch": 1.4378431803206677, + "grad_norm": 2.327294111251831, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7064903974533081, + "num_tokens": 333681630.0, + "step": 13093 + }, + { + "epoch": 1.4379529980232815, + "grad_norm": 2.136549949645996, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7174615263938904, + "num_tokens": 333710122.0, + "step": 13094 + }, + { + "epoch": 1.438062815725895, + "grad_norm": 2.1600399017333984, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7324165105819702, + "num_tokens": 333737263.0, + "step": 13095 + }, + { + "epoch": 1.4381726334285085, + "grad_norm": 2.318833827972412, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7239594459533691, + "num_tokens": 333760945.0, + "step": 13096 + }, + { + "epoch": 1.4382824511311223, + "grad_norm": 2.148787021636963, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7050671577453613, + "num_tokens": 333790198.0, + "step": 13097 + }, + { + "epoch": 1.438392268833736, + "grad_norm": 2.3548004627227783, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7143484354019165, + "num_tokens": 333813811.0, + "step": 13098 + }, + { + "epoch": 1.4385020865363496, + "grad_norm": 2.849470853805542, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7288177013397217, + "num_tokens": 333833251.0, + "step": 13099 + }, + { + "epoch": 1.4386119042389633, + "grad_norm": 2.3815157413482666, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.728691041469574, + "num_tokens": 333856138.0, + "step": 13100 + }, + { + "epoch": 1.4387217219415769, + "grad_norm": 2.2449800968170166, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.721183180809021, + "num_tokens": 333881079.0, + "step": 13101 + }, + { + "epoch": 1.4388315396441906, + "grad_norm": 2.135016441345215, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7189364433288574, + "num_tokens": 333909746.0, + "step": 13102 + }, + { + "epoch": 1.4389413573468044, + "grad_norm": 2.294177293777466, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7170133590698242, + "num_tokens": 333934650.0, + "step": 13103 + }, + { + "epoch": 1.439051175049418, + "grad_norm": 1.9436924457550049, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7392407655715942, + "num_tokens": 333965393.0, + "step": 13104 + }, + { + "epoch": 1.4391609927520317, + "grad_norm": 1.9194263219833374, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7395376563072205, + "num_tokens": 333997129.0, + "step": 13105 + }, + { + "epoch": 1.4392708104546452, + "grad_norm": 1.9360623359680176, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7121782302856445, + "num_tokens": 334030219.0, + "step": 13106 + }, + { + "epoch": 1.439380628157259, + "grad_norm": 2.1683993339538574, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.724004328250885, + "num_tokens": 334058862.0, + "step": 13107 + }, + { + "epoch": 1.4394904458598727, + "grad_norm": 2.045316696166992, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.7017413377761841, + "num_tokens": 334088115.0, + "step": 13108 + }, + { + "epoch": 1.4396002635624863, + "grad_norm": 2.115614891052246, + "learning_rate": 1e-06, + "loss": 1.0805, + "mean_token_accuracy": 0.6803076863288879, + "num_tokens": 334116306.0, + "step": 13109 + }, + { + "epoch": 1.4397100812650998, + "grad_norm": 2.1530418395996094, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7072871923446655, + "num_tokens": 334144163.0, + "step": 13110 + }, + { + "epoch": 1.4398198989677136, + "grad_norm": 2.1563780307769775, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7369859218597412, + "num_tokens": 334168427.0, + "step": 13111 + }, + { + "epoch": 1.4399297166703273, + "grad_norm": 2.1772735118865967, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7137464284896851, + "num_tokens": 334196060.0, + "step": 13112 + }, + { + "epoch": 1.4400395343729409, + "grad_norm": 2.203702688217163, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7148187160491943, + "num_tokens": 334221611.0, + "step": 13113 + }, + { + "epoch": 1.4401493520755546, + "grad_norm": 2.297146797180176, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7110801935195923, + "num_tokens": 334247782.0, + "step": 13114 + }, + { + "epoch": 1.4402591697781681, + "grad_norm": 2.1078596115112305, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.710256814956665, + "num_tokens": 334275184.0, + "step": 13115 + }, + { + "epoch": 1.440368987480782, + "grad_norm": 2.37199068069458, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7243088483810425, + "num_tokens": 334299089.0, + "step": 13116 + }, + { + "epoch": 1.4404788051833957, + "grad_norm": 2.4137582778930664, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.7380409240722656, + "num_tokens": 334320898.0, + "step": 13117 + }, + { + "epoch": 1.4405886228860092, + "grad_norm": 2.220344066619873, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.718246579170227, + "num_tokens": 334345694.0, + "step": 13118 + }, + { + "epoch": 1.440698440588623, + "grad_norm": 2.2111830711364746, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6974713802337646, + "num_tokens": 334374470.0, + "step": 13119 + }, + { + "epoch": 1.4408082582912365, + "grad_norm": 2.590238332748413, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.721343994140625, + "num_tokens": 334395547.0, + "step": 13120 + }, + { + "epoch": 1.4409180759938502, + "grad_norm": 2.3590962886810303, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7139867544174194, + "num_tokens": 334419911.0, + "step": 13121 + }, + { + "epoch": 1.441027893696464, + "grad_norm": 2.2102315425872803, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7110414505004883, + "num_tokens": 334445954.0, + "step": 13122 + }, + { + "epoch": 1.4411377113990775, + "grad_norm": 2.5235414505004883, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7298263311386108, + "num_tokens": 334466407.0, + "step": 13123 + }, + { + "epoch": 1.441247529101691, + "grad_norm": 2.1402251720428467, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7135501503944397, + "num_tokens": 334493473.0, + "step": 13124 + }, + { + "epoch": 1.4413573468043048, + "grad_norm": 2.366325616836548, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.7009783983230591, + "num_tokens": 334517860.0, + "step": 13125 + }, + { + "epoch": 1.4414671645069186, + "grad_norm": 2.0975565910339355, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7173832058906555, + "num_tokens": 334546297.0, + "step": 13126 + }, + { + "epoch": 1.4415769822095321, + "grad_norm": 2.5132408142089844, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7206515073776245, + "num_tokens": 334570368.0, + "step": 13127 + }, + { + "epoch": 1.4416867999121459, + "grad_norm": 2.178840398788452, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7062486410140991, + "num_tokens": 334596278.0, + "step": 13128 + }, + { + "epoch": 1.4417966176147594, + "grad_norm": 2.149216890335083, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7022873759269714, + "num_tokens": 334626832.0, + "step": 13129 + }, + { + "epoch": 1.4419064353173732, + "grad_norm": 2.345186948776245, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7336706519126892, + "num_tokens": 334650298.0, + "step": 13130 + }, + { + "epoch": 1.442016253019987, + "grad_norm": 2.1575891971588135, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7147432565689087, + "num_tokens": 334677335.0, + "step": 13131 + }, + { + "epoch": 1.4421260707226005, + "grad_norm": 2.2811083793640137, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7229380011558533, + "num_tokens": 334700270.0, + "step": 13132 + }, + { + "epoch": 1.4422358884252142, + "grad_norm": 2.3445470333099365, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7280478477478027, + "num_tokens": 334723658.0, + "step": 13133 + }, + { + "epoch": 1.4423457061278278, + "grad_norm": 2.230865240097046, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7070522308349609, + "num_tokens": 334750656.0, + "step": 13134 + }, + { + "epoch": 1.4424555238304415, + "grad_norm": 2.2272651195526123, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7286303043365479, + "num_tokens": 334775696.0, + "step": 13135 + }, + { + "epoch": 1.442565341533055, + "grad_norm": 2.1002109050750732, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7214558124542236, + "num_tokens": 334803641.0, + "step": 13136 + }, + { + "epoch": 1.4426751592356688, + "grad_norm": 2.223083019256592, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7053979635238647, + "num_tokens": 334829777.0, + "step": 13137 + }, + { + "epoch": 1.4427849769382823, + "grad_norm": 2.4178218841552734, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7252349853515625, + "num_tokens": 334851852.0, + "step": 13138 + }, + { + "epoch": 1.442894794640896, + "grad_norm": 2.17216157913208, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7354756593704224, + "num_tokens": 334878291.0, + "step": 13139 + }, + { + "epoch": 1.4430046123435099, + "grad_norm": 2.3897719383239746, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7204715609550476, + "num_tokens": 334901735.0, + "step": 13140 + }, + { + "epoch": 1.4431144300461234, + "grad_norm": 2.6586318016052246, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7324529886245728, + "num_tokens": 334920767.0, + "step": 13141 + }, + { + "epoch": 1.4432242477487371, + "grad_norm": 2.0001254081726074, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.6984982490539551, + "num_tokens": 334954596.0, + "step": 13142 + }, + { + "epoch": 1.4433340654513507, + "grad_norm": 2.5117275714874268, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.723766565322876, + "num_tokens": 334976453.0, + "step": 13143 + }, + { + "epoch": 1.4434438831539644, + "grad_norm": 2.3899354934692383, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7195096015930176, + "num_tokens": 335001370.0, + "step": 13144 + }, + { + "epoch": 1.4435537008565782, + "grad_norm": 1.9309860467910767, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7067886590957642, + "num_tokens": 335037861.0, + "step": 13145 + }, + { + "epoch": 1.4436635185591917, + "grad_norm": 2.310903549194336, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7089946269989014, + "num_tokens": 335063427.0, + "step": 13146 + }, + { + "epoch": 1.4437733362618055, + "grad_norm": 2.642263889312744, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7525445818901062, + "num_tokens": 335081308.0, + "step": 13147 + }, + { + "epoch": 1.443883153964419, + "grad_norm": 2.070444345474243, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.6918961405754089, + "num_tokens": 335109341.0, + "step": 13148 + }, + { + "epoch": 1.4439929716670328, + "grad_norm": 2.086850166320801, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7335418462753296, + "num_tokens": 335137649.0, + "step": 13149 + }, + { + "epoch": 1.4441027893696463, + "grad_norm": 2.3664743900299072, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7066382169723511, + "num_tokens": 335161193.0, + "step": 13150 + }, + { + "epoch": 1.44421260707226, + "grad_norm": 2.0818052291870117, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7261511087417603, + "num_tokens": 335188916.0, + "step": 13151 + }, + { + "epoch": 1.4443224247748736, + "grad_norm": 2.0870442390441895, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7222508192062378, + "num_tokens": 335217770.0, + "step": 13152 + }, + { + "epoch": 1.4444322424774874, + "grad_norm": 2.307488203048706, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7298202514648438, + "num_tokens": 335242516.0, + "step": 13153 + }, + { + "epoch": 1.4445420601801011, + "grad_norm": 2.7556307315826416, + "learning_rate": 1e-06, + "loss": 0.7664, + "mean_token_accuracy": 0.7481775283813477, + "num_tokens": 335260649.0, + "step": 13154 + }, + { + "epoch": 1.4446518778827147, + "grad_norm": 2.4901702404022217, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7231162190437317, + "num_tokens": 335283295.0, + "step": 13155 + }, + { + "epoch": 1.4447616955853284, + "grad_norm": 2.038496255874634, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.694026529788971, + "num_tokens": 335314152.0, + "step": 13156 + }, + { + "epoch": 1.444871513287942, + "grad_norm": 2.3927652835845947, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7347964644432068, + "num_tokens": 335336546.0, + "step": 13157 + }, + { + "epoch": 1.4449813309905557, + "grad_norm": 2.1381728649139404, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.732901930809021, + "num_tokens": 335365070.0, + "step": 13158 + }, + { + "epoch": 1.4450911486931695, + "grad_norm": 2.565519332885742, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7404108643531799, + "num_tokens": 335387660.0, + "step": 13159 + }, + { + "epoch": 1.445200966395783, + "grad_norm": 2.124708890914917, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7319988012313843, + "num_tokens": 335413677.0, + "step": 13160 + }, + { + "epoch": 1.4453107840983965, + "grad_norm": 2.2327444553375244, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7404814958572388, + "num_tokens": 335437500.0, + "step": 13161 + }, + { + "epoch": 1.4454206018010103, + "grad_norm": 2.4614462852478027, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7328767776489258, + "num_tokens": 335460435.0, + "step": 13162 + }, + { + "epoch": 1.445530419503624, + "grad_norm": 2.232300043106079, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.728651225566864, + "num_tokens": 335488231.0, + "step": 13163 + }, + { + "epoch": 1.4456402372062376, + "grad_norm": 2.109208345413208, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7277922630310059, + "num_tokens": 335518310.0, + "step": 13164 + }, + { + "epoch": 1.4457500549088513, + "grad_norm": 2.331380605697632, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7324331998825073, + "num_tokens": 335542702.0, + "step": 13165 + }, + { + "epoch": 1.4458598726114649, + "grad_norm": 2.6451780796051025, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7440510392189026, + "num_tokens": 335562095.0, + "step": 13166 + }, + { + "epoch": 1.4459696903140786, + "grad_norm": 2.337259531021118, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7254924774169922, + "num_tokens": 335585583.0, + "step": 13167 + }, + { + "epoch": 1.4460795080166924, + "grad_norm": 2.2861952781677246, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7133839726448059, + "num_tokens": 335611214.0, + "step": 13168 + }, + { + "epoch": 1.446189325719306, + "grad_norm": 2.1744604110717773, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7265224456787109, + "num_tokens": 335636407.0, + "step": 13169 + }, + { + "epoch": 1.4462991434219197, + "grad_norm": 2.1825902462005615, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7060543298721313, + "num_tokens": 335664699.0, + "step": 13170 + }, + { + "epoch": 1.4464089611245332, + "grad_norm": 2.513794422149658, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7091202735900879, + "num_tokens": 335686106.0, + "step": 13171 + }, + { + "epoch": 1.446518778827147, + "grad_norm": 2.389357805252075, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.727419376373291, + "num_tokens": 335709422.0, + "step": 13172 + }, + { + "epoch": 1.4466285965297607, + "grad_norm": 2.156346082687378, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7338518500328064, + "num_tokens": 335735535.0, + "step": 13173 + }, + { + "epoch": 1.4467384142323743, + "grad_norm": 2.6447207927703857, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7438924908638, + "num_tokens": 335753568.0, + "step": 13174 + }, + { + "epoch": 1.4468482319349878, + "grad_norm": 2.207035779953003, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.711604118347168, + "num_tokens": 335783609.0, + "step": 13175 + }, + { + "epoch": 1.4469580496376016, + "grad_norm": 2.1292014122009277, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7334872484207153, + "num_tokens": 335813419.0, + "step": 13176 + }, + { + "epoch": 1.4470678673402153, + "grad_norm": 2.363755464553833, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7129175662994385, + "num_tokens": 335835955.0, + "step": 13177 + }, + { + "epoch": 1.4471776850428288, + "grad_norm": 2.2304039001464844, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7269811630249023, + "num_tokens": 335861110.0, + "step": 13178 + }, + { + "epoch": 1.4472875027454426, + "grad_norm": 2.6509830951690674, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7263773679733276, + "num_tokens": 335881637.0, + "step": 13179 + }, + { + "epoch": 1.4473973204480561, + "grad_norm": 2.1690783500671387, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7149040699005127, + "num_tokens": 335910118.0, + "step": 13180 + }, + { + "epoch": 1.44750713815067, + "grad_norm": 2.713020086288452, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7521148324012756, + "num_tokens": 335929455.0, + "step": 13181 + }, + { + "epoch": 1.4476169558532836, + "grad_norm": 2.3490490913391113, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7388540506362915, + "num_tokens": 335951375.0, + "step": 13182 + }, + { + "epoch": 1.4477267735558972, + "grad_norm": 2.2739744186401367, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7121467590332031, + "num_tokens": 335976997.0, + "step": 13183 + }, + { + "epoch": 1.447836591258511, + "grad_norm": 2.1729631423950195, + "learning_rate": 1e-06, + "loss": 0.7783, + "mean_token_accuracy": 0.7477329969406128, + "num_tokens": 336002474.0, + "step": 13184 + }, + { + "epoch": 1.4479464089611245, + "grad_norm": 2.0569679737091064, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.741245687007904, + "num_tokens": 336029912.0, + "step": 13185 + }, + { + "epoch": 1.4480562266637382, + "grad_norm": 2.2776055335998535, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7229995131492615, + "num_tokens": 336056106.0, + "step": 13186 + }, + { + "epoch": 1.448166044366352, + "grad_norm": 2.313074827194214, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7315239906311035, + "num_tokens": 336079916.0, + "step": 13187 + }, + { + "epoch": 1.4482758620689655, + "grad_norm": 2.2657604217529297, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7125712633132935, + "num_tokens": 336107003.0, + "step": 13188 + }, + { + "epoch": 1.448385679771579, + "grad_norm": 2.009427070617676, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.708948016166687, + "num_tokens": 336135517.0, + "step": 13189 + }, + { + "epoch": 1.4484954974741928, + "grad_norm": 2.163860321044922, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.727506160736084, + "num_tokens": 336163104.0, + "step": 13190 + }, + { + "epoch": 1.4486053151768066, + "grad_norm": 2.145214319229126, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7045103907585144, + "num_tokens": 336189320.0, + "step": 13191 + }, + { + "epoch": 1.44871513287942, + "grad_norm": 2.2493526935577393, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7281317114830017, + "num_tokens": 336214132.0, + "step": 13192 + }, + { + "epoch": 1.4488249505820339, + "grad_norm": 2.256558895111084, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7278714776039124, + "num_tokens": 336239184.0, + "step": 13193 + }, + { + "epoch": 1.4489347682846474, + "grad_norm": 2.1595852375030518, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6902594566345215, + "num_tokens": 336268177.0, + "step": 13194 + }, + { + "epoch": 1.4490445859872612, + "grad_norm": 2.030731678009033, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7043310403823853, + "num_tokens": 336299358.0, + "step": 13195 + }, + { + "epoch": 1.449154403689875, + "grad_norm": 2.3990511894226074, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7221951484680176, + "num_tokens": 336322277.0, + "step": 13196 + }, + { + "epoch": 1.4492642213924884, + "grad_norm": 2.1392054557800293, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7086703777313232, + "num_tokens": 336349593.0, + "step": 13197 + }, + { + "epoch": 1.4493740390951022, + "grad_norm": 2.1750378608703613, + "learning_rate": 1e-06, + "loss": 0.8266, + "mean_token_accuracy": 0.7423953413963318, + "num_tokens": 336377362.0, + "step": 13198 + }, + { + "epoch": 1.4494838567977157, + "grad_norm": 2.2057433128356934, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.713119387626648, + "num_tokens": 336403500.0, + "step": 13199 + }, + { + "epoch": 1.4495936745003295, + "grad_norm": 1.9949135780334473, + "learning_rate": 1e-06, + "loss": 0.7859, + "mean_token_accuracy": 0.7595053911209106, + "num_tokens": 336431438.0, + "step": 13200 + }, + { + "epoch": 1.449703492202943, + "grad_norm": 2.363255739212036, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7284888029098511, + "num_tokens": 336455014.0, + "step": 13201 + }, + { + "epoch": 1.4498133099055568, + "grad_norm": 2.2527542114257812, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.6980912685394287, + "num_tokens": 336480785.0, + "step": 13202 + }, + { + "epoch": 1.4499231276081703, + "grad_norm": 2.112393856048584, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7015711069107056, + "num_tokens": 336509584.0, + "step": 13203 + }, + { + "epoch": 1.450032945310784, + "grad_norm": 2.0908403396606445, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7171510457992554, + "num_tokens": 336538494.0, + "step": 13204 + }, + { + "epoch": 1.4501427630133978, + "grad_norm": 2.3318824768066406, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7040365934371948, + "num_tokens": 336562806.0, + "step": 13205 + }, + { + "epoch": 1.4502525807160114, + "grad_norm": 2.5128586292266846, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7505232691764832, + "num_tokens": 336587146.0, + "step": 13206 + }, + { + "epoch": 1.4503623984186251, + "grad_norm": 2.235792875289917, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6940574049949646, + "num_tokens": 336613462.0, + "step": 13207 + }, + { + "epoch": 1.4504722161212387, + "grad_norm": 2.0349905490875244, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7303561568260193, + "num_tokens": 336642899.0, + "step": 13208 + }, + { + "epoch": 1.4505820338238524, + "grad_norm": 2.0419857501983643, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7305488586425781, + "num_tokens": 336671884.0, + "step": 13209 + }, + { + "epoch": 1.4506918515264662, + "grad_norm": 2.3400611877441406, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7293047904968262, + "num_tokens": 336694653.0, + "step": 13210 + }, + { + "epoch": 1.4508016692290797, + "grad_norm": 2.258816719055176, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7178082466125488, + "num_tokens": 336719482.0, + "step": 13211 + }, + { + "epoch": 1.4509114869316933, + "grad_norm": 2.2609875202178955, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7097787857055664, + "num_tokens": 336746801.0, + "step": 13212 + }, + { + "epoch": 1.451021304634307, + "grad_norm": 2.314107894897461, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7182146906852722, + "num_tokens": 336771683.0, + "step": 13213 + }, + { + "epoch": 1.4511311223369208, + "grad_norm": 2.1852478981018066, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7225033044815063, + "num_tokens": 336799042.0, + "step": 13214 + }, + { + "epoch": 1.4512409400395343, + "grad_norm": 2.076503038406372, + "learning_rate": 1e-06, + "loss": 1.0562, + "mean_token_accuracy": 0.6794525384902954, + "num_tokens": 336828961.0, + "step": 13215 + }, + { + "epoch": 1.451350757742148, + "grad_norm": 2.4352970123291016, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7266116142272949, + "num_tokens": 336850829.0, + "step": 13216 + }, + { + "epoch": 1.4514605754447616, + "grad_norm": 2.0595309734344482, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7206443548202515, + "num_tokens": 336879649.0, + "step": 13217 + }, + { + "epoch": 1.4515703931473753, + "grad_norm": 2.3507330417633057, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7227470278739929, + "num_tokens": 336903119.0, + "step": 13218 + }, + { + "epoch": 1.451680210849989, + "grad_norm": 2.3351714611053467, + "learning_rate": 1e-06, + "loss": 0.7452, + "mean_token_accuracy": 0.7579430341720581, + "num_tokens": 336925791.0, + "step": 13219 + }, + { + "epoch": 1.4517900285526026, + "grad_norm": 2.6242337226867676, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7206103205680847, + "num_tokens": 336946481.0, + "step": 13220 + }, + { + "epoch": 1.4518998462552164, + "grad_norm": 2.1675405502319336, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7167009711265564, + "num_tokens": 336973349.0, + "step": 13221 + }, + { + "epoch": 1.45200966395783, + "grad_norm": 2.4796249866485596, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6903579235076904, + "num_tokens": 336997229.0, + "step": 13222 + }, + { + "epoch": 1.4521194816604437, + "grad_norm": 2.3087117671966553, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7119808197021484, + "num_tokens": 337021365.0, + "step": 13223 + }, + { + "epoch": 1.4522292993630574, + "grad_norm": 2.1049282550811768, + "learning_rate": 1e-06, + "loss": 1.0722, + "mean_token_accuracy": 0.6782163381576538, + "num_tokens": 337050700.0, + "step": 13224 + }, + { + "epoch": 1.452339117065671, + "grad_norm": 2.1621248722076416, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.6963831186294556, + "num_tokens": 337079488.0, + "step": 13225 + }, + { + "epoch": 1.4524489347682845, + "grad_norm": 2.591980218887329, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7273829579353333, + "num_tokens": 337101133.0, + "step": 13226 + }, + { + "epoch": 1.4525587524708983, + "grad_norm": 2.3349051475524902, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7326589822769165, + "num_tokens": 337122627.0, + "step": 13227 + }, + { + "epoch": 1.452668570173512, + "grad_norm": 2.046527624130249, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7258976697921753, + "num_tokens": 337152421.0, + "step": 13228 + }, + { + "epoch": 1.4527783878761256, + "grad_norm": 2.2077085971832275, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.719141960144043, + "num_tokens": 337179273.0, + "step": 13229 + }, + { + "epoch": 1.4528882055787393, + "grad_norm": 2.3485498428344727, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7161555290222168, + "num_tokens": 337204361.0, + "step": 13230 + }, + { + "epoch": 1.4529980232813529, + "grad_norm": 2.442366123199463, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7127951383590698, + "num_tokens": 337226898.0, + "step": 13231 + }, + { + "epoch": 1.4531078409839666, + "grad_norm": 2.1741878986358643, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7018197774887085, + "num_tokens": 337253596.0, + "step": 13232 + }, + { + "epoch": 1.4532176586865804, + "grad_norm": 2.325838327407837, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7345775365829468, + "num_tokens": 337276335.0, + "step": 13233 + }, + { + "epoch": 1.453327476389194, + "grad_norm": 1.8909828662872314, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7281829714775085, + "num_tokens": 337307977.0, + "step": 13234 + }, + { + "epoch": 1.4534372940918077, + "grad_norm": 2.308558940887451, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7113910913467407, + "num_tokens": 337333565.0, + "step": 13235 + }, + { + "epoch": 1.4535471117944212, + "grad_norm": 2.025972366333008, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7311495542526245, + "num_tokens": 337362265.0, + "step": 13236 + }, + { + "epoch": 1.453656929497035, + "grad_norm": 2.0649733543395996, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.7022853493690491, + "num_tokens": 337393638.0, + "step": 13237 + }, + { + "epoch": 1.4537667471996487, + "grad_norm": 1.9418628215789795, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7298868894577026, + "num_tokens": 337427075.0, + "step": 13238 + }, + { + "epoch": 1.4538765649022622, + "grad_norm": 2.1070282459259033, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.6926977634429932, + "num_tokens": 337457173.0, + "step": 13239 + }, + { + "epoch": 1.4539863826048758, + "grad_norm": 2.155348539352417, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7307772636413574, + "num_tokens": 337484378.0, + "step": 13240 + }, + { + "epoch": 1.4540962003074895, + "grad_norm": 2.2451648712158203, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7233378887176514, + "num_tokens": 337510990.0, + "step": 13241 + }, + { + "epoch": 1.4542060180101033, + "grad_norm": 2.3315396308898926, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6989458203315735, + "num_tokens": 337536437.0, + "step": 13242 + }, + { + "epoch": 1.4543158357127168, + "grad_norm": 2.2978358268737793, + "learning_rate": 1e-06, + "loss": 0.8069, + "mean_token_accuracy": 0.7398771643638611, + "num_tokens": 337560614.0, + "step": 13243 + }, + { + "epoch": 1.4544256534153306, + "grad_norm": 2.2249741554260254, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7160714268684387, + "num_tokens": 337587121.0, + "step": 13244 + }, + { + "epoch": 1.4545354711179441, + "grad_norm": 2.348517894744873, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7341516613960266, + "num_tokens": 337611116.0, + "step": 13245 + }, + { + "epoch": 1.4546452888205579, + "grad_norm": 2.1293225288391113, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7186577916145325, + "num_tokens": 337640257.0, + "step": 13246 + }, + { + "epoch": 1.4547551065231716, + "grad_norm": 2.362799882888794, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.729894757270813, + "num_tokens": 337665015.0, + "step": 13247 + }, + { + "epoch": 1.4548649242257852, + "grad_norm": 2.330252170562744, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7116772532463074, + "num_tokens": 337689288.0, + "step": 13248 + }, + { + "epoch": 1.454974741928399, + "grad_norm": 2.028000831604004, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7103593945503235, + "num_tokens": 337720645.0, + "step": 13249 + }, + { + "epoch": 1.4550845596310125, + "grad_norm": 2.2871696949005127, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7152931690216064, + "num_tokens": 337745707.0, + "step": 13250 + }, + { + "epoch": 1.4551943773336262, + "grad_norm": 2.0601601600646973, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7223345637321472, + "num_tokens": 337774813.0, + "step": 13251 + }, + { + "epoch": 1.45530419503624, + "grad_norm": 2.1847927570343018, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7062157392501831, + "num_tokens": 337801556.0, + "step": 13252 + }, + { + "epoch": 1.4554140127388535, + "grad_norm": 2.285660743713379, + "learning_rate": 1e-06, + "loss": 0.7738, + "mean_token_accuracy": 0.7514805793762207, + "num_tokens": 337823871.0, + "step": 13253 + }, + { + "epoch": 1.455523830441467, + "grad_norm": 2.0143113136291504, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6976727247238159, + "num_tokens": 337855705.0, + "step": 13254 + }, + { + "epoch": 1.4556336481440808, + "grad_norm": 2.056408166885376, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7044787406921387, + "num_tokens": 337886486.0, + "step": 13255 + }, + { + "epoch": 1.4557434658466946, + "grad_norm": 2.3490121364593506, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.72409987449646, + "num_tokens": 337909150.0, + "step": 13256 + }, + { + "epoch": 1.455853283549308, + "grad_norm": 2.1349868774414062, + "learning_rate": 1e-06, + "loss": 0.8049, + "mean_token_accuracy": 0.7400193214416504, + "num_tokens": 337935387.0, + "step": 13257 + }, + { + "epoch": 1.4559631012519219, + "grad_norm": 2.314051866531372, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.717897891998291, + "num_tokens": 337962104.0, + "step": 13258 + }, + { + "epoch": 1.4560729189545354, + "grad_norm": 2.3746941089630127, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7337298393249512, + "num_tokens": 337985596.0, + "step": 13259 + }, + { + "epoch": 1.4561827366571491, + "grad_norm": 2.42356014251709, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7513805031776428, + "num_tokens": 338007821.0, + "step": 13260 + }, + { + "epoch": 1.456292554359763, + "grad_norm": 2.417215585708618, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7229559421539307, + "num_tokens": 338032358.0, + "step": 13261 + }, + { + "epoch": 1.4564023720623764, + "grad_norm": 2.2952349185943604, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7230778932571411, + "num_tokens": 338059046.0, + "step": 13262 + }, + { + "epoch": 1.4565121897649902, + "grad_norm": 2.6350204944610596, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7280749082565308, + "num_tokens": 338079589.0, + "step": 13263 + }, + { + "epoch": 1.4566220074676037, + "grad_norm": 2.2651007175445557, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7359112501144409, + "num_tokens": 338103689.0, + "step": 13264 + }, + { + "epoch": 1.4567318251702175, + "grad_norm": 2.283144474029541, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.6944344639778137, + "num_tokens": 338129019.0, + "step": 13265 + }, + { + "epoch": 1.456841642872831, + "grad_norm": 2.454477548599243, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.715451717376709, + "num_tokens": 338151725.0, + "step": 13266 + }, + { + "epoch": 1.4569514605754448, + "grad_norm": 2.2244067192077637, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7345407009124756, + "num_tokens": 338176014.0, + "step": 13267 + }, + { + "epoch": 1.4570612782780583, + "grad_norm": 1.8492127656936646, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.6943691968917847, + "num_tokens": 338214462.0, + "step": 13268 + }, + { + "epoch": 1.457171095980672, + "grad_norm": 2.2567901611328125, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7246609926223755, + "num_tokens": 338239613.0, + "step": 13269 + }, + { + "epoch": 1.4572809136832858, + "grad_norm": 2.1299777030944824, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7182735800743103, + "num_tokens": 338267797.0, + "step": 13270 + }, + { + "epoch": 1.4573907313858994, + "grad_norm": 2.2499444484710693, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7218169569969177, + "num_tokens": 338292363.0, + "step": 13271 + }, + { + "epoch": 1.4575005490885131, + "grad_norm": 2.125401496887207, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7148608565330505, + "num_tokens": 338319806.0, + "step": 13272 + }, + { + "epoch": 1.4576103667911267, + "grad_norm": 2.3406097888946533, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7054377794265747, + "num_tokens": 338346524.0, + "step": 13273 + }, + { + "epoch": 1.4577201844937404, + "grad_norm": 2.3246963024139404, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7059870958328247, + "num_tokens": 338372548.0, + "step": 13274 + }, + { + "epoch": 1.4578300021963542, + "grad_norm": 2.1202962398529053, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7280717492103577, + "num_tokens": 338401806.0, + "step": 13275 + }, + { + "epoch": 1.4579398198989677, + "grad_norm": 2.1233558654785156, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7258047461509705, + "num_tokens": 338428295.0, + "step": 13276 + }, + { + "epoch": 1.4580496376015812, + "grad_norm": 1.9592208862304688, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7159980535507202, + "num_tokens": 338461370.0, + "step": 13277 + }, + { + "epoch": 1.458159455304195, + "grad_norm": 2.0919759273529053, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7397699952125549, + "num_tokens": 338489709.0, + "step": 13278 + }, + { + "epoch": 1.4582692730068088, + "grad_norm": 2.205142021179199, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.6954160928726196, + "num_tokens": 338517129.0, + "step": 13279 + }, + { + "epoch": 1.4583790907094223, + "grad_norm": 2.482123374938965, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7223777770996094, + "num_tokens": 338539231.0, + "step": 13280 + }, + { + "epoch": 1.458488908412036, + "grad_norm": 2.0993173122406006, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7055677771568298, + "num_tokens": 338568306.0, + "step": 13281 + }, + { + "epoch": 1.4585987261146496, + "grad_norm": 2.3410844802856445, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.7057655453681946, + "num_tokens": 338592621.0, + "step": 13282 + }, + { + "epoch": 1.4587085438172633, + "grad_norm": 2.325198173522949, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7154618501663208, + "num_tokens": 338617144.0, + "step": 13283 + }, + { + "epoch": 1.458818361519877, + "grad_norm": 2.373356819152832, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7278838157653809, + "num_tokens": 338641120.0, + "step": 13284 + }, + { + "epoch": 1.4589281792224906, + "grad_norm": 2.8382368087768555, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7283394932746887, + "num_tokens": 338661523.0, + "step": 13285 + }, + { + "epoch": 1.4590379969251044, + "grad_norm": 2.545822858810425, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7196564078330994, + "num_tokens": 338683055.0, + "step": 13286 + }, + { + "epoch": 1.459147814627718, + "grad_norm": 2.374180555343628, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7005932927131653, + "num_tokens": 338708177.0, + "step": 13287 + }, + { + "epoch": 1.4592576323303317, + "grad_norm": 2.2327849864959717, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7195424437522888, + "num_tokens": 338733251.0, + "step": 13288 + }, + { + "epoch": 1.4593674500329454, + "grad_norm": 2.273671865463257, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7234802842140198, + "num_tokens": 338760933.0, + "step": 13289 + }, + { + "epoch": 1.459477267735559, + "grad_norm": 2.3035757541656494, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.7033272981643677, + "num_tokens": 338787338.0, + "step": 13290 + }, + { + "epoch": 1.4595870854381725, + "grad_norm": 2.0205414295196533, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.6963290572166443, + "num_tokens": 338817186.0, + "step": 13291 + }, + { + "epoch": 1.4596969031407863, + "grad_norm": 2.186923027038574, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.73111891746521, + "num_tokens": 338843649.0, + "step": 13292 + }, + { + "epoch": 1.4598067208434, + "grad_norm": 2.351067066192627, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7221869826316833, + "num_tokens": 338865795.0, + "step": 13293 + }, + { + "epoch": 1.4599165385460136, + "grad_norm": 2.0748932361602783, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.738288164138794, + "num_tokens": 338893213.0, + "step": 13294 + }, + { + "epoch": 1.4600263562486273, + "grad_norm": 2.5038836002349854, + "learning_rate": 1e-06, + "loss": 0.7935, + "mean_token_accuracy": 0.7498563528060913, + "num_tokens": 338911987.0, + "step": 13295 + }, + { + "epoch": 1.4601361739512408, + "grad_norm": 2.0482306480407715, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.6861464977264404, + "num_tokens": 338944320.0, + "step": 13296 + }, + { + "epoch": 1.4602459916538546, + "grad_norm": 2.141498565673828, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7126615047454834, + "num_tokens": 338972263.0, + "step": 13297 + }, + { + "epoch": 1.4603558093564684, + "grad_norm": 2.1795599460601807, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7295308709144592, + "num_tokens": 339001682.0, + "step": 13298 + }, + { + "epoch": 1.460465627059082, + "grad_norm": 1.8733452558517456, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7287732362747192, + "num_tokens": 339037895.0, + "step": 13299 + }, + { + "epoch": 1.4605754447616957, + "grad_norm": 2.482180595397949, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7312097549438477, + "num_tokens": 339059221.0, + "step": 13300 + }, + { + "epoch": 1.4606852624643092, + "grad_norm": 1.9934463500976562, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7179164290428162, + "num_tokens": 339088239.0, + "step": 13301 + }, + { + "epoch": 1.460795080166923, + "grad_norm": 2.4075303077697754, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7237148284912109, + "num_tokens": 339112601.0, + "step": 13302 + }, + { + "epoch": 1.4609048978695367, + "grad_norm": 2.2003190517425537, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7264248132705688, + "num_tokens": 339140303.0, + "step": 13303 + }, + { + "epoch": 1.4610147155721502, + "grad_norm": 2.483527183532715, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.726670503616333, + "num_tokens": 339161282.0, + "step": 13304 + }, + { + "epoch": 1.4611245332747638, + "grad_norm": 2.290344715118408, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7260870933532715, + "num_tokens": 339187021.0, + "step": 13305 + }, + { + "epoch": 1.4612343509773775, + "grad_norm": 2.237792491912842, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6936067938804626, + "num_tokens": 339217044.0, + "step": 13306 + }, + { + "epoch": 1.4613441686799913, + "grad_norm": 2.1637277603149414, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7210298776626587, + "num_tokens": 339244070.0, + "step": 13307 + }, + { + "epoch": 1.4614539863826048, + "grad_norm": 2.0866174697875977, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7397297024726868, + "num_tokens": 339271712.0, + "step": 13308 + }, + { + "epoch": 1.4615638040852186, + "grad_norm": 2.558378219604492, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.7466787099838257, + "num_tokens": 339290747.0, + "step": 13309 + }, + { + "epoch": 1.4616736217878321, + "grad_norm": 2.181142807006836, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7142047882080078, + "num_tokens": 339315518.0, + "step": 13310 + }, + { + "epoch": 1.4617834394904459, + "grad_norm": 2.051456928253174, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7027424573898315, + "num_tokens": 339345891.0, + "step": 13311 + }, + { + "epoch": 1.4618932571930596, + "grad_norm": 2.1703810691833496, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7165029048919678, + "num_tokens": 339373715.0, + "step": 13312 + }, + { + "epoch": 1.4620030748956732, + "grad_norm": 2.2508912086486816, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.71774822473526, + "num_tokens": 339400476.0, + "step": 13313 + }, + { + "epoch": 1.462112892598287, + "grad_norm": 2.4096083641052246, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6925638318061829, + "num_tokens": 339424176.0, + "step": 13314 + }, + { + "epoch": 1.4622227103009005, + "grad_norm": 2.2652740478515625, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7176869511604309, + "num_tokens": 339448951.0, + "step": 13315 + }, + { + "epoch": 1.4623325280035142, + "grad_norm": 2.368536949157715, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7184419631958008, + "num_tokens": 339472573.0, + "step": 13316 + }, + { + "epoch": 1.462442345706128, + "grad_norm": 2.3983101844787598, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7198209762573242, + "num_tokens": 339496853.0, + "step": 13317 + }, + { + "epoch": 1.4625521634087415, + "grad_norm": 2.369256019592285, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7115716338157654, + "num_tokens": 339521282.0, + "step": 13318 + }, + { + "epoch": 1.462661981111355, + "grad_norm": 2.4029910564422607, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7156386971473694, + "num_tokens": 339546377.0, + "step": 13319 + }, + { + "epoch": 1.4627717988139688, + "grad_norm": 2.253521680831909, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7235726118087769, + "num_tokens": 339574221.0, + "step": 13320 + }, + { + "epoch": 1.4628816165165826, + "grad_norm": 2.3956592082977295, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7307356595993042, + "num_tokens": 339596682.0, + "step": 13321 + }, + { + "epoch": 1.462991434219196, + "grad_norm": 2.043262481689453, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7103512287139893, + "num_tokens": 339629436.0, + "step": 13322 + }, + { + "epoch": 1.4631012519218098, + "grad_norm": 2.294863224029541, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7183324098587036, + "num_tokens": 339654057.0, + "step": 13323 + }, + { + "epoch": 1.4632110696244234, + "grad_norm": 2.413270950317383, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7450497150421143, + "num_tokens": 339677213.0, + "step": 13324 + }, + { + "epoch": 1.4633208873270371, + "grad_norm": 2.45471453666687, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7288302779197693, + "num_tokens": 339700701.0, + "step": 13325 + }, + { + "epoch": 1.463430705029651, + "grad_norm": 2.472581148147583, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7321812510490417, + "num_tokens": 339721778.0, + "step": 13326 + }, + { + "epoch": 1.4635405227322644, + "grad_norm": 2.2086336612701416, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7166801691055298, + "num_tokens": 339749253.0, + "step": 13327 + }, + { + "epoch": 1.4636503404348782, + "grad_norm": 2.1534407138824463, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7092471122741699, + "num_tokens": 339778181.0, + "step": 13328 + }, + { + "epoch": 1.4637601581374917, + "grad_norm": 2.2694413661956787, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7142960429191589, + "num_tokens": 339802764.0, + "step": 13329 + }, + { + "epoch": 1.4638699758401055, + "grad_norm": 2.4127719402313232, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7376800179481506, + "num_tokens": 339824836.0, + "step": 13330 + }, + { + "epoch": 1.463979793542719, + "grad_norm": 2.3135526180267334, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.716708242893219, + "num_tokens": 339851777.0, + "step": 13331 + }, + { + "epoch": 1.4640896112453328, + "grad_norm": 2.213325023651123, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7129663825035095, + "num_tokens": 339878416.0, + "step": 13332 + }, + { + "epoch": 1.4641994289479463, + "grad_norm": 2.5591177940368652, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7192308902740479, + "num_tokens": 339899715.0, + "step": 13333 + }, + { + "epoch": 1.46430924665056, + "grad_norm": 2.3558855056762695, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7351992726325989, + "num_tokens": 339923566.0, + "step": 13334 + }, + { + "epoch": 1.4644190643531738, + "grad_norm": 2.357255220413208, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7019035220146179, + "num_tokens": 339947944.0, + "step": 13335 + }, + { + "epoch": 1.4645288820557874, + "grad_norm": 2.5092720985412598, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7371081709861755, + "num_tokens": 339969037.0, + "step": 13336 + }, + { + "epoch": 1.464638699758401, + "grad_norm": 2.2085983753204346, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7277500629425049, + "num_tokens": 339994689.0, + "step": 13337 + }, + { + "epoch": 1.4647485174610146, + "grad_norm": 2.313568592071533, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7481805086135864, + "num_tokens": 340018160.0, + "step": 13338 + }, + { + "epoch": 1.4648583351636284, + "grad_norm": 2.1151015758514404, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7187214493751526, + "num_tokens": 340048036.0, + "step": 13339 + }, + { + "epoch": 1.4649681528662422, + "grad_norm": 2.5950231552124023, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7157834768295288, + "num_tokens": 340068850.0, + "step": 13340 + }, + { + "epoch": 1.4650779705688557, + "grad_norm": 2.6096065044403076, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7254812717437744, + "num_tokens": 340089906.0, + "step": 13341 + }, + { + "epoch": 1.4651877882714692, + "grad_norm": 2.5165624618530273, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7018049955368042, + "num_tokens": 340112382.0, + "step": 13342 + }, + { + "epoch": 1.465297605974083, + "grad_norm": 2.240706205368042, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.711277961730957, + "num_tokens": 340139054.0, + "step": 13343 + }, + { + "epoch": 1.4654074236766967, + "grad_norm": 2.17842698097229, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7226887941360474, + "num_tokens": 340167028.0, + "step": 13344 + }, + { + "epoch": 1.4655172413793103, + "grad_norm": 2.055828094482422, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6950286626815796, + "num_tokens": 340197824.0, + "step": 13345 + }, + { + "epoch": 1.465627059081924, + "grad_norm": 2.1148273944854736, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7227449417114258, + "num_tokens": 340225728.0, + "step": 13346 + }, + { + "epoch": 1.4657368767845376, + "grad_norm": 2.5061137676239014, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7457860708236694, + "num_tokens": 340245357.0, + "step": 13347 + }, + { + "epoch": 1.4658466944871513, + "grad_norm": 2.0793557167053223, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7149381637573242, + "num_tokens": 340273947.0, + "step": 13348 + }, + { + "epoch": 1.465956512189765, + "grad_norm": 2.167858123779297, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7216564416885376, + "num_tokens": 340300640.0, + "step": 13349 + }, + { + "epoch": 1.4660663298923786, + "grad_norm": 2.193957567214966, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7433808445930481, + "num_tokens": 340325905.0, + "step": 13350 + }, + { + "epoch": 1.4661761475949924, + "grad_norm": 1.9845640659332275, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7099822759628296, + "num_tokens": 340356459.0, + "step": 13351 + }, + { + "epoch": 1.466285965297606, + "grad_norm": 2.304589033126831, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7137347459793091, + "num_tokens": 340381841.0, + "step": 13352 + }, + { + "epoch": 1.4663957830002197, + "grad_norm": 2.4908487796783447, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.726502537727356, + "num_tokens": 340402970.0, + "step": 13353 + }, + { + "epoch": 1.4665056007028334, + "grad_norm": 2.19743275642395, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7203099727630615, + "num_tokens": 340431122.0, + "step": 13354 + }, + { + "epoch": 1.466615418405447, + "grad_norm": 2.177737236022949, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7348496913909912, + "num_tokens": 340459885.0, + "step": 13355 + }, + { + "epoch": 1.4667252361080605, + "grad_norm": 2.2566354274749756, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7206665277481079, + "num_tokens": 340486361.0, + "step": 13356 + }, + { + "epoch": 1.4668350538106742, + "grad_norm": 2.1698312759399414, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.6968759894371033, + "num_tokens": 340516735.0, + "step": 13357 + }, + { + "epoch": 1.466944871513288, + "grad_norm": 2.1793670654296875, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7119115591049194, + "num_tokens": 340543305.0, + "step": 13358 + }, + { + "epoch": 1.4670546892159015, + "grad_norm": 2.4521281719207764, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7404335141181946, + "num_tokens": 340564458.0, + "step": 13359 + }, + { + "epoch": 1.4671645069185153, + "grad_norm": 2.3270246982574463, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7086837887763977, + "num_tokens": 340588725.0, + "step": 13360 + }, + { + "epoch": 1.4672743246211288, + "grad_norm": 2.524771213531494, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7042303681373596, + "num_tokens": 340609560.0, + "step": 13361 + }, + { + "epoch": 1.4673841423237426, + "grad_norm": 2.3639814853668213, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7383120656013489, + "num_tokens": 340631065.0, + "step": 13362 + }, + { + "epoch": 1.4674939600263563, + "grad_norm": 2.0615389347076416, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.6942498087882996, + "num_tokens": 340659438.0, + "step": 13363 + }, + { + "epoch": 1.4676037777289699, + "grad_norm": 2.2416203022003174, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7265986204147339, + "num_tokens": 340684722.0, + "step": 13364 + }, + { + "epoch": 1.4677135954315836, + "grad_norm": 1.9512381553649902, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7282637357711792, + "num_tokens": 340715695.0, + "step": 13365 + }, + { + "epoch": 1.4678234131341972, + "grad_norm": 2.1736061573028564, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7007703185081482, + "num_tokens": 340743009.0, + "step": 13366 + }, + { + "epoch": 1.467933230836811, + "grad_norm": 2.3871757984161377, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7331584095954895, + "num_tokens": 340766546.0, + "step": 13367 + }, + { + "epoch": 1.4680430485394247, + "grad_norm": 2.1824982166290283, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7096960544586182, + "num_tokens": 340792445.0, + "step": 13368 + }, + { + "epoch": 1.4681528662420382, + "grad_norm": 2.593080997467041, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.702474057674408, + "num_tokens": 340813405.0, + "step": 13369 + }, + { + "epoch": 1.4682626839446518, + "grad_norm": 2.2636513710021973, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7230738401412964, + "num_tokens": 340839798.0, + "step": 13370 + }, + { + "epoch": 1.4683725016472655, + "grad_norm": 2.273282289505005, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7346392869949341, + "num_tokens": 340865364.0, + "step": 13371 + }, + { + "epoch": 1.4684823193498793, + "grad_norm": 2.3816795349121094, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.724589467048645, + "num_tokens": 340887423.0, + "step": 13372 + }, + { + "epoch": 1.4685921370524928, + "grad_norm": 2.0965464115142822, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7051090002059937, + "num_tokens": 340916250.0, + "step": 13373 + }, + { + "epoch": 1.4687019547551066, + "grad_norm": 2.489672899246216, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7160618305206299, + "num_tokens": 340937864.0, + "step": 13374 + }, + { + "epoch": 1.46881177245772, + "grad_norm": 2.2885966300964355, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7158917188644409, + "num_tokens": 340963283.0, + "step": 13375 + }, + { + "epoch": 1.4689215901603339, + "grad_norm": 2.3209428787231445, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7059353590011597, + "num_tokens": 340989680.0, + "step": 13376 + }, + { + "epoch": 1.4690314078629476, + "grad_norm": 2.036059856414795, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.715620756149292, + "num_tokens": 341020547.0, + "step": 13377 + }, + { + "epoch": 1.4691412255655611, + "grad_norm": 2.3234951496124268, + "learning_rate": 1e-06, + "loss": 0.768, + "mean_token_accuracy": 0.7502763271331787, + "num_tokens": 341042807.0, + "step": 13378 + }, + { + "epoch": 1.469251043268175, + "grad_norm": 2.1693005561828613, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.719164252281189, + "num_tokens": 341068603.0, + "step": 13379 + }, + { + "epoch": 1.4693608609707884, + "grad_norm": 2.227111339569092, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7360841035842896, + "num_tokens": 341094415.0, + "step": 13380 + }, + { + "epoch": 1.4694706786734022, + "grad_norm": 2.2928175926208496, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7175897359848022, + "num_tokens": 341118972.0, + "step": 13381 + }, + { + "epoch": 1.4695804963760157, + "grad_norm": 2.3604624271392822, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7278147339820862, + "num_tokens": 341143003.0, + "step": 13382 + }, + { + "epoch": 1.4696903140786295, + "grad_norm": 2.0581259727478027, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7177107930183411, + "num_tokens": 341174128.0, + "step": 13383 + }, + { + "epoch": 1.469800131781243, + "grad_norm": 1.9350718259811401, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7233942151069641, + "num_tokens": 341206578.0, + "step": 13384 + }, + { + "epoch": 1.4699099494838568, + "grad_norm": 2.127399206161499, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.717781662940979, + "num_tokens": 341232976.0, + "step": 13385 + }, + { + "epoch": 1.4700197671864705, + "grad_norm": 2.034393072128296, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6971547603607178, + "num_tokens": 341264722.0, + "step": 13386 + }, + { + "epoch": 1.470129584889084, + "grad_norm": 2.2842936515808105, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7170974612236023, + "num_tokens": 341291145.0, + "step": 13387 + }, + { + "epoch": 1.4702394025916978, + "grad_norm": 2.208322286605835, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7320393323898315, + "num_tokens": 341315512.0, + "step": 13388 + }, + { + "epoch": 1.4703492202943114, + "grad_norm": 2.0359110832214355, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7050347328186035, + "num_tokens": 341348573.0, + "step": 13389 + }, + { + "epoch": 1.4704590379969251, + "grad_norm": 2.370300769805908, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7249512672424316, + "num_tokens": 341373362.0, + "step": 13390 + }, + { + "epoch": 1.4705688556995389, + "grad_norm": 2.4104623794555664, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7104025483131409, + "num_tokens": 341398658.0, + "step": 13391 + }, + { + "epoch": 1.4706786734021524, + "grad_norm": 2.1604225635528564, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7205339074134827, + "num_tokens": 341427106.0, + "step": 13392 + }, + { + "epoch": 1.4707884911047662, + "grad_norm": 2.3184165954589844, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7124731540679932, + "num_tokens": 341450809.0, + "step": 13393 + }, + { + "epoch": 1.4708983088073797, + "grad_norm": 2.037524700164795, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6944409608840942, + "num_tokens": 341481846.0, + "step": 13394 + }, + { + "epoch": 1.4710081265099935, + "grad_norm": 2.7167060375213623, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.74056077003479, + "num_tokens": 341500835.0, + "step": 13395 + }, + { + "epoch": 1.471117944212607, + "grad_norm": 2.1608049869537354, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7135346531867981, + "num_tokens": 341529261.0, + "step": 13396 + }, + { + "epoch": 1.4712277619152208, + "grad_norm": 2.5439765453338623, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.719752311706543, + "num_tokens": 341550588.0, + "step": 13397 + }, + { + "epoch": 1.4713375796178343, + "grad_norm": 2.163875102996826, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7174330949783325, + "num_tokens": 341577201.0, + "step": 13398 + }, + { + "epoch": 1.471447397320448, + "grad_norm": 2.6403026580810547, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7216446399688721, + "num_tokens": 341595377.0, + "step": 13399 + }, + { + "epoch": 1.4715572150230618, + "grad_norm": 1.9856648445129395, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7130991816520691, + "num_tokens": 341626879.0, + "step": 13400 + }, + { + "epoch": 1.4716670327256753, + "grad_norm": 2.499145269393921, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7416564226150513, + "num_tokens": 341648140.0, + "step": 13401 + }, + { + "epoch": 1.471776850428289, + "grad_norm": 2.1577072143554688, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7244205474853516, + "num_tokens": 341674944.0, + "step": 13402 + }, + { + "epoch": 1.4718866681309026, + "grad_norm": 2.4318435192108154, + "learning_rate": 1e-06, + "loss": 0.7454, + "mean_token_accuracy": 0.7669327259063721, + "num_tokens": 341696097.0, + "step": 13403 + }, + { + "epoch": 1.4719964858335164, + "grad_norm": 2.411921501159668, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7076807022094727, + "num_tokens": 341720902.0, + "step": 13404 + }, + { + "epoch": 1.4721063035361301, + "grad_norm": 2.3226752281188965, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.711843729019165, + "num_tokens": 341745583.0, + "step": 13405 + }, + { + "epoch": 1.4722161212387437, + "grad_norm": 2.6180641651153564, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7163825035095215, + "num_tokens": 341766549.0, + "step": 13406 + }, + { + "epoch": 1.4723259389413572, + "grad_norm": 2.405853748321533, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7255938053131104, + "num_tokens": 341790297.0, + "step": 13407 + }, + { + "epoch": 1.472435756643971, + "grad_norm": 2.1942837238311768, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.6962248086929321, + "num_tokens": 341821117.0, + "step": 13408 + }, + { + "epoch": 1.4725455743465847, + "grad_norm": 2.058316469192505, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7315165996551514, + "num_tokens": 341848751.0, + "step": 13409 + }, + { + "epoch": 1.4726553920491983, + "grad_norm": 2.011620283126831, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7148020267486572, + "num_tokens": 341879401.0, + "step": 13410 + }, + { + "epoch": 1.472765209751812, + "grad_norm": 2.329775333404541, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7203188538551331, + "num_tokens": 341902353.0, + "step": 13411 + }, + { + "epoch": 1.4728750274544256, + "grad_norm": 1.840370535850525, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7128754258155823, + "num_tokens": 341935319.0, + "step": 13412 + }, + { + "epoch": 1.4729848451570393, + "grad_norm": 2.1236283779144287, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7276707291603088, + "num_tokens": 341963262.0, + "step": 13413 + }, + { + "epoch": 1.473094662859653, + "grad_norm": 2.032543659210205, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7192478179931641, + "num_tokens": 341991080.0, + "step": 13414 + }, + { + "epoch": 1.4732044805622666, + "grad_norm": 2.0749447345733643, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7128227353096008, + "num_tokens": 342021614.0, + "step": 13415 + }, + { + "epoch": 1.4733142982648804, + "grad_norm": 2.2400686740875244, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7456122636795044, + "num_tokens": 342046641.0, + "step": 13416 + }, + { + "epoch": 1.473424115967494, + "grad_norm": 2.3487448692321777, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7419881224632263, + "num_tokens": 342069416.0, + "step": 13417 + }, + { + "epoch": 1.4735339336701077, + "grad_norm": 2.1233556270599365, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.718259334564209, + "num_tokens": 342096301.0, + "step": 13418 + }, + { + "epoch": 1.4736437513727214, + "grad_norm": 1.969368577003479, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.712533712387085, + "num_tokens": 342127893.0, + "step": 13419 + }, + { + "epoch": 1.473753569075335, + "grad_norm": 2.1158976554870605, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7177767753601074, + "num_tokens": 342154211.0, + "step": 13420 + }, + { + "epoch": 1.4738633867779485, + "grad_norm": 2.3567070960998535, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7168415784835815, + "num_tokens": 342178079.0, + "step": 13421 + }, + { + "epoch": 1.4739732044805622, + "grad_norm": 2.0710930824279785, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7075149416923523, + "num_tokens": 342208236.0, + "step": 13422 + }, + { + "epoch": 1.474083022183176, + "grad_norm": 2.823117971420288, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7420082092285156, + "num_tokens": 342224510.0, + "step": 13423 + }, + { + "epoch": 1.4741928398857895, + "grad_norm": 2.580070734024048, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7226569056510925, + "num_tokens": 342246308.0, + "step": 13424 + }, + { + "epoch": 1.4743026575884033, + "grad_norm": 2.005776882171631, + "learning_rate": 1e-06, + "loss": 1.012, + "mean_token_accuracy": 0.6820963621139526, + "num_tokens": 342279201.0, + "step": 13425 + }, + { + "epoch": 1.4744124752910168, + "grad_norm": 2.2958619594573975, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7027363777160645, + "num_tokens": 342305685.0, + "step": 13426 + }, + { + "epoch": 1.4745222929936306, + "grad_norm": 2.47925066947937, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7139948010444641, + "num_tokens": 342328192.0, + "step": 13427 + }, + { + "epoch": 1.4746321106962443, + "grad_norm": 2.1286373138427734, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7168713808059692, + "num_tokens": 342356676.0, + "step": 13428 + }, + { + "epoch": 1.4747419283988579, + "grad_norm": 2.0626981258392334, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7212147116661072, + "num_tokens": 342384896.0, + "step": 13429 + }, + { + "epoch": 1.4748517461014716, + "grad_norm": 2.337036609649658, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7054733037948608, + "num_tokens": 342411485.0, + "step": 13430 + }, + { + "epoch": 1.4749615638040852, + "grad_norm": 2.301331043243408, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7222769260406494, + "num_tokens": 342436177.0, + "step": 13431 + }, + { + "epoch": 1.475071381506699, + "grad_norm": 2.5378332138061523, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.721371054649353, + "num_tokens": 342456550.0, + "step": 13432 + }, + { + "epoch": 1.4751811992093127, + "grad_norm": 2.276787042617798, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7128402590751648, + "num_tokens": 342481312.0, + "step": 13433 + }, + { + "epoch": 1.4752910169119262, + "grad_norm": 1.9862606525421143, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7018910050392151, + "num_tokens": 342512181.0, + "step": 13434 + }, + { + "epoch": 1.4754008346145397, + "grad_norm": 2.454042434692383, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.7468386888504028, + "num_tokens": 342533127.0, + "step": 13435 + }, + { + "epoch": 1.4755106523171535, + "grad_norm": 2.438972234725952, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7349104881286621, + "num_tokens": 342554866.0, + "step": 13436 + }, + { + "epoch": 1.4756204700197673, + "grad_norm": 2.2181460857391357, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7167091369628906, + "num_tokens": 342581303.0, + "step": 13437 + }, + { + "epoch": 1.4757302877223808, + "grad_norm": 2.3921005725860596, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7439623475074768, + "num_tokens": 342602667.0, + "step": 13438 + }, + { + "epoch": 1.4758401054249946, + "grad_norm": 2.1094629764556885, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7042670249938965, + "num_tokens": 342631131.0, + "step": 13439 + }, + { + "epoch": 1.475949923127608, + "grad_norm": 2.035895347595215, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.726082444190979, + "num_tokens": 342662829.0, + "step": 13440 + }, + { + "epoch": 1.4760597408302218, + "grad_norm": 2.453542709350586, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7239673733711243, + "num_tokens": 342684879.0, + "step": 13441 + }, + { + "epoch": 1.4761695585328356, + "grad_norm": 2.1147496700286865, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6906349658966064, + "num_tokens": 342715711.0, + "step": 13442 + }, + { + "epoch": 1.4762793762354491, + "grad_norm": 2.5120294094085693, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7288184762001038, + "num_tokens": 342738175.0, + "step": 13443 + }, + { + "epoch": 1.476389193938063, + "grad_norm": 2.0720558166503906, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7303942441940308, + "num_tokens": 342767410.0, + "step": 13444 + }, + { + "epoch": 1.4764990116406764, + "grad_norm": 2.4718313217163086, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7072561383247375, + "num_tokens": 342791663.0, + "step": 13445 + }, + { + "epoch": 1.4766088293432902, + "grad_norm": 2.324209451675415, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7177219390869141, + "num_tokens": 342818208.0, + "step": 13446 + }, + { + "epoch": 1.4767186470459037, + "grad_norm": 2.5273940563201904, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7315071225166321, + "num_tokens": 342838320.0, + "step": 13447 + }, + { + "epoch": 1.4768284647485175, + "grad_norm": 2.261580228805542, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7215708494186401, + "num_tokens": 342864032.0, + "step": 13448 + }, + { + "epoch": 1.476938282451131, + "grad_norm": 2.2043302059173584, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7272785902023315, + "num_tokens": 342890652.0, + "step": 13449 + }, + { + "epoch": 1.4770481001537448, + "grad_norm": 2.7676682472229004, + "learning_rate": 1e-06, + "loss": 0.7683, + "mean_token_accuracy": 0.7451848983764648, + "num_tokens": 342908064.0, + "step": 13450 + }, + { + "epoch": 1.4771579178563585, + "grad_norm": 2.395550012588501, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7169156074523926, + "num_tokens": 342932128.0, + "step": 13451 + }, + { + "epoch": 1.477267735558972, + "grad_norm": 2.481361150741577, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7198455929756165, + "num_tokens": 342953696.0, + "step": 13452 + }, + { + "epoch": 1.4773775532615858, + "grad_norm": 2.522265672683716, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7281454205513, + "num_tokens": 342974310.0, + "step": 13453 + }, + { + "epoch": 1.4774873709641994, + "grad_norm": 2.090841770172119, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7031148076057434, + "num_tokens": 343002876.0, + "step": 13454 + }, + { + "epoch": 1.477597188666813, + "grad_norm": 2.6410486698150635, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7265047430992126, + "num_tokens": 343022350.0, + "step": 13455 + }, + { + "epoch": 1.4777070063694269, + "grad_norm": 2.625396251678467, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7066936492919922, + "num_tokens": 343042669.0, + "step": 13456 + }, + { + "epoch": 1.4778168240720404, + "grad_norm": 1.9615259170532227, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7102679014205933, + "num_tokens": 343073312.0, + "step": 13457 + }, + { + "epoch": 1.477926641774654, + "grad_norm": 2.1676766872406006, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7285526990890503, + "num_tokens": 343099440.0, + "step": 13458 + }, + { + "epoch": 1.4780364594772677, + "grad_norm": 2.2280755043029785, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7216252684593201, + "num_tokens": 343123326.0, + "step": 13459 + }, + { + "epoch": 1.4781462771798815, + "grad_norm": 1.903928279876709, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7046375274658203, + "num_tokens": 343159435.0, + "step": 13460 + }, + { + "epoch": 1.478256094882495, + "grad_norm": 2.169646739959717, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.701130747795105, + "num_tokens": 343186201.0, + "step": 13461 + }, + { + "epoch": 1.4783659125851087, + "grad_norm": 2.2507517337799072, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7040920257568359, + "num_tokens": 343212558.0, + "step": 13462 + }, + { + "epoch": 1.4784757302877223, + "grad_norm": 2.10479474067688, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.717782735824585, + "num_tokens": 343240728.0, + "step": 13463 + }, + { + "epoch": 1.478585547990336, + "grad_norm": 2.2805557250976562, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.708195686340332, + "num_tokens": 343266472.0, + "step": 13464 + }, + { + "epoch": 1.4786953656929498, + "grad_norm": 2.4119925498962402, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.7351058721542358, + "num_tokens": 343289075.0, + "step": 13465 + }, + { + "epoch": 1.4788051833955633, + "grad_norm": 2.1623215675354004, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6932117938995361, + "num_tokens": 343317092.0, + "step": 13466 + }, + { + "epoch": 1.478915001098177, + "grad_norm": 2.1728389263153076, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7383050918579102, + "num_tokens": 343343484.0, + "step": 13467 + }, + { + "epoch": 1.4790248188007906, + "grad_norm": 2.3221189975738525, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7219668626785278, + "num_tokens": 343368473.0, + "step": 13468 + }, + { + "epoch": 1.4791346365034044, + "grad_norm": 2.232743501663208, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7061752080917358, + "num_tokens": 343395749.0, + "step": 13469 + }, + { + "epoch": 1.4792444542060181, + "grad_norm": 2.089508533477783, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7065021395683289, + "num_tokens": 343427973.0, + "step": 13470 + }, + { + "epoch": 1.4793542719086317, + "grad_norm": 2.1198549270629883, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7296761274337769, + "num_tokens": 343454771.0, + "step": 13471 + }, + { + "epoch": 1.4794640896112452, + "grad_norm": 2.4446778297424316, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7060248255729675, + "num_tokens": 343480798.0, + "step": 13472 + }, + { + "epoch": 1.479573907313859, + "grad_norm": 2.4408936500549316, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7363840341567993, + "num_tokens": 343504201.0, + "step": 13473 + }, + { + "epoch": 1.4796837250164727, + "grad_norm": 2.425992488861084, + "learning_rate": 1e-06, + "loss": 0.8115, + "mean_token_accuracy": 0.7422910928726196, + "num_tokens": 343524838.0, + "step": 13474 + }, + { + "epoch": 1.4797935427190863, + "grad_norm": 2.060788631439209, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7226783037185669, + "num_tokens": 343556747.0, + "step": 13475 + }, + { + "epoch": 1.4799033604217, + "grad_norm": 2.1487538814544678, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7329367399215698, + "num_tokens": 343582369.0, + "step": 13476 + }, + { + "epoch": 1.4800131781243135, + "grad_norm": 2.7495973110198975, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7088713645935059, + "num_tokens": 343602206.0, + "step": 13477 + }, + { + "epoch": 1.4801229958269273, + "grad_norm": 2.2063896656036377, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7211723327636719, + "num_tokens": 343631056.0, + "step": 13478 + }, + { + "epoch": 1.480232813529541, + "grad_norm": 2.1710140705108643, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7234695553779602, + "num_tokens": 343656396.0, + "step": 13479 + }, + { + "epoch": 1.4803426312321546, + "grad_norm": 2.5206542015075684, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7417151927947998, + "num_tokens": 343676439.0, + "step": 13480 + }, + { + "epoch": 1.4804524489347684, + "grad_norm": 2.591822385787964, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7281047105789185, + "num_tokens": 343697087.0, + "step": 13481 + }, + { + "epoch": 1.4805622666373819, + "grad_norm": 2.1403861045837402, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6932241916656494, + "num_tokens": 343727416.0, + "step": 13482 + }, + { + "epoch": 1.4806720843399956, + "grad_norm": 2.4904823303222656, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7266521453857422, + "num_tokens": 343749069.0, + "step": 13483 + }, + { + "epoch": 1.4807819020426094, + "grad_norm": 2.121248960494995, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7221893668174744, + "num_tokens": 343778920.0, + "step": 13484 + }, + { + "epoch": 1.480891719745223, + "grad_norm": 2.242295503616333, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7459679841995239, + "num_tokens": 343804061.0, + "step": 13485 + }, + { + "epoch": 1.4810015374478365, + "grad_norm": 2.4254636764526367, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7306939959526062, + "num_tokens": 343826314.0, + "step": 13486 + }, + { + "epoch": 1.4811113551504502, + "grad_norm": 2.2166876792907715, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7139829397201538, + "num_tokens": 343853004.0, + "step": 13487 + }, + { + "epoch": 1.481221172853064, + "grad_norm": 2.3629748821258545, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7305697202682495, + "num_tokens": 343876176.0, + "step": 13488 + }, + { + "epoch": 1.4813309905556775, + "grad_norm": 2.219372034072876, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7233965992927551, + "num_tokens": 343899699.0, + "step": 13489 + }, + { + "epoch": 1.4814408082582913, + "grad_norm": 2.220966100692749, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7187824845314026, + "num_tokens": 343923448.0, + "step": 13490 + }, + { + "epoch": 1.4815506259609048, + "grad_norm": 2.4109489917755127, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7274796366691589, + "num_tokens": 343946986.0, + "step": 13491 + }, + { + "epoch": 1.4816604436635186, + "grad_norm": 2.3597309589385986, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7226398587226868, + "num_tokens": 343970512.0, + "step": 13492 + }, + { + "epoch": 1.4817702613661323, + "grad_norm": 2.1876156330108643, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7139852046966553, + "num_tokens": 343997421.0, + "step": 13493 + }, + { + "epoch": 1.4818800790687459, + "grad_norm": 2.7729101181030273, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7319293022155762, + "num_tokens": 344016150.0, + "step": 13494 + }, + { + "epoch": 1.4819898967713596, + "grad_norm": 2.386521100997925, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.7591428756713867, + "num_tokens": 344035557.0, + "step": 13495 + }, + { + "epoch": 1.4820997144739732, + "grad_norm": 2.125105381011963, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.718191385269165, + "num_tokens": 344062508.0, + "step": 13496 + }, + { + "epoch": 1.482209532176587, + "grad_norm": 2.163994550704956, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.728744387626648, + "num_tokens": 344089428.0, + "step": 13497 + }, + { + "epoch": 1.4823193498792007, + "grad_norm": 2.339231252670288, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7391378879547119, + "num_tokens": 344112241.0, + "step": 13498 + }, + { + "epoch": 1.4824291675818142, + "grad_norm": 2.2887139320373535, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7174288034439087, + "num_tokens": 344138521.0, + "step": 13499 + }, + { + "epoch": 1.4825389852844277, + "grad_norm": 2.195746898651123, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7079994082450867, + "num_tokens": 344166155.0, + "step": 13500 + }, + { + "epoch": 1.4826488029870415, + "grad_norm": 2.328047513961792, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7208549976348877, + "num_tokens": 344190126.0, + "step": 13501 + }, + { + "epoch": 1.4827586206896552, + "grad_norm": 2.163295030593872, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7235295176506042, + "num_tokens": 344217851.0, + "step": 13502 + }, + { + "epoch": 1.4828684383922688, + "grad_norm": 2.387174367904663, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7244398593902588, + "num_tokens": 344241572.0, + "step": 13503 + }, + { + "epoch": 1.4829782560948825, + "grad_norm": 2.1347744464874268, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7078507542610168, + "num_tokens": 344270719.0, + "step": 13504 + }, + { + "epoch": 1.483088073797496, + "grad_norm": 2.0207767486572266, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7168567180633545, + "num_tokens": 344301521.0, + "step": 13505 + }, + { + "epoch": 1.4831978915001098, + "grad_norm": 2.050868034362793, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7097086906433105, + "num_tokens": 344330664.0, + "step": 13506 + }, + { + "epoch": 1.4833077092027236, + "grad_norm": 2.073716402053833, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.694992184638977, + "num_tokens": 344362237.0, + "step": 13507 + }, + { + "epoch": 1.4834175269053371, + "grad_norm": 2.27756929397583, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7088903784751892, + "num_tokens": 344386049.0, + "step": 13508 + }, + { + "epoch": 1.4835273446079509, + "grad_norm": 2.056999444961548, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7111522555351257, + "num_tokens": 344416564.0, + "step": 13509 + }, + { + "epoch": 1.4836371623105644, + "grad_norm": 2.291412591934204, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7359057664871216, + "num_tokens": 344442713.0, + "step": 13510 + }, + { + "epoch": 1.4837469800131782, + "grad_norm": 2.035818576812744, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7201181650161743, + "num_tokens": 344472752.0, + "step": 13511 + }, + { + "epoch": 1.4838567977157917, + "grad_norm": 2.8567144870758057, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7321056127548218, + "num_tokens": 344491277.0, + "step": 13512 + }, + { + "epoch": 1.4839666154184055, + "grad_norm": 2.452131509780884, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7314239740371704, + "num_tokens": 344513510.0, + "step": 13513 + }, + { + "epoch": 1.484076433121019, + "grad_norm": 2.2239749431610107, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.732309103012085, + "num_tokens": 344539924.0, + "step": 13514 + }, + { + "epoch": 1.4841862508236328, + "grad_norm": 2.241159439086914, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7176846265792847, + "num_tokens": 344565373.0, + "step": 13515 + }, + { + "epoch": 1.4842960685262465, + "grad_norm": 2.5094566345214844, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7165483832359314, + "num_tokens": 344588675.0, + "step": 13516 + }, + { + "epoch": 1.48440588622886, + "grad_norm": 2.5360238552093506, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7394533753395081, + "num_tokens": 344610028.0, + "step": 13517 + }, + { + "epoch": 1.4845157039314738, + "grad_norm": 2.2057433128356934, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7167466878890991, + "num_tokens": 344635379.0, + "step": 13518 + }, + { + "epoch": 1.4846255216340873, + "grad_norm": 2.2987279891967773, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7233659625053406, + "num_tokens": 344659607.0, + "step": 13519 + }, + { + "epoch": 1.484735339336701, + "grad_norm": 2.439765691757202, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7357780933380127, + "num_tokens": 344682354.0, + "step": 13520 + }, + { + "epoch": 1.4848451570393149, + "grad_norm": 2.499067783355713, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.730025053024292, + "num_tokens": 344704217.0, + "step": 13521 + }, + { + "epoch": 1.4849549747419284, + "grad_norm": 2.11259126663208, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7207848429679871, + "num_tokens": 344732069.0, + "step": 13522 + }, + { + "epoch": 1.485064792444542, + "grad_norm": 2.109614610671997, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.713470995426178, + "num_tokens": 344759980.0, + "step": 13523 + }, + { + "epoch": 1.4851746101471557, + "grad_norm": 2.0831634998321533, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7142205834388733, + "num_tokens": 344786661.0, + "step": 13524 + }, + { + "epoch": 1.4852844278497694, + "grad_norm": 2.2848544120788574, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7017086744308472, + "num_tokens": 344814328.0, + "step": 13525 + }, + { + "epoch": 1.485394245552383, + "grad_norm": 2.3763585090637207, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7101057171821594, + "num_tokens": 344837776.0, + "step": 13526 + }, + { + "epoch": 1.4855040632549967, + "grad_norm": 2.3219666481018066, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7315993309020996, + "num_tokens": 344860929.0, + "step": 13527 + }, + { + "epoch": 1.4856138809576103, + "grad_norm": 2.290469169616699, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7053290009498596, + "num_tokens": 344886867.0, + "step": 13528 + }, + { + "epoch": 1.485723698660224, + "grad_norm": 2.2881312370300293, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7116714715957642, + "num_tokens": 344912748.0, + "step": 13529 + }, + { + "epoch": 1.4858335163628378, + "grad_norm": 2.2802674770355225, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7067582607269287, + "num_tokens": 344937777.0, + "step": 13530 + }, + { + "epoch": 1.4859433340654513, + "grad_norm": 2.3967411518096924, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7165858745574951, + "num_tokens": 344961088.0, + "step": 13531 + }, + { + "epoch": 1.486053151768065, + "grad_norm": 2.4428045749664307, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7240405082702637, + "num_tokens": 344982429.0, + "step": 13532 + }, + { + "epoch": 1.4861629694706786, + "grad_norm": 2.4303014278411865, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7099853754043579, + "num_tokens": 345004862.0, + "step": 13533 + }, + { + "epoch": 1.4862727871732924, + "grad_norm": 2.1575725078582764, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7190840840339661, + "num_tokens": 345034573.0, + "step": 13534 + }, + { + "epoch": 1.4863826048759061, + "grad_norm": 2.358802318572998, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7376893758773804, + "num_tokens": 345059052.0, + "step": 13535 + }, + { + "epoch": 1.4864924225785197, + "grad_norm": 2.1991209983825684, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7049980163574219, + "num_tokens": 345087815.0, + "step": 13536 + }, + { + "epoch": 1.4866022402811332, + "grad_norm": 2.235071897506714, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7335535287857056, + "num_tokens": 345110637.0, + "step": 13537 + }, + { + "epoch": 1.486712057983747, + "grad_norm": 2.234401226043701, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.7591253519058228, + "num_tokens": 345136359.0, + "step": 13538 + }, + { + "epoch": 1.4868218756863607, + "grad_norm": 2.1289470195770264, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7228783369064331, + "num_tokens": 345165651.0, + "step": 13539 + }, + { + "epoch": 1.4869316933889742, + "grad_norm": 2.0652153491973877, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7039589881896973, + "num_tokens": 345197076.0, + "step": 13540 + }, + { + "epoch": 1.487041511091588, + "grad_norm": 2.157721757888794, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6933988332748413, + "num_tokens": 345226422.0, + "step": 13541 + }, + { + "epoch": 1.4871513287942015, + "grad_norm": 2.248575210571289, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7161040306091309, + "num_tokens": 345250418.0, + "step": 13542 + }, + { + "epoch": 1.4872611464968153, + "grad_norm": 1.9769455194473267, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7157526612281799, + "num_tokens": 345283164.0, + "step": 13543 + }, + { + "epoch": 1.487370964199429, + "grad_norm": 2.5237343311309814, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.7512786388397217, + "num_tokens": 345303646.0, + "step": 13544 + }, + { + "epoch": 1.4874807819020426, + "grad_norm": 2.3700690269470215, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7529603242874146, + "num_tokens": 345326280.0, + "step": 13545 + }, + { + "epoch": 1.4875905996046563, + "grad_norm": 2.076171875, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7108169198036194, + "num_tokens": 345355025.0, + "step": 13546 + }, + { + "epoch": 1.4877004173072699, + "grad_norm": 2.248500347137451, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7090516090393066, + "num_tokens": 345382754.0, + "step": 13547 + }, + { + "epoch": 1.4878102350098836, + "grad_norm": 2.4702227115631104, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7194657325744629, + "num_tokens": 345404228.0, + "step": 13548 + }, + { + "epoch": 1.4879200527124974, + "grad_norm": 2.3243558406829834, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7105754017829895, + "num_tokens": 345429639.0, + "step": 13549 + }, + { + "epoch": 1.488029870415111, + "grad_norm": 2.378722667694092, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7259317636489868, + "num_tokens": 345451884.0, + "step": 13550 + }, + { + "epoch": 1.4881396881177245, + "grad_norm": 2.18464732170105, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7206403017044067, + "num_tokens": 345477907.0, + "step": 13551 + }, + { + "epoch": 1.4882495058203382, + "grad_norm": 2.222113847732544, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7170441150665283, + "num_tokens": 345501989.0, + "step": 13552 + }, + { + "epoch": 1.488359323522952, + "grad_norm": 2.231980800628662, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7345253825187683, + "num_tokens": 345527234.0, + "step": 13553 + }, + { + "epoch": 1.4884691412255655, + "grad_norm": 2.4865667819976807, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7253788709640503, + "num_tokens": 345548831.0, + "step": 13554 + }, + { + "epoch": 1.4885789589281793, + "grad_norm": 2.372556447982788, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7217720746994019, + "num_tokens": 345574029.0, + "step": 13555 + }, + { + "epoch": 1.4886887766307928, + "grad_norm": 2.353175640106201, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.71484375, + "num_tokens": 345598335.0, + "step": 13556 + }, + { + "epoch": 1.4887985943334066, + "grad_norm": 2.0286803245544434, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7079336643218994, + "num_tokens": 345630107.0, + "step": 13557 + }, + { + "epoch": 1.4889084120360203, + "grad_norm": 2.2919163703918457, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7155231833457947, + "num_tokens": 345654048.0, + "step": 13558 + }, + { + "epoch": 1.4890182297386338, + "grad_norm": 2.344219446182251, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7207995653152466, + "num_tokens": 345677489.0, + "step": 13559 + }, + { + "epoch": 1.4891280474412476, + "grad_norm": 2.348745346069336, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7188172936439514, + "num_tokens": 345700865.0, + "step": 13560 + }, + { + "epoch": 1.4892378651438611, + "grad_norm": 2.2685225009918213, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6900191307067871, + "num_tokens": 345727806.0, + "step": 13561 + }, + { + "epoch": 1.489347682846475, + "grad_norm": 2.1822311878204346, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7080894708633423, + "num_tokens": 345754467.0, + "step": 13562 + }, + { + "epoch": 1.4894575005490884, + "grad_norm": 2.4699835777282715, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.7045601606369019, + "num_tokens": 345778692.0, + "step": 13563 + }, + { + "epoch": 1.4895673182517022, + "grad_norm": 2.511767864227295, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7214391231536865, + "num_tokens": 345801778.0, + "step": 13564 + }, + { + "epoch": 1.4896771359543157, + "grad_norm": 2.0880508422851562, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.6972398161888123, + "num_tokens": 345831172.0, + "step": 13565 + }, + { + "epoch": 1.4897869536569295, + "grad_norm": 2.495093584060669, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7262139320373535, + "num_tokens": 345852709.0, + "step": 13566 + }, + { + "epoch": 1.4898967713595432, + "grad_norm": 2.254202365875244, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.723290741443634, + "num_tokens": 345877481.0, + "step": 13567 + }, + { + "epoch": 1.4900065890621568, + "grad_norm": 2.0505549907684326, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.730839729309082, + "num_tokens": 345908895.0, + "step": 13568 + }, + { + "epoch": 1.4901164067647705, + "grad_norm": 2.0498430728912354, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7205552458763123, + "num_tokens": 345939221.0, + "step": 13569 + }, + { + "epoch": 1.490226224467384, + "grad_norm": 2.2711939811706543, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7224360704421997, + "num_tokens": 345965009.0, + "step": 13570 + }, + { + "epoch": 1.4903360421699978, + "grad_norm": 2.2099902629852295, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7173164486885071, + "num_tokens": 345991558.0, + "step": 13571 + }, + { + "epoch": 1.4904458598726116, + "grad_norm": 2.6314432621002197, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.7544032335281372, + "num_tokens": 346010450.0, + "step": 13572 + }, + { + "epoch": 1.4905556775752251, + "grad_norm": 1.9338304996490479, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7214323878288269, + "num_tokens": 346041762.0, + "step": 13573 + }, + { + "epoch": 1.4906654952778389, + "grad_norm": 2.3113670349121094, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7173427939414978, + "num_tokens": 346067568.0, + "step": 13574 + }, + { + "epoch": 1.4907753129804524, + "grad_norm": 2.6503469944000244, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7105107307434082, + "num_tokens": 346087122.0, + "step": 13575 + }, + { + "epoch": 1.4908851306830662, + "grad_norm": 2.227729082107544, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7131249904632568, + "num_tokens": 346111822.0, + "step": 13576 + }, + { + "epoch": 1.4909949483856797, + "grad_norm": 2.1764779090881348, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.705669641494751, + "num_tokens": 346139938.0, + "step": 13577 + }, + { + "epoch": 1.4911047660882935, + "grad_norm": 2.047715663909912, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7083749771118164, + "num_tokens": 346172443.0, + "step": 13578 + }, + { + "epoch": 1.491214583790907, + "grad_norm": 2.0337014198303223, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7208369374275208, + "num_tokens": 346202229.0, + "step": 13579 + }, + { + "epoch": 1.4913244014935207, + "grad_norm": 2.338174819946289, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.6987093687057495, + "num_tokens": 346225821.0, + "step": 13580 + }, + { + "epoch": 1.4914342191961345, + "grad_norm": 2.5925915241241455, + "learning_rate": 1e-06, + "loss": 0.7742, + "mean_token_accuracy": 0.7496852874755859, + "num_tokens": 346244782.0, + "step": 13581 + }, + { + "epoch": 1.491544036898748, + "grad_norm": 2.459975004196167, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7371947765350342, + "num_tokens": 346266451.0, + "step": 13582 + }, + { + "epoch": 1.4916538546013618, + "grad_norm": 2.2899177074432373, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7212939262390137, + "num_tokens": 346290806.0, + "step": 13583 + }, + { + "epoch": 1.4917636723039753, + "grad_norm": 2.4374828338623047, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7360247373580933, + "num_tokens": 346313381.0, + "step": 13584 + }, + { + "epoch": 1.491873490006589, + "grad_norm": 2.250507354736328, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6931842565536499, + "num_tokens": 346339799.0, + "step": 13585 + }, + { + "epoch": 1.4919833077092028, + "grad_norm": 2.1801087856292725, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7410112023353577, + "num_tokens": 346366179.0, + "step": 13586 + }, + { + "epoch": 1.4920931254118164, + "grad_norm": 2.427436113357544, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7229222059249878, + "num_tokens": 346390157.0, + "step": 13587 + }, + { + "epoch": 1.49220294311443, + "grad_norm": 2.4004125595092773, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7223149538040161, + "num_tokens": 346414237.0, + "step": 13588 + }, + { + "epoch": 1.4923127608170437, + "grad_norm": 2.223526954650879, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7097949981689453, + "num_tokens": 346441024.0, + "step": 13589 + }, + { + "epoch": 1.4924225785196574, + "grad_norm": 2.275723695755005, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7266943454742432, + "num_tokens": 346466403.0, + "step": 13590 + }, + { + "epoch": 1.492532396222271, + "grad_norm": 2.0440337657928467, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7233538627624512, + "num_tokens": 346495040.0, + "step": 13591 + }, + { + "epoch": 1.4926422139248847, + "grad_norm": 2.4153494834899902, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.742048978805542, + "num_tokens": 346517584.0, + "step": 13592 + }, + { + "epoch": 1.4927520316274983, + "grad_norm": 2.5484347343444824, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7304076552391052, + "num_tokens": 346538734.0, + "step": 13593 + }, + { + "epoch": 1.492861849330112, + "grad_norm": 2.081078290939331, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7534196972846985, + "num_tokens": 346564806.0, + "step": 13594 + }, + { + "epoch": 1.4929716670327258, + "grad_norm": 2.34482741355896, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7248786687850952, + "num_tokens": 346590241.0, + "step": 13595 + }, + { + "epoch": 1.4930814847353393, + "grad_norm": 2.2377824783325195, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7146306037902832, + "num_tokens": 346615459.0, + "step": 13596 + }, + { + "epoch": 1.493191302437953, + "grad_norm": 2.32900071144104, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7393605709075928, + "num_tokens": 346639099.0, + "step": 13597 + }, + { + "epoch": 1.4933011201405666, + "grad_norm": 2.262608766555786, + "learning_rate": 1e-06, + "loss": 0.8128, + "mean_token_accuracy": 0.7424502968788147, + "num_tokens": 346663010.0, + "step": 13598 + }, + { + "epoch": 1.4934109378431804, + "grad_norm": 2.2060160636901855, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7047653794288635, + "num_tokens": 346691695.0, + "step": 13599 + }, + { + "epoch": 1.493520755545794, + "grad_norm": 2.13088321685791, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6949409246444702, + "num_tokens": 346720284.0, + "step": 13600 + }, + { + "epoch": 1.4936305732484076, + "grad_norm": 2.6058037281036377, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7264825105667114, + "num_tokens": 346740293.0, + "step": 13601 + }, + { + "epoch": 1.4937403909510212, + "grad_norm": 2.0928940773010254, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7379159927368164, + "num_tokens": 346767793.0, + "step": 13602 + }, + { + "epoch": 1.493850208653635, + "grad_norm": 2.079972505569458, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7095582485198975, + "num_tokens": 346798550.0, + "step": 13603 + }, + { + "epoch": 1.4939600263562487, + "grad_norm": 2.4649734497070312, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7314355373382568, + "num_tokens": 346818519.0, + "step": 13604 + }, + { + "epoch": 1.4940698440588622, + "grad_norm": 2.110264539718628, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7203032970428467, + "num_tokens": 346846432.0, + "step": 13605 + }, + { + "epoch": 1.494179661761476, + "grad_norm": 2.219101905822754, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7559158802032471, + "num_tokens": 346870951.0, + "step": 13606 + }, + { + "epoch": 1.4942894794640895, + "grad_norm": 2.2741477489471436, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7315563559532166, + "num_tokens": 346895225.0, + "step": 13607 + }, + { + "epoch": 1.4943992971667033, + "grad_norm": 2.2376904487609863, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7204082012176514, + "num_tokens": 346922444.0, + "step": 13608 + }, + { + "epoch": 1.494509114869317, + "grad_norm": 2.2536792755126953, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7251958250999451, + "num_tokens": 346949917.0, + "step": 13609 + }, + { + "epoch": 1.4946189325719306, + "grad_norm": 2.3006067276000977, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7111486196517944, + "num_tokens": 346974606.0, + "step": 13610 + }, + { + "epoch": 1.4947287502745443, + "grad_norm": 2.4820728302001953, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7137093544006348, + "num_tokens": 346995102.0, + "step": 13611 + }, + { + "epoch": 1.4948385679771579, + "grad_norm": 2.071260690689087, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6920809745788574, + "num_tokens": 347024628.0, + "step": 13612 + }, + { + "epoch": 1.4949483856797716, + "grad_norm": 2.131929874420166, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7041969299316406, + "num_tokens": 347054584.0, + "step": 13613 + }, + { + "epoch": 1.4950582033823854, + "grad_norm": 2.2673771381378174, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7187173366546631, + "num_tokens": 347079474.0, + "step": 13614 + }, + { + "epoch": 1.495168021084999, + "grad_norm": 2.1655495166778564, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6981822848320007, + "num_tokens": 347107593.0, + "step": 13615 + }, + { + "epoch": 1.4952778387876124, + "grad_norm": 2.4580986499786377, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7272984981536865, + "num_tokens": 347130025.0, + "step": 13616 + }, + { + "epoch": 1.4953876564902262, + "grad_norm": 2.216233015060425, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.6971112489700317, + "num_tokens": 347157213.0, + "step": 13617 + }, + { + "epoch": 1.49549747419284, + "grad_norm": 2.2393815517425537, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7315758466720581, + "num_tokens": 347183260.0, + "step": 13618 + }, + { + "epoch": 1.4956072918954535, + "grad_norm": 2.075495481491089, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.704024612903595, + "num_tokens": 347211801.0, + "step": 13619 + }, + { + "epoch": 1.4957171095980673, + "grad_norm": 2.278864622116089, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7024296522140503, + "num_tokens": 347238634.0, + "step": 13620 + }, + { + "epoch": 1.4958269273006808, + "grad_norm": 2.3526546955108643, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7201551795005798, + "num_tokens": 347264136.0, + "step": 13621 + }, + { + "epoch": 1.4959367450032945, + "grad_norm": 2.3281805515289307, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7352351546287537, + "num_tokens": 347288455.0, + "step": 13622 + }, + { + "epoch": 1.4960465627059083, + "grad_norm": 2.1254892349243164, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7250454425811768, + "num_tokens": 347316228.0, + "step": 13623 + }, + { + "epoch": 1.4961563804085218, + "grad_norm": 2.3153252601623535, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7432044744491577, + "num_tokens": 347339995.0, + "step": 13624 + }, + { + "epoch": 1.4962661981111356, + "grad_norm": 2.0384888648986816, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6852719783782959, + "num_tokens": 347372498.0, + "step": 13625 + }, + { + "epoch": 1.4963760158137491, + "grad_norm": 2.061272144317627, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7134060263633728, + "num_tokens": 347402132.0, + "step": 13626 + }, + { + "epoch": 1.4964858335163629, + "grad_norm": 2.0974040031433105, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7113770842552185, + "num_tokens": 347432791.0, + "step": 13627 + }, + { + "epoch": 1.4965956512189764, + "grad_norm": 2.458415985107422, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7258782386779785, + "num_tokens": 347455147.0, + "step": 13628 + }, + { + "epoch": 1.4967054689215902, + "grad_norm": 2.563896894454956, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7195821404457092, + "num_tokens": 347476546.0, + "step": 13629 + }, + { + "epoch": 1.4968152866242037, + "grad_norm": 2.805861473083496, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7266279458999634, + "num_tokens": 347494135.0, + "step": 13630 + }, + { + "epoch": 1.4969251043268175, + "grad_norm": 2.139765739440918, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6836472153663635, + "num_tokens": 347523045.0, + "step": 13631 + }, + { + "epoch": 1.4970349220294312, + "grad_norm": 2.371328592300415, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7564941644668579, + "num_tokens": 347544264.0, + "step": 13632 + }, + { + "epoch": 1.4971447397320448, + "grad_norm": 2.564467191696167, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7203965187072754, + "num_tokens": 347566161.0, + "step": 13633 + }, + { + "epoch": 1.4972545574346585, + "grad_norm": 2.635406255722046, + "learning_rate": 1e-06, + "loss": 0.7936, + "mean_token_accuracy": 0.7404470443725586, + "num_tokens": 347585351.0, + "step": 13634 + }, + { + "epoch": 1.497364375137272, + "grad_norm": 2.0602991580963135, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7198508977890015, + "num_tokens": 347616264.0, + "step": 13635 + }, + { + "epoch": 1.4974741928398858, + "grad_norm": 2.347837448120117, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.735408365726471, + "num_tokens": 347639075.0, + "step": 13636 + }, + { + "epoch": 1.4975840105424996, + "grad_norm": 2.4500653743743896, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7268894910812378, + "num_tokens": 347661740.0, + "step": 13637 + }, + { + "epoch": 1.497693828245113, + "grad_norm": 2.1196725368499756, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7179584503173828, + "num_tokens": 347691657.0, + "step": 13638 + }, + { + "epoch": 1.4978036459477266, + "grad_norm": 2.1058149337768555, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.6982246041297913, + "num_tokens": 347720711.0, + "step": 13639 + }, + { + "epoch": 1.4979134636503404, + "grad_norm": 2.2130181789398193, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7350795865058899, + "num_tokens": 347748405.0, + "step": 13640 + }, + { + "epoch": 1.4980232813529542, + "grad_norm": 1.9569422006607056, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.712185263633728, + "num_tokens": 347782203.0, + "step": 13641 + }, + { + "epoch": 1.4981330990555677, + "grad_norm": 2.733621835708618, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7142543792724609, + "num_tokens": 347801296.0, + "step": 13642 + }, + { + "epoch": 1.4982429167581814, + "grad_norm": 2.690390110015869, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7371126413345337, + "num_tokens": 347820387.0, + "step": 13643 + }, + { + "epoch": 1.498352734460795, + "grad_norm": 2.3124117851257324, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7188470959663391, + "num_tokens": 347842887.0, + "step": 13644 + }, + { + "epoch": 1.4984625521634087, + "grad_norm": 2.1083486080169678, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7307182550430298, + "num_tokens": 347869873.0, + "step": 13645 + }, + { + "epoch": 1.4985723698660225, + "grad_norm": 2.4798007011413574, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7374609112739563, + "num_tokens": 347889214.0, + "step": 13646 + }, + { + "epoch": 1.498682187568636, + "grad_norm": 2.1983494758605957, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.734570324420929, + "num_tokens": 347913903.0, + "step": 13647 + }, + { + "epoch": 1.4987920052712498, + "grad_norm": 2.249870777130127, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7158658504486084, + "num_tokens": 347941105.0, + "step": 13648 + }, + { + "epoch": 1.4989018229738633, + "grad_norm": 2.744706630706787, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7183424830436707, + "num_tokens": 347958928.0, + "step": 13649 + }, + { + "epoch": 1.499011640676477, + "grad_norm": 2.5055487155914307, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7263981103897095, + "num_tokens": 347979636.0, + "step": 13650 + }, + { + "epoch": 1.4991214583790908, + "grad_norm": 2.1503922939300537, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7230157256126404, + "num_tokens": 348006655.0, + "step": 13651 + }, + { + "epoch": 1.4992312760817044, + "grad_norm": 2.2548046112060547, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7071240544319153, + "num_tokens": 348034366.0, + "step": 13652 + }, + { + "epoch": 1.499341093784318, + "grad_norm": 2.552826166152954, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7198928594589233, + "num_tokens": 348057777.0, + "step": 13653 + }, + { + "epoch": 1.4994509114869317, + "grad_norm": 2.027099847793579, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7127411365509033, + "num_tokens": 348085575.0, + "step": 13654 + }, + { + "epoch": 1.4995607291895454, + "grad_norm": 2.4641900062561035, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.729399561882019, + "num_tokens": 348107297.0, + "step": 13655 + }, + { + "epoch": 1.499670546892159, + "grad_norm": 2.2923009395599365, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7341009378433228, + "num_tokens": 348131401.0, + "step": 13656 + }, + { + "epoch": 1.4997803645947727, + "grad_norm": 2.2103755474090576, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7178643941879272, + "num_tokens": 348156767.0, + "step": 13657 + }, + { + "epoch": 1.4998901822973862, + "grad_norm": 2.309095621109009, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7197800874710083, + "num_tokens": 348181088.0, + "step": 13658 + }, + { + "epoch": 1.5, + "grad_norm": 2.441951036453247, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7166493535041809, + "num_tokens": 348206151.0, + "step": 13659 + }, + { + "epoch": 1.5001098177026138, + "grad_norm": 2.41839861869812, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7073756456375122, + "num_tokens": 348229726.0, + "step": 13660 + }, + { + "epoch": 1.5002196354052273, + "grad_norm": 2.507645845413208, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7293881177902222, + "num_tokens": 348251088.0, + "step": 13661 + }, + { + "epoch": 1.5003294531078408, + "grad_norm": 1.7909327745437622, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.725601851940155, + "num_tokens": 348287340.0, + "step": 13662 + }, + { + "epoch": 1.5004392708104546, + "grad_norm": 2.238729476928711, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7259609699249268, + "num_tokens": 348311839.0, + "step": 13663 + }, + { + "epoch": 1.5005490885130683, + "grad_norm": 2.0625603199005127, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7158179879188538, + "num_tokens": 348340114.0, + "step": 13664 + }, + { + "epoch": 1.500658906215682, + "grad_norm": 2.053021192550659, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7171770334243774, + "num_tokens": 348369674.0, + "step": 13665 + }, + { + "epoch": 1.5007687239182956, + "grad_norm": 1.9888982772827148, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7109752297401428, + "num_tokens": 348401382.0, + "step": 13666 + }, + { + "epoch": 1.5008785416209092, + "grad_norm": 2.2773823738098145, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.6968619227409363, + "num_tokens": 348428767.0, + "step": 13667 + }, + { + "epoch": 1.500988359323523, + "grad_norm": 1.9807130098342896, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7104406952857971, + "num_tokens": 348461521.0, + "step": 13668 + }, + { + "epoch": 1.5010981770261367, + "grad_norm": 2.5062172412872314, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7257205247879028, + "num_tokens": 348481853.0, + "step": 13669 + }, + { + "epoch": 1.5012079947287504, + "grad_norm": 2.2938125133514404, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7485736608505249, + "num_tokens": 348506377.0, + "step": 13670 + }, + { + "epoch": 1.501317812431364, + "grad_norm": 2.388544797897339, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7276488542556763, + "num_tokens": 348528319.0, + "step": 13671 + }, + { + "epoch": 1.5014276301339775, + "grad_norm": 2.2831854820251465, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7274730205535889, + "num_tokens": 348554634.0, + "step": 13672 + }, + { + "epoch": 1.5015374478365913, + "grad_norm": 2.460482597351074, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7065659761428833, + "num_tokens": 348578324.0, + "step": 13673 + }, + { + "epoch": 1.501647265539205, + "grad_norm": 2.1403489112854004, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.720977783203125, + "num_tokens": 348606217.0, + "step": 13674 + }, + { + "epoch": 1.5017570832418186, + "grad_norm": 2.837196111679077, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.737646222114563, + "num_tokens": 348623267.0, + "step": 13675 + }, + { + "epoch": 1.501866900944432, + "grad_norm": 2.1853482723236084, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7229788303375244, + "num_tokens": 348648727.0, + "step": 13676 + }, + { + "epoch": 1.5019767186470458, + "grad_norm": 2.4783267974853516, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7170054912567139, + "num_tokens": 348670422.0, + "step": 13677 + }, + { + "epoch": 1.5020865363496596, + "grad_norm": 2.623769998550415, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7376857995986938, + "num_tokens": 348689615.0, + "step": 13678 + }, + { + "epoch": 1.5021963540522734, + "grad_norm": 2.4119861125946045, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7195113301277161, + "num_tokens": 348714131.0, + "step": 13679 + }, + { + "epoch": 1.502306171754887, + "grad_norm": 2.1591362953186035, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6931695342063904, + "num_tokens": 348743589.0, + "step": 13680 + }, + { + "epoch": 1.5024159894575004, + "grad_norm": 2.7595999240875244, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7239578366279602, + "num_tokens": 348761490.0, + "step": 13681 + }, + { + "epoch": 1.5025258071601142, + "grad_norm": 2.005481481552124, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7358909249305725, + "num_tokens": 348791929.0, + "step": 13682 + }, + { + "epoch": 1.502635624862728, + "grad_norm": 2.7904915809631348, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7564035654067993, + "num_tokens": 348808664.0, + "step": 13683 + }, + { + "epoch": 1.5027454425653415, + "grad_norm": 2.2932779788970947, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7211209535598755, + "num_tokens": 348833095.0, + "step": 13684 + }, + { + "epoch": 1.5028552602679552, + "grad_norm": 2.620454788208008, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7147201299667358, + "num_tokens": 348853702.0, + "step": 13685 + }, + { + "epoch": 1.5029650779705688, + "grad_norm": 2.1249215602874756, + "learning_rate": 1e-06, + "loss": 0.8264, + "mean_token_accuracy": 0.7437189221382141, + "num_tokens": 348881070.0, + "step": 13686 + }, + { + "epoch": 1.5030748956731825, + "grad_norm": 2.3659582138061523, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7183928489685059, + "num_tokens": 348904847.0, + "step": 13687 + }, + { + "epoch": 1.5031847133757963, + "grad_norm": 2.710111379623413, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7166028022766113, + "num_tokens": 348924820.0, + "step": 13688 + }, + { + "epoch": 1.5032945310784098, + "grad_norm": 2.521765947341919, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7337992191314697, + "num_tokens": 348946851.0, + "step": 13689 + }, + { + "epoch": 1.5034043487810234, + "grad_norm": 2.4365687370300293, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7321563363075256, + "num_tokens": 348967932.0, + "step": 13690 + }, + { + "epoch": 1.5035141664836371, + "grad_norm": 2.2858827114105225, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7290404438972473, + "num_tokens": 348991618.0, + "step": 13691 + }, + { + "epoch": 1.5036239841862509, + "grad_norm": 2.006906509399414, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.720714807510376, + "num_tokens": 349022505.0, + "step": 13692 + }, + { + "epoch": 1.5037338018888646, + "grad_norm": 2.0538322925567627, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7089378833770752, + "num_tokens": 349053498.0, + "step": 13693 + }, + { + "epoch": 1.5038436195914782, + "grad_norm": 2.3094921112060547, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7166929244995117, + "num_tokens": 349077650.0, + "step": 13694 + }, + { + "epoch": 1.5039534372940917, + "grad_norm": 2.2934224605560303, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7305177450180054, + "num_tokens": 349102068.0, + "step": 13695 + }, + { + "epoch": 1.5040632549967055, + "grad_norm": 2.09169340133667, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7097122073173523, + "num_tokens": 349130873.0, + "step": 13696 + }, + { + "epoch": 1.5041730726993192, + "grad_norm": 2.3189165592193604, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7175443172454834, + "num_tokens": 349157697.0, + "step": 13697 + }, + { + "epoch": 1.5042828904019327, + "grad_norm": 2.1375577449798584, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7090351581573486, + "num_tokens": 349187784.0, + "step": 13698 + }, + { + "epoch": 1.5043927081045465, + "grad_norm": 2.1803030967712402, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7297702431678772, + "num_tokens": 349214565.0, + "step": 13699 + }, + { + "epoch": 1.50450252580716, + "grad_norm": 2.180126428604126, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7122546434402466, + "num_tokens": 349241178.0, + "step": 13700 + }, + { + "epoch": 1.5046123435097738, + "grad_norm": 1.9637243747711182, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7030807733535767, + "num_tokens": 349273146.0, + "step": 13701 + }, + { + "epoch": 1.5047221612123876, + "grad_norm": 2.3576977252960205, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7210254669189453, + "num_tokens": 349297586.0, + "step": 13702 + }, + { + "epoch": 1.504831978915001, + "grad_norm": 2.0995185375213623, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7281597852706909, + "num_tokens": 349325219.0, + "step": 13703 + }, + { + "epoch": 1.5049417966176146, + "grad_norm": 2.4409940242767334, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7442814111709595, + "num_tokens": 349345825.0, + "step": 13704 + }, + { + "epoch": 1.5050516143202284, + "grad_norm": 2.5883405208587646, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7294430732727051, + "num_tokens": 349365341.0, + "step": 13705 + }, + { + "epoch": 1.5051614320228421, + "grad_norm": 2.5158591270446777, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7406576871871948, + "num_tokens": 349387326.0, + "step": 13706 + }, + { + "epoch": 1.505271249725456, + "grad_norm": 2.472294330596924, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7223643064498901, + "num_tokens": 349409689.0, + "step": 13707 + }, + { + "epoch": 1.5053810674280694, + "grad_norm": 2.4182381629943848, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7450580596923828, + "num_tokens": 349431286.0, + "step": 13708 + }, + { + "epoch": 1.505490885130683, + "grad_norm": 2.4498443603515625, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7206358313560486, + "num_tokens": 349454152.0, + "step": 13709 + }, + { + "epoch": 1.5056007028332967, + "grad_norm": 2.615067481994629, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7230014204978943, + "num_tokens": 349474480.0, + "step": 13710 + }, + { + "epoch": 1.5057105205359105, + "grad_norm": 2.471277952194214, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7337929606437683, + "num_tokens": 349495110.0, + "step": 13711 + }, + { + "epoch": 1.505820338238524, + "grad_norm": 2.0981392860412598, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7345432639122009, + "num_tokens": 349522160.0, + "step": 13712 + }, + { + "epoch": 1.5059301559411375, + "grad_norm": 2.3607194423675537, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7313625812530518, + "num_tokens": 349546122.0, + "step": 13713 + }, + { + "epoch": 1.5060399736437513, + "grad_norm": 2.2461094856262207, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7078844308853149, + "num_tokens": 349572519.0, + "step": 13714 + }, + { + "epoch": 1.506149791346365, + "grad_norm": 2.1244120597839355, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7280617952346802, + "num_tokens": 349598447.0, + "step": 13715 + }, + { + "epoch": 1.5062596090489788, + "grad_norm": 2.390899658203125, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7395954728126526, + "num_tokens": 349621482.0, + "step": 13716 + }, + { + "epoch": 1.5063694267515924, + "grad_norm": 2.56152606010437, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.719901442527771, + "num_tokens": 349642904.0, + "step": 13717 + }, + { + "epoch": 1.506479244454206, + "grad_norm": 1.9933820962905884, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7375384569168091, + "num_tokens": 349671817.0, + "step": 13718 + }, + { + "epoch": 1.5065890621568196, + "grad_norm": 2.1313698291778564, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.713070273399353, + "num_tokens": 349698019.0, + "step": 13719 + }, + { + "epoch": 1.5066988798594334, + "grad_norm": 2.237847089767456, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7148770093917847, + "num_tokens": 349723164.0, + "step": 13720 + }, + { + "epoch": 1.5068086975620472, + "grad_norm": 2.2266087532043457, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7190830707550049, + "num_tokens": 349748666.0, + "step": 13721 + }, + { + "epoch": 1.5069185152646607, + "grad_norm": 1.9335227012634277, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7073248624801636, + "num_tokens": 349780634.0, + "step": 13722 + }, + { + "epoch": 1.5070283329672742, + "grad_norm": 2.804776668548584, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7238984107971191, + "num_tokens": 349799339.0, + "step": 13723 + }, + { + "epoch": 1.507138150669888, + "grad_norm": 2.276296377182007, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7145060300827026, + "num_tokens": 349825701.0, + "step": 13724 + }, + { + "epoch": 1.5072479683725017, + "grad_norm": 2.0923702716827393, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.714018702507019, + "num_tokens": 349853057.0, + "step": 13725 + }, + { + "epoch": 1.5073577860751153, + "grad_norm": 2.2258589267730713, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7238155603408813, + "num_tokens": 349880542.0, + "step": 13726 + }, + { + "epoch": 1.5074676037777288, + "grad_norm": 2.281604051589966, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7130682468414307, + "num_tokens": 349907251.0, + "step": 13727 + }, + { + "epoch": 1.5075774214803426, + "grad_norm": 2.0810134410858154, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7134883999824524, + "num_tokens": 349936302.0, + "step": 13728 + }, + { + "epoch": 1.5076872391829563, + "grad_norm": 2.1029651165008545, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7112941741943359, + "num_tokens": 349963655.0, + "step": 13729 + }, + { + "epoch": 1.50779705688557, + "grad_norm": 2.1568117141723633, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7219459414482117, + "num_tokens": 349991577.0, + "step": 13730 + }, + { + "epoch": 1.5079068745881836, + "grad_norm": 1.97003972530365, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.748752236366272, + "num_tokens": 350023371.0, + "step": 13731 + }, + { + "epoch": 1.5080166922907972, + "grad_norm": 2.0530025959014893, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7127492427825928, + "num_tokens": 350053122.0, + "step": 13732 + }, + { + "epoch": 1.508126509993411, + "grad_norm": 2.5925564765930176, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7277711629867554, + "num_tokens": 350073814.0, + "step": 13733 + }, + { + "epoch": 1.5082363276960247, + "grad_norm": 2.1281850337982178, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7089085578918457, + "num_tokens": 350101466.0, + "step": 13734 + }, + { + "epoch": 1.5083461453986382, + "grad_norm": 2.2539095878601074, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.716381311416626, + "num_tokens": 350125673.0, + "step": 13735 + }, + { + "epoch": 1.508455963101252, + "grad_norm": 2.2669312953948975, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7111438512802124, + "num_tokens": 350153287.0, + "step": 13736 + }, + { + "epoch": 1.5085657808038655, + "grad_norm": 2.3092026710510254, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6971802711486816, + "num_tokens": 350180507.0, + "step": 13737 + }, + { + "epoch": 1.5086755985064793, + "grad_norm": 2.218949556350708, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7291958928108215, + "num_tokens": 350205699.0, + "step": 13738 + }, + { + "epoch": 1.508785416209093, + "grad_norm": 2.1263201236724854, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7170078754425049, + "num_tokens": 350235619.0, + "step": 13739 + }, + { + "epoch": 1.5088952339117065, + "grad_norm": 1.8865675926208496, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.695509135723114, + "num_tokens": 350273601.0, + "step": 13740 + }, + { + "epoch": 1.50900505161432, + "grad_norm": 2.4787590503692627, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7281285524368286, + "num_tokens": 350295834.0, + "step": 13741 + }, + { + "epoch": 1.5091148693169338, + "grad_norm": 2.3495075702667236, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7104899883270264, + "num_tokens": 350320450.0, + "step": 13742 + }, + { + "epoch": 1.5092246870195476, + "grad_norm": 2.114408016204834, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7006660103797913, + "num_tokens": 350349972.0, + "step": 13743 + }, + { + "epoch": 1.5093345047221614, + "grad_norm": 2.3413593769073486, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7017488479614258, + "num_tokens": 350375360.0, + "step": 13744 + }, + { + "epoch": 1.5094443224247749, + "grad_norm": 2.260145664215088, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7165685892105103, + "num_tokens": 350401166.0, + "step": 13745 + }, + { + "epoch": 1.5095541401273884, + "grad_norm": 2.237708330154419, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7235572338104248, + "num_tokens": 350426358.0, + "step": 13746 + }, + { + "epoch": 1.5096639578300022, + "grad_norm": 2.1340067386627197, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7280763983726501, + "num_tokens": 350452905.0, + "step": 13747 + }, + { + "epoch": 1.509773775532616, + "grad_norm": 2.1003990173339844, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7208086252212524, + "num_tokens": 350482811.0, + "step": 13748 + }, + { + "epoch": 1.5098835932352295, + "grad_norm": 2.1928622722625732, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7202162742614746, + "num_tokens": 350508827.0, + "step": 13749 + }, + { + "epoch": 1.5099934109378432, + "grad_norm": 2.2878570556640625, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7375681400299072, + "num_tokens": 350532928.0, + "step": 13750 + }, + { + "epoch": 1.5101032286404568, + "grad_norm": 2.1999258995056152, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7175829410552979, + "num_tokens": 350560727.0, + "step": 13751 + }, + { + "epoch": 1.5102130463430705, + "grad_norm": 2.0285840034484863, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7227899432182312, + "num_tokens": 350592057.0, + "step": 13752 + }, + { + "epoch": 1.5103228640456843, + "grad_norm": 2.29293155670166, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.7456709146499634, + "num_tokens": 350615386.0, + "step": 13753 + }, + { + "epoch": 1.5104326817482978, + "grad_norm": 2.0955140590667725, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.704947292804718, + "num_tokens": 350646058.0, + "step": 13754 + }, + { + "epoch": 1.5105424994509113, + "grad_norm": 2.0845022201538086, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7209866642951965, + "num_tokens": 350673473.0, + "step": 13755 + }, + { + "epoch": 1.510652317153525, + "grad_norm": 2.554216146469116, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7373417615890503, + "num_tokens": 350694869.0, + "step": 13756 + }, + { + "epoch": 1.5107621348561389, + "grad_norm": 2.2419376373291016, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7191503047943115, + "num_tokens": 350718577.0, + "step": 13757 + }, + { + "epoch": 1.5108719525587526, + "grad_norm": 2.064985990524292, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7033834457397461, + "num_tokens": 350746246.0, + "step": 13758 + }, + { + "epoch": 1.5109817702613662, + "grad_norm": 2.5579514503479004, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7623976469039917, + "num_tokens": 350766514.0, + "step": 13759 + }, + { + "epoch": 1.5110915879639797, + "grad_norm": 2.2281253337860107, + "learning_rate": 1e-06, + "loss": 1.065, + "mean_token_accuracy": 0.6791923642158508, + "num_tokens": 350794249.0, + "step": 13760 + }, + { + "epoch": 1.5112014056665934, + "grad_norm": 2.340167284011841, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7419911026954651, + "num_tokens": 350816732.0, + "step": 13761 + }, + { + "epoch": 1.5113112233692072, + "grad_norm": 2.445894241333008, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7350778579711914, + "num_tokens": 350840214.0, + "step": 13762 + }, + { + "epoch": 1.5114210410718207, + "grad_norm": 2.139930248260498, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6911930441856384, + "num_tokens": 350869447.0, + "step": 13763 + }, + { + "epoch": 1.5115308587744345, + "grad_norm": 2.262831211090088, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7258107662200928, + "num_tokens": 350895546.0, + "step": 13764 + }, + { + "epoch": 1.511640676477048, + "grad_norm": 2.215583562850952, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7224808931350708, + "num_tokens": 350922782.0, + "step": 13765 + }, + { + "epoch": 1.5117504941796618, + "grad_norm": 2.1894216537475586, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7374112606048584, + "num_tokens": 350947701.0, + "step": 13766 + }, + { + "epoch": 1.5118603118822755, + "grad_norm": 2.136871814727783, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.722233235836029, + "num_tokens": 350975663.0, + "step": 13767 + }, + { + "epoch": 1.511970129584889, + "grad_norm": 1.9897478818893433, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7261962890625, + "num_tokens": 351005900.0, + "step": 13768 + }, + { + "epoch": 1.5120799472875026, + "grad_norm": 2.575174570083618, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7213093042373657, + "num_tokens": 351027417.0, + "step": 13769 + }, + { + "epoch": 1.5121897649901164, + "grad_norm": 2.3544492721557617, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7219842672348022, + "num_tokens": 351050957.0, + "step": 13770 + }, + { + "epoch": 1.5122995826927301, + "grad_norm": 1.9524681568145752, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7161225080490112, + "num_tokens": 351084158.0, + "step": 13771 + }, + { + "epoch": 1.5124094003953439, + "grad_norm": 2.3847129344940186, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7206082344055176, + "num_tokens": 351106432.0, + "step": 13772 + }, + { + "epoch": 1.5125192180979574, + "grad_norm": 2.1139702796936035, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7028301954269409, + "num_tokens": 351134183.0, + "step": 13773 + }, + { + "epoch": 1.512629035800571, + "grad_norm": 2.1636343002319336, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7114628553390503, + "num_tokens": 351162752.0, + "step": 13774 + }, + { + "epoch": 1.5127388535031847, + "grad_norm": 2.5768723487854004, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.731063723564148, + "num_tokens": 351182905.0, + "step": 13775 + }, + { + "epoch": 1.5128486712057985, + "grad_norm": 1.9711970090866089, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7203868627548218, + "num_tokens": 351214311.0, + "step": 13776 + }, + { + "epoch": 1.512958488908412, + "grad_norm": 2.113276958465576, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7227436900138855, + "num_tokens": 351241815.0, + "step": 13777 + }, + { + "epoch": 1.5130683066110255, + "grad_norm": 2.4685823917388916, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7159386873245239, + "num_tokens": 351264994.0, + "step": 13778 + }, + { + "epoch": 1.5131781243136393, + "grad_norm": 2.493279218673706, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7100232839584351, + "num_tokens": 351287639.0, + "step": 13779 + }, + { + "epoch": 1.513287942016253, + "grad_norm": 2.453700304031372, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7341523170471191, + "num_tokens": 351310180.0, + "step": 13780 + }, + { + "epoch": 1.5133977597188668, + "grad_norm": 2.365586996078491, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7416823506355286, + "num_tokens": 351333813.0, + "step": 13781 + }, + { + "epoch": 1.5135075774214803, + "grad_norm": 2.34564208984375, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.723089337348938, + "num_tokens": 351357633.0, + "step": 13782 + }, + { + "epoch": 1.5136173951240939, + "grad_norm": 2.0924808979034424, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7310224771499634, + "num_tokens": 351386890.0, + "step": 13783 + }, + { + "epoch": 1.5137272128267076, + "grad_norm": 2.5478200912475586, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7288796901702881, + "num_tokens": 351407495.0, + "step": 13784 + }, + { + "epoch": 1.5138370305293214, + "grad_norm": 1.9393630027770996, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.6999093294143677, + "num_tokens": 351438963.0, + "step": 13785 + }, + { + "epoch": 1.5139468482319351, + "grad_norm": 2.3187203407287598, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7409310340881348, + "num_tokens": 351464100.0, + "step": 13786 + }, + { + "epoch": 1.5140566659345487, + "grad_norm": 2.4700989723205566, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7305355072021484, + "num_tokens": 351484974.0, + "step": 13787 + }, + { + "epoch": 1.5141664836371622, + "grad_norm": 2.369913339614868, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7066501975059509, + "num_tokens": 351509468.0, + "step": 13788 + }, + { + "epoch": 1.514276301339776, + "grad_norm": 2.24312686920166, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7230755090713501, + "num_tokens": 351536056.0, + "step": 13789 + }, + { + "epoch": 1.5143861190423897, + "grad_norm": 1.977975606918335, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7284421920776367, + "num_tokens": 351565866.0, + "step": 13790 + }, + { + "epoch": 1.5144959367450033, + "grad_norm": 2.297070264816284, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7033334970474243, + "num_tokens": 351590634.0, + "step": 13791 + }, + { + "epoch": 1.5146057544476168, + "grad_norm": 2.10697340965271, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.7022066712379456, + "num_tokens": 351618482.0, + "step": 13792 + }, + { + "epoch": 1.5147155721502306, + "grad_norm": 2.2948923110961914, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7120606899261475, + "num_tokens": 351644025.0, + "step": 13793 + }, + { + "epoch": 1.5148253898528443, + "grad_norm": 2.0012028217315674, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.7064282298088074, + "num_tokens": 351678146.0, + "step": 13794 + }, + { + "epoch": 1.514935207555458, + "grad_norm": 2.315351963043213, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7279478311538696, + "num_tokens": 351701518.0, + "step": 13795 + }, + { + "epoch": 1.5150450252580716, + "grad_norm": 2.385652542114258, + "learning_rate": 1e-06, + "loss": 0.8019, + "mean_token_accuracy": 0.741859495639801, + "num_tokens": 351725307.0, + "step": 13796 + }, + { + "epoch": 1.5151548429606851, + "grad_norm": 2.2088446617126465, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.703561544418335, + "num_tokens": 351754889.0, + "step": 13797 + }, + { + "epoch": 1.515264660663299, + "grad_norm": 2.1979618072509766, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7205812335014343, + "num_tokens": 351779344.0, + "step": 13798 + }, + { + "epoch": 1.5153744783659127, + "grad_norm": 2.5742928981781006, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7394155263900757, + "num_tokens": 351798584.0, + "step": 13799 + }, + { + "epoch": 1.5154842960685262, + "grad_norm": 2.2311770915985107, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7232024073600769, + "num_tokens": 351823646.0, + "step": 13800 + }, + { + "epoch": 1.51559411377114, + "grad_norm": 2.2319393157958984, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7173926830291748, + "num_tokens": 351848370.0, + "step": 13801 + }, + { + "epoch": 1.5157039314737535, + "grad_norm": 2.2689871788024902, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7137588262557983, + "num_tokens": 351875505.0, + "step": 13802 + }, + { + "epoch": 1.5158137491763672, + "grad_norm": 2.135244607925415, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7242686152458191, + "num_tokens": 351902841.0, + "step": 13803 + }, + { + "epoch": 1.515923566878981, + "grad_norm": 2.0750732421875, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7121135592460632, + "num_tokens": 351931744.0, + "step": 13804 + }, + { + "epoch": 1.5160333845815945, + "grad_norm": 2.046046733856201, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7157365083694458, + "num_tokens": 351961275.0, + "step": 13805 + }, + { + "epoch": 1.516143202284208, + "grad_norm": 2.551941156387329, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7329320311546326, + "num_tokens": 351981991.0, + "step": 13806 + }, + { + "epoch": 1.5162530199868218, + "grad_norm": 2.2318592071533203, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7170807123184204, + "num_tokens": 352008253.0, + "step": 13807 + }, + { + "epoch": 1.5163628376894356, + "grad_norm": 2.3398187160491943, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7250370979309082, + "num_tokens": 352031198.0, + "step": 13808 + }, + { + "epoch": 1.5164726553920493, + "grad_norm": 2.3705012798309326, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7194743156433105, + "num_tokens": 352055191.0, + "step": 13809 + }, + { + "epoch": 1.5165824730946629, + "grad_norm": 2.9567394256591797, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7442455887794495, + "num_tokens": 352072993.0, + "step": 13810 + }, + { + "epoch": 1.5166922907972764, + "grad_norm": 2.3411571979522705, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7204951643943787, + "num_tokens": 352096365.0, + "step": 13811 + }, + { + "epoch": 1.5168021084998902, + "grad_norm": 2.261970281600952, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7354927659034729, + "num_tokens": 352120286.0, + "step": 13812 + }, + { + "epoch": 1.516911926202504, + "grad_norm": 2.1377146244049072, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.7064365148544312, + "num_tokens": 352147953.0, + "step": 13813 + }, + { + "epoch": 1.5170217439051175, + "grad_norm": 2.1620399951934814, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7324845194816589, + "num_tokens": 352175273.0, + "step": 13814 + }, + { + "epoch": 1.5171315616077312, + "grad_norm": 2.262247085571289, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7069458365440369, + "num_tokens": 352200452.0, + "step": 13815 + }, + { + "epoch": 1.5172413793103448, + "grad_norm": 2.256335735321045, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7265987396240234, + "num_tokens": 352225440.0, + "step": 13816 + }, + { + "epoch": 1.5173511970129585, + "grad_norm": 2.1157472133636475, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7138974666595459, + "num_tokens": 352257004.0, + "step": 13817 + }, + { + "epoch": 1.5174610147155723, + "grad_norm": 2.251163959503174, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7132082581520081, + "num_tokens": 352283993.0, + "step": 13818 + }, + { + "epoch": 1.5175708324181858, + "grad_norm": 2.3072822093963623, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7208141684532166, + "num_tokens": 352308198.0, + "step": 13819 + }, + { + "epoch": 1.5176806501207993, + "grad_norm": 2.2960739135742188, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7246626019477844, + "num_tokens": 352331485.0, + "step": 13820 + }, + { + "epoch": 1.517790467823413, + "grad_norm": 2.1748850345611572, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7392891645431519, + "num_tokens": 352358404.0, + "step": 13821 + }, + { + "epoch": 1.5179002855260268, + "grad_norm": 2.1384122371673584, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.724425196647644, + "num_tokens": 352385428.0, + "step": 13822 + }, + { + "epoch": 1.5180101032286406, + "grad_norm": 2.105490207672119, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7246988415718079, + "num_tokens": 352414349.0, + "step": 13823 + }, + { + "epoch": 1.5181199209312541, + "grad_norm": 2.7425899505615234, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7516535520553589, + "num_tokens": 352432980.0, + "step": 13824 + }, + { + "epoch": 1.5182297386338677, + "grad_norm": 2.721292734146118, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7158987522125244, + "num_tokens": 352452443.0, + "step": 13825 + }, + { + "epoch": 1.5183395563364814, + "grad_norm": 2.2463412284851074, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7220584154129028, + "num_tokens": 352479001.0, + "step": 13826 + }, + { + "epoch": 1.5184493740390952, + "grad_norm": 2.4033899307250977, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.748025119304657, + "num_tokens": 352501316.0, + "step": 13827 + }, + { + "epoch": 1.5185591917417087, + "grad_norm": 2.173891544342041, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7223284244537354, + "num_tokens": 352527924.0, + "step": 13828 + }, + { + "epoch": 1.5186690094443223, + "grad_norm": 2.496009349822998, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7018872499465942, + "num_tokens": 352550071.0, + "step": 13829 + }, + { + "epoch": 1.518778827146936, + "grad_norm": 2.535701274871826, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7469366788864136, + "num_tokens": 352570216.0, + "step": 13830 + }, + { + "epoch": 1.5188886448495498, + "grad_norm": 2.318713426589966, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7412651777267456, + "num_tokens": 352592588.0, + "step": 13831 + }, + { + "epoch": 1.5189984625521635, + "grad_norm": 2.1819028854370117, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7410024404525757, + "num_tokens": 352618444.0, + "step": 13832 + }, + { + "epoch": 1.519108280254777, + "grad_norm": 2.6400933265686035, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7213179469108582, + "num_tokens": 352639785.0, + "step": 13833 + }, + { + "epoch": 1.5192180979573906, + "grad_norm": 2.377242088317871, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7131733894348145, + "num_tokens": 352665017.0, + "step": 13834 + }, + { + "epoch": 1.5193279156600044, + "grad_norm": 2.336663007736206, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.71701580286026, + "num_tokens": 352688919.0, + "step": 13835 + }, + { + "epoch": 1.5194377333626181, + "grad_norm": 2.2479991912841797, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.734182596206665, + "num_tokens": 352711698.0, + "step": 13836 + }, + { + "epoch": 1.5195475510652319, + "grad_norm": 2.3088462352752686, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.720157265663147, + "num_tokens": 352734855.0, + "step": 13837 + }, + { + "epoch": 1.5196573687678454, + "grad_norm": 2.3714232444763184, + "learning_rate": 1e-06, + "loss": 0.8, + "mean_token_accuracy": 0.7420409917831421, + "num_tokens": 352755954.0, + "step": 13838 + }, + { + "epoch": 1.519767186470459, + "grad_norm": 2.291884183883667, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7155578136444092, + "num_tokens": 352782298.0, + "step": 13839 + }, + { + "epoch": 1.5198770041730727, + "grad_norm": 2.375169038772583, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7176141738891602, + "num_tokens": 352804955.0, + "step": 13840 + }, + { + "epoch": 1.5199868218756865, + "grad_norm": 2.026033401489258, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7222265601158142, + "num_tokens": 352834716.0, + "step": 13841 + }, + { + "epoch": 1.5200966395783, + "grad_norm": 2.1481592655181885, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6904868483543396, + "num_tokens": 352862270.0, + "step": 13842 + }, + { + "epoch": 1.5202064572809135, + "grad_norm": 2.2018697261810303, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7014575004577637, + "num_tokens": 352888628.0, + "step": 13843 + }, + { + "epoch": 1.5203162749835273, + "grad_norm": 2.1686108112335205, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7109233140945435, + "num_tokens": 352916674.0, + "step": 13844 + }, + { + "epoch": 1.520426092686141, + "grad_norm": 2.36659836769104, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7452844381332397, + "num_tokens": 352938838.0, + "step": 13845 + }, + { + "epoch": 1.5205359103887548, + "grad_norm": 2.3709232807159424, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7090259790420532, + "num_tokens": 352965234.0, + "step": 13846 + }, + { + "epoch": 1.5206457280913683, + "grad_norm": 2.1975133419036865, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7159727811813354, + "num_tokens": 352991116.0, + "step": 13847 + }, + { + "epoch": 1.5207555457939819, + "grad_norm": 2.0540919303894043, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7143252491950989, + "num_tokens": 353020735.0, + "step": 13848 + }, + { + "epoch": 1.5208653634965956, + "grad_norm": 2.3807597160339355, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7032454609870911, + "num_tokens": 353047445.0, + "step": 13849 + }, + { + "epoch": 1.5209751811992094, + "grad_norm": 2.3772099018096924, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7273108959197998, + "num_tokens": 353070364.0, + "step": 13850 + }, + { + "epoch": 1.5210849989018231, + "grad_norm": 2.0050127506256104, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6950108408927917, + "num_tokens": 353103418.0, + "step": 13851 + }, + { + "epoch": 1.5211948166044367, + "grad_norm": 2.3911526203155518, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7190772294998169, + "num_tokens": 353127551.0, + "step": 13852 + }, + { + "epoch": 1.5213046343070502, + "grad_norm": 2.324347734451294, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7279475927352905, + "num_tokens": 353150790.0, + "step": 13853 + }, + { + "epoch": 1.521414452009664, + "grad_norm": 2.454101800918579, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.6944304704666138, + "num_tokens": 353173891.0, + "step": 13854 + }, + { + "epoch": 1.5215242697122777, + "grad_norm": 2.039846897125244, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7280430793762207, + "num_tokens": 353204882.0, + "step": 13855 + }, + { + "epoch": 1.5216340874148913, + "grad_norm": 2.7403950691223145, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7232832312583923, + "num_tokens": 353223061.0, + "step": 13856 + }, + { + "epoch": 1.5217439051175048, + "grad_norm": 2.615971326828003, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7331212162971497, + "num_tokens": 353242901.0, + "step": 13857 + }, + { + "epoch": 1.5218537228201185, + "grad_norm": 2.409365177154541, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7221003174781799, + "num_tokens": 353265900.0, + "step": 13858 + }, + { + "epoch": 1.5219635405227323, + "grad_norm": 2.2599875926971436, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7154809832572937, + "num_tokens": 353291412.0, + "step": 13859 + }, + { + "epoch": 1.522073358225346, + "grad_norm": 2.185152530670166, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7064217329025269, + "num_tokens": 353317858.0, + "step": 13860 + }, + { + "epoch": 1.5221831759279596, + "grad_norm": 2.323946952819824, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7047262787818909, + "num_tokens": 353342827.0, + "step": 13861 + }, + { + "epoch": 1.5222929936305731, + "grad_norm": 2.2868473529815674, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7546401619911194, + "num_tokens": 353364832.0, + "step": 13862 + }, + { + "epoch": 1.522402811333187, + "grad_norm": 2.481948137283325, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7128874659538269, + "num_tokens": 353386437.0, + "step": 13863 + }, + { + "epoch": 1.5225126290358006, + "grad_norm": 1.9802383184432983, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7126116752624512, + "num_tokens": 353417283.0, + "step": 13864 + }, + { + "epoch": 1.5226224467384142, + "grad_norm": 2.0017433166503906, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.701035737991333, + "num_tokens": 353447783.0, + "step": 13865 + }, + { + "epoch": 1.522732264441028, + "grad_norm": 2.257211685180664, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7116520404815674, + "num_tokens": 353472111.0, + "step": 13866 + }, + { + "epoch": 1.5228420821436415, + "grad_norm": 2.142711639404297, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6982375383377075, + "num_tokens": 353499782.0, + "step": 13867 + }, + { + "epoch": 1.5229518998462552, + "grad_norm": 2.152432680130005, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7099706530570984, + "num_tokens": 353525999.0, + "step": 13868 + }, + { + "epoch": 1.523061717548869, + "grad_norm": 2.3102307319641113, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7362567782402039, + "num_tokens": 353548205.0, + "step": 13869 + }, + { + "epoch": 1.5231715352514825, + "grad_norm": 2.5768535137176514, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7360051870346069, + "num_tokens": 353569311.0, + "step": 13870 + }, + { + "epoch": 1.523281352954096, + "grad_norm": 2.5468497276306152, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7302284240722656, + "num_tokens": 353590637.0, + "step": 13871 + }, + { + "epoch": 1.5233911706567098, + "grad_norm": 2.4159910678863525, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.731147289276123, + "num_tokens": 353611919.0, + "step": 13872 + }, + { + "epoch": 1.5235009883593236, + "grad_norm": 2.2798032760620117, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7052537798881531, + "num_tokens": 353638821.0, + "step": 13873 + }, + { + "epoch": 1.5236108060619373, + "grad_norm": 2.3728115558624268, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7362561225891113, + "num_tokens": 353661010.0, + "step": 13874 + }, + { + "epoch": 1.5237206237645509, + "grad_norm": 2.302241325378418, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.710752010345459, + "num_tokens": 353686730.0, + "step": 13875 + }, + { + "epoch": 1.5238304414671644, + "grad_norm": 2.292856216430664, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7025231719017029, + "num_tokens": 353716585.0, + "step": 13876 + }, + { + "epoch": 1.5239402591697782, + "grad_norm": 2.583897829055786, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7249220609664917, + "num_tokens": 353736256.0, + "step": 13877 + }, + { + "epoch": 1.524050076872392, + "grad_norm": 2.2788867950439453, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7105157375335693, + "num_tokens": 353761048.0, + "step": 13878 + }, + { + "epoch": 1.5241598945750054, + "grad_norm": 2.0957562923431396, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.6977370381355286, + "num_tokens": 353791899.0, + "step": 13879 + }, + { + "epoch": 1.5242697122776192, + "grad_norm": 2.2821240425109863, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7040351629257202, + "num_tokens": 353815124.0, + "step": 13880 + }, + { + "epoch": 1.5243795299802327, + "grad_norm": 2.2951653003692627, + "learning_rate": 1e-06, + "loss": 0.7342, + "mean_token_accuracy": 0.7584218978881836, + "num_tokens": 353836510.0, + "step": 13881 + }, + { + "epoch": 1.5244893476828465, + "grad_norm": 2.315195322036743, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7173600792884827, + "num_tokens": 353860455.0, + "step": 13882 + }, + { + "epoch": 1.5245991653854603, + "grad_norm": 2.4307804107666016, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7040398120880127, + "num_tokens": 353882545.0, + "step": 13883 + }, + { + "epoch": 1.5247089830880738, + "grad_norm": 2.368972063064575, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7286981344223022, + "num_tokens": 353904374.0, + "step": 13884 + }, + { + "epoch": 1.5248188007906873, + "grad_norm": 2.3097448348999023, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7283477187156677, + "num_tokens": 353929027.0, + "step": 13885 + }, + { + "epoch": 1.524928618493301, + "grad_norm": 2.397433280944824, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7516478896141052, + "num_tokens": 353950209.0, + "step": 13886 + }, + { + "epoch": 1.5250384361959148, + "grad_norm": 2.421046018600464, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7237681150436401, + "num_tokens": 353971242.0, + "step": 13887 + }, + { + "epoch": 1.5251482538985286, + "grad_norm": 2.2754573822021484, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7169293761253357, + "num_tokens": 353996008.0, + "step": 13888 + }, + { + "epoch": 1.5252580716011421, + "grad_norm": 2.2668395042419434, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7224832773208618, + "num_tokens": 354020920.0, + "step": 13889 + }, + { + "epoch": 1.5253678893037557, + "grad_norm": 2.2447102069854736, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7152278423309326, + "num_tokens": 354047282.0, + "step": 13890 + }, + { + "epoch": 1.5254777070063694, + "grad_norm": 2.2841882705688477, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.7003980875015259, + "num_tokens": 354071951.0, + "step": 13891 + }, + { + "epoch": 1.5255875247089832, + "grad_norm": 2.405656337738037, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7171869874000549, + "num_tokens": 354095290.0, + "step": 13892 + }, + { + "epoch": 1.5256973424115967, + "grad_norm": 2.383803129196167, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7145892381668091, + "num_tokens": 354119117.0, + "step": 13893 + }, + { + "epoch": 1.5258071601142102, + "grad_norm": 2.2353670597076416, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7025361061096191, + "num_tokens": 354145044.0, + "step": 13894 + }, + { + "epoch": 1.525916977816824, + "grad_norm": 2.147958755493164, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7113147377967834, + "num_tokens": 354172327.0, + "step": 13895 + }, + { + "epoch": 1.5260267955194378, + "grad_norm": 2.0885813236236572, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7081531286239624, + "num_tokens": 354200711.0, + "step": 13896 + }, + { + "epoch": 1.5261366132220515, + "grad_norm": 2.440764904022217, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7307401299476624, + "num_tokens": 354222058.0, + "step": 13897 + }, + { + "epoch": 1.526246430924665, + "grad_norm": 2.516652822494507, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7285422086715698, + "num_tokens": 354242804.0, + "step": 13898 + }, + { + "epoch": 1.5263562486272786, + "grad_norm": 2.3129634857177734, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7160643935203552, + "num_tokens": 354266749.0, + "step": 13899 + }, + { + "epoch": 1.5264660663298923, + "grad_norm": 2.497931480407715, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7327655553817749, + "num_tokens": 354287351.0, + "step": 13900 + }, + { + "epoch": 1.526575884032506, + "grad_norm": 2.319486618041992, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7146936655044556, + "num_tokens": 354311672.0, + "step": 13901 + }, + { + "epoch": 1.5266857017351199, + "grad_norm": 2.3853366374969482, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.745822012424469, + "num_tokens": 354334170.0, + "step": 13902 + }, + { + "epoch": 1.5267955194377334, + "grad_norm": 2.315991163253784, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7139850854873657, + "num_tokens": 354358843.0, + "step": 13903 + }, + { + "epoch": 1.526905337140347, + "grad_norm": 2.217849016189575, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7136831879615784, + "num_tokens": 354384120.0, + "step": 13904 + }, + { + "epoch": 1.5270151548429607, + "grad_norm": 2.556575059890747, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7157618999481201, + "num_tokens": 354404963.0, + "step": 13905 + }, + { + "epoch": 1.5271249725455744, + "grad_norm": 2.3953354358673096, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7179497480392456, + "num_tokens": 354427570.0, + "step": 13906 + }, + { + "epoch": 1.527234790248188, + "grad_norm": 2.0858142375946045, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7297998666763306, + "num_tokens": 354457583.0, + "step": 13907 + }, + { + "epoch": 1.5273446079508015, + "grad_norm": 2.503769874572754, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.728753924369812, + "num_tokens": 354479890.0, + "step": 13908 + }, + { + "epoch": 1.5274544256534153, + "grad_norm": 2.316582441329956, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7428504824638367, + "num_tokens": 354503548.0, + "step": 13909 + }, + { + "epoch": 1.527564243356029, + "grad_norm": 2.7259531021118164, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7367948293685913, + "num_tokens": 354522442.0, + "step": 13910 + }, + { + "epoch": 1.5276740610586428, + "grad_norm": 2.410609722137451, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7364934682846069, + "num_tokens": 354544620.0, + "step": 13911 + }, + { + "epoch": 1.5277838787612563, + "grad_norm": 2.1438698768615723, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7242855429649353, + "num_tokens": 354571585.0, + "step": 13912 + }, + { + "epoch": 1.5278936964638699, + "grad_norm": 2.2395076751708984, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7140750885009766, + "num_tokens": 354598139.0, + "step": 13913 + }, + { + "epoch": 1.5280035141664836, + "grad_norm": 2.5524466037750244, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7299079895019531, + "num_tokens": 354618797.0, + "step": 13914 + }, + { + "epoch": 1.5281133318690974, + "grad_norm": 2.2160496711730957, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7240006923675537, + "num_tokens": 354645037.0, + "step": 13915 + }, + { + "epoch": 1.5282231495717111, + "grad_norm": 2.338799476623535, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7073920965194702, + "num_tokens": 354670911.0, + "step": 13916 + }, + { + "epoch": 1.5283329672743247, + "grad_norm": 2.1511800289154053, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7120929956436157, + "num_tokens": 354700296.0, + "step": 13917 + }, + { + "epoch": 1.5284427849769382, + "grad_norm": 2.3581669330596924, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7298803329467773, + "num_tokens": 354724293.0, + "step": 13918 + }, + { + "epoch": 1.528552602679552, + "grad_norm": 2.5755913257598877, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7451021671295166, + "num_tokens": 354745319.0, + "step": 13919 + }, + { + "epoch": 1.5286624203821657, + "grad_norm": 2.405426263809204, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7265833020210266, + "num_tokens": 354768025.0, + "step": 13920 + }, + { + "epoch": 1.5287722380847792, + "grad_norm": 1.8551515340805054, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7050740718841553, + "num_tokens": 354805823.0, + "step": 13921 + }, + { + "epoch": 1.5288820557873928, + "grad_norm": 2.4555532932281494, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7287484407424927, + "num_tokens": 354827178.0, + "step": 13922 + }, + { + "epoch": 1.5289918734900065, + "grad_norm": 2.398845672607422, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7527192831039429, + "num_tokens": 354850549.0, + "step": 13923 + }, + { + "epoch": 1.5291016911926203, + "grad_norm": 2.1947219371795654, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7026050090789795, + "num_tokens": 354879179.0, + "step": 13924 + }, + { + "epoch": 1.529211508895234, + "grad_norm": 2.6193408966064453, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7537835240364075, + "num_tokens": 354897492.0, + "step": 13925 + }, + { + "epoch": 1.5293213265978476, + "grad_norm": 2.162454843521118, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.749057948589325, + "num_tokens": 354921828.0, + "step": 13926 + }, + { + "epoch": 1.5294311443004611, + "grad_norm": 2.2909646034240723, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7220997214317322, + "num_tokens": 354946194.0, + "step": 13927 + }, + { + "epoch": 1.5295409620030749, + "grad_norm": 2.092419385910034, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7149332165718079, + "num_tokens": 354974950.0, + "step": 13928 + }, + { + "epoch": 1.5296507797056886, + "grad_norm": 2.3910114765167236, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7343463897705078, + "num_tokens": 354998748.0, + "step": 13929 + }, + { + "epoch": 1.5297605974083022, + "grad_norm": 2.139781951904297, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7111753821372986, + "num_tokens": 355027216.0, + "step": 13930 + }, + { + "epoch": 1.529870415110916, + "grad_norm": 2.1425561904907227, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7350623607635498, + "num_tokens": 355054185.0, + "step": 13931 + }, + { + "epoch": 1.5299802328135295, + "grad_norm": 2.345569372177124, + "learning_rate": 1e-06, + "loss": 0.8075, + "mean_token_accuracy": 0.7412800788879395, + "num_tokens": 355075818.0, + "step": 13932 + }, + { + "epoch": 1.5300900505161432, + "grad_norm": 2.4472763538360596, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7330042123794556, + "num_tokens": 355096190.0, + "step": 13933 + }, + { + "epoch": 1.530199868218757, + "grad_norm": 2.504960775375366, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7378056049346924, + "num_tokens": 355118737.0, + "step": 13934 + }, + { + "epoch": 1.5303096859213705, + "grad_norm": 2.6513559818267822, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7291377782821655, + "num_tokens": 355140461.0, + "step": 13935 + }, + { + "epoch": 1.530419503623984, + "grad_norm": 2.230712413787842, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7109526991844177, + "num_tokens": 355169305.0, + "step": 13936 + }, + { + "epoch": 1.5305293213265978, + "grad_norm": 2.2354962825775146, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7154830694198608, + "num_tokens": 355194592.0, + "step": 13937 + }, + { + "epoch": 1.5306391390292116, + "grad_norm": 2.147218942642212, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7392096519470215, + "num_tokens": 355220360.0, + "step": 13938 + }, + { + "epoch": 1.5307489567318253, + "grad_norm": 2.248976469039917, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7222775220870972, + "num_tokens": 355245187.0, + "step": 13939 + }, + { + "epoch": 1.5308587744344389, + "grad_norm": 1.9485533237457275, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7145196199417114, + "num_tokens": 355277080.0, + "step": 13940 + }, + { + "epoch": 1.5309685921370524, + "grad_norm": 2.060462474822998, + "learning_rate": 1e-06, + "loss": 1.0524, + "mean_token_accuracy": 0.6806555986404419, + "num_tokens": 355311459.0, + "step": 13941 + }, + { + "epoch": 1.5310784098396661, + "grad_norm": 2.4997072219848633, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7132706046104431, + "num_tokens": 355332966.0, + "step": 13942 + }, + { + "epoch": 1.53118822754228, + "grad_norm": 2.184507369995117, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7300369739532471, + "num_tokens": 355359288.0, + "step": 13943 + }, + { + "epoch": 1.5312980452448934, + "grad_norm": 2.103142738342285, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7057831883430481, + "num_tokens": 355388792.0, + "step": 13944 + }, + { + "epoch": 1.5314078629475072, + "grad_norm": 2.151219606399536, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7156813740730286, + "num_tokens": 355417494.0, + "step": 13945 + }, + { + "epoch": 1.5315176806501207, + "grad_norm": 2.4113664627075195, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.715678334236145, + "num_tokens": 355439314.0, + "step": 13946 + }, + { + "epoch": 1.5316274983527345, + "grad_norm": 2.030975341796875, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7296231389045715, + "num_tokens": 355469587.0, + "step": 13947 + }, + { + "epoch": 1.5317373160553482, + "grad_norm": 2.0449512004852295, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7246772050857544, + "num_tokens": 355503270.0, + "step": 13948 + }, + { + "epoch": 1.5318471337579618, + "grad_norm": 2.1745119094848633, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.6999236345291138, + "num_tokens": 355530181.0, + "step": 13949 + }, + { + "epoch": 1.5319569514605753, + "grad_norm": 2.305565357208252, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7160329818725586, + "num_tokens": 355555191.0, + "step": 13950 + }, + { + "epoch": 1.532066769163189, + "grad_norm": 2.3536789417266846, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7188164591789246, + "num_tokens": 355578770.0, + "step": 13951 + }, + { + "epoch": 1.5321765868658028, + "grad_norm": 1.9927678108215332, + "learning_rate": 1e-06, + "loss": 1.0327, + "mean_token_accuracy": 0.6804347038269043, + "num_tokens": 355613616.0, + "step": 13952 + }, + { + "epoch": 1.5322864045684166, + "grad_norm": 2.294480323791504, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7108928561210632, + "num_tokens": 355640633.0, + "step": 13953 + }, + { + "epoch": 1.5323962222710301, + "grad_norm": 2.2457704544067383, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7398974895477295, + "num_tokens": 355665599.0, + "step": 13954 + }, + { + "epoch": 1.5325060399736437, + "grad_norm": 2.499992609024048, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7123361825942993, + "num_tokens": 355689604.0, + "step": 13955 + }, + { + "epoch": 1.5326158576762574, + "grad_norm": 2.239464282989502, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7167896628379822, + "num_tokens": 355715325.0, + "step": 13956 + }, + { + "epoch": 1.5327256753788712, + "grad_norm": 2.1849606037139893, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.71236252784729, + "num_tokens": 355743230.0, + "step": 13957 + }, + { + "epoch": 1.5328354930814847, + "grad_norm": 2.049797296524048, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7106658816337585, + "num_tokens": 355773035.0, + "step": 13958 + }, + { + "epoch": 1.5329453107840982, + "grad_norm": 2.3333442211151123, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7244973182678223, + "num_tokens": 355796581.0, + "step": 13959 + }, + { + "epoch": 1.533055128486712, + "grad_norm": 2.1514759063720703, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7258834838867188, + "num_tokens": 355824450.0, + "step": 13960 + }, + { + "epoch": 1.5331649461893258, + "grad_norm": 2.1547181606292725, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7253018617630005, + "num_tokens": 355852070.0, + "step": 13961 + }, + { + "epoch": 1.5332747638919395, + "grad_norm": 2.106778621673584, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.6974710822105408, + "num_tokens": 355879879.0, + "step": 13962 + }, + { + "epoch": 1.533384581594553, + "grad_norm": 2.541628360748291, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7397777438163757, + "num_tokens": 355901748.0, + "step": 13963 + }, + { + "epoch": 1.5334943992971666, + "grad_norm": 2.210212469100952, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.6973543167114258, + "num_tokens": 355933313.0, + "step": 13964 + }, + { + "epoch": 1.5336042169997803, + "grad_norm": 2.569251775741577, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7122535705566406, + "num_tokens": 355953076.0, + "step": 13965 + }, + { + "epoch": 1.533714034702394, + "grad_norm": 2.4353628158569336, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7169195413589478, + "num_tokens": 355976772.0, + "step": 13966 + }, + { + "epoch": 1.5338238524050078, + "grad_norm": 2.3823788166046143, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6916522979736328, + "num_tokens": 356002581.0, + "step": 13967 + }, + { + "epoch": 1.5339336701076214, + "grad_norm": 2.6422746181488037, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7240540981292725, + "num_tokens": 356022745.0, + "step": 13968 + }, + { + "epoch": 1.534043487810235, + "grad_norm": 2.393695116043091, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.727666437625885, + "num_tokens": 356043577.0, + "step": 13969 + }, + { + "epoch": 1.5341533055128487, + "grad_norm": 2.080772638320923, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7017626166343689, + "num_tokens": 356073467.0, + "step": 13970 + }, + { + "epoch": 1.5342631232154624, + "grad_norm": 2.7968640327453613, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7276896238327026, + "num_tokens": 356092468.0, + "step": 13971 + }, + { + "epoch": 1.534372940918076, + "grad_norm": 2.030113697052002, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.6961785554885864, + "num_tokens": 356124565.0, + "step": 13972 + }, + { + "epoch": 1.5344827586206895, + "grad_norm": 2.840273141860962, + "learning_rate": 1e-06, + "loss": 0.7698, + "mean_token_accuracy": 0.7517853379249573, + "num_tokens": 356142047.0, + "step": 13973 + }, + { + "epoch": 1.5345925763233033, + "grad_norm": 2.006194591522217, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.701694905757904, + "num_tokens": 356172559.0, + "step": 13974 + }, + { + "epoch": 1.534702394025917, + "grad_norm": 2.5344717502593994, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7323105931282043, + "num_tokens": 356192673.0, + "step": 13975 + }, + { + "epoch": 1.5348122117285308, + "grad_norm": 2.2454874515533447, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7220399975776672, + "num_tokens": 356217127.0, + "step": 13976 + }, + { + "epoch": 1.5349220294311443, + "grad_norm": 1.9693354368209839, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.6924970149993896, + "num_tokens": 356248789.0, + "step": 13977 + }, + { + "epoch": 1.5350318471337578, + "grad_norm": 2.2991700172424316, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7298151254653931, + "num_tokens": 356272478.0, + "step": 13978 + }, + { + "epoch": 1.5351416648363716, + "grad_norm": 2.3430862426757812, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7120268940925598, + "num_tokens": 356298328.0, + "step": 13979 + }, + { + "epoch": 1.5352514825389854, + "grad_norm": 2.4580976963043213, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7233679294586182, + "num_tokens": 356321242.0, + "step": 13980 + }, + { + "epoch": 1.535361300241599, + "grad_norm": 2.715446949005127, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7278825044631958, + "num_tokens": 356341166.0, + "step": 13981 + }, + { + "epoch": 1.5354711179442126, + "grad_norm": 2.2537477016448975, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7211600542068481, + "num_tokens": 356366403.0, + "step": 13982 + }, + { + "epoch": 1.5355809356468262, + "grad_norm": 2.3096923828125, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7083589434623718, + "num_tokens": 356393223.0, + "step": 13983 + }, + { + "epoch": 1.53569075334944, + "grad_norm": 2.149977207183838, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7074267864227295, + "num_tokens": 356421830.0, + "step": 13984 + }, + { + "epoch": 1.5358005710520537, + "grad_norm": 2.3288168907165527, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.722723126411438, + "num_tokens": 356445354.0, + "step": 13985 + }, + { + "epoch": 1.5359103887546672, + "grad_norm": 2.3307952880859375, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7032114863395691, + "num_tokens": 356470342.0, + "step": 13986 + }, + { + "epoch": 1.5360202064572808, + "grad_norm": 2.2034153938293457, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7120794653892517, + "num_tokens": 356495294.0, + "step": 13987 + }, + { + "epoch": 1.5361300241598945, + "grad_norm": 2.448035717010498, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7370214462280273, + "num_tokens": 356517322.0, + "step": 13988 + }, + { + "epoch": 1.5362398418625083, + "grad_norm": 2.003894329071045, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7296558618545532, + "num_tokens": 356546280.0, + "step": 13989 + }, + { + "epoch": 1.536349659565122, + "grad_norm": 2.2036020755767822, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7054144144058228, + "num_tokens": 356573209.0, + "step": 13990 + }, + { + "epoch": 1.5364594772677356, + "grad_norm": 2.1489970684051514, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7161586880683899, + "num_tokens": 356601050.0, + "step": 13991 + }, + { + "epoch": 1.536569294970349, + "grad_norm": 2.223870038986206, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.6979414820671082, + "num_tokens": 356626709.0, + "step": 13992 + }, + { + "epoch": 1.5366791126729629, + "grad_norm": 2.43697190284729, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7127701044082642, + "num_tokens": 356649920.0, + "step": 13993 + }, + { + "epoch": 1.5367889303755766, + "grad_norm": 2.4693386554718018, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7298288345336914, + "num_tokens": 356671795.0, + "step": 13994 + }, + { + "epoch": 1.5368987480781902, + "grad_norm": 2.2366886138916016, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7029823064804077, + "num_tokens": 356697864.0, + "step": 13995 + }, + { + "epoch": 1.537008565780804, + "grad_norm": 2.225823163986206, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7239891886711121, + "num_tokens": 356723304.0, + "step": 13996 + }, + { + "epoch": 1.5371183834834174, + "grad_norm": 2.344468593597412, + "learning_rate": 1e-06, + "loss": 0.8015, + "mean_token_accuracy": 0.7542541027069092, + "num_tokens": 356745109.0, + "step": 13997 + }, + { + "epoch": 1.5372282011860312, + "grad_norm": 1.9485206604003906, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7425953149795532, + "num_tokens": 356777433.0, + "step": 13998 + }, + { + "epoch": 1.537338018888645, + "grad_norm": 2.2447593212127686, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7253089547157288, + "num_tokens": 356802564.0, + "step": 13999 + }, + { + "epoch": 1.5374478365912585, + "grad_norm": 2.2496790885925293, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.705436110496521, + "num_tokens": 356826383.0, + "step": 14000 + }, + { + "epoch": 1.537557654293872, + "grad_norm": 2.6544182300567627, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7192479968070984, + "num_tokens": 356846053.0, + "step": 14001 + }, + { + "epoch": 1.5376674719964858, + "grad_norm": 1.9598851203918457, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7155581712722778, + "num_tokens": 356876559.0, + "step": 14002 + }, + { + "epoch": 1.5377772896990995, + "grad_norm": 2.2489967346191406, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7202540636062622, + "num_tokens": 356902492.0, + "step": 14003 + }, + { + "epoch": 1.5378871074017133, + "grad_norm": 2.543799877166748, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7083834409713745, + "num_tokens": 356923950.0, + "step": 14004 + }, + { + "epoch": 1.5379969251043268, + "grad_norm": 2.2620906829833984, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7248207330703735, + "num_tokens": 356949396.0, + "step": 14005 + }, + { + "epoch": 1.5381067428069404, + "grad_norm": 2.221059799194336, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7020643949508667, + "num_tokens": 356977381.0, + "step": 14006 + }, + { + "epoch": 1.5382165605095541, + "grad_norm": 2.108330726623535, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7475220561027527, + "num_tokens": 357002895.0, + "step": 14007 + }, + { + "epoch": 1.5383263782121679, + "grad_norm": 2.4017248153686523, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7268521189689636, + "num_tokens": 357025267.0, + "step": 14008 + }, + { + "epoch": 1.5384361959147814, + "grad_norm": 2.157222032546997, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7394827604293823, + "num_tokens": 357053525.0, + "step": 14009 + }, + { + "epoch": 1.538546013617395, + "grad_norm": 1.958733081817627, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.6877286434173584, + "num_tokens": 357086477.0, + "step": 14010 + }, + { + "epoch": 1.5386558313200087, + "grad_norm": 2.217090606689453, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7312371730804443, + "num_tokens": 357111064.0, + "step": 14011 + }, + { + "epoch": 1.5387656490226225, + "grad_norm": 2.4230546951293945, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7427877187728882, + "num_tokens": 357132768.0, + "step": 14012 + }, + { + "epoch": 1.5388754667252362, + "grad_norm": 2.3096578121185303, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.705500602722168, + "num_tokens": 357156938.0, + "step": 14013 + }, + { + "epoch": 1.5389852844278498, + "grad_norm": 2.0661768913269043, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7138913869857788, + "num_tokens": 357186617.0, + "step": 14014 + }, + { + "epoch": 1.5390951021304633, + "grad_norm": 2.5023269653320312, + "learning_rate": 1e-06, + "loss": 0.7926, + "mean_token_accuracy": 0.7525204420089722, + "num_tokens": 357209153.0, + "step": 14015 + }, + { + "epoch": 1.539204919833077, + "grad_norm": 2.529585838317871, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7468236088752747, + "num_tokens": 357231027.0, + "step": 14016 + }, + { + "epoch": 1.5393147375356908, + "grad_norm": 2.3515188694000244, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7253749966621399, + "num_tokens": 357254325.0, + "step": 14017 + }, + { + "epoch": 1.5394245552383046, + "grad_norm": 2.1615262031555176, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7176852226257324, + "num_tokens": 357280913.0, + "step": 14018 + }, + { + "epoch": 1.539534372940918, + "grad_norm": 2.6159000396728516, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7422255277633667, + "num_tokens": 357301875.0, + "step": 14019 + }, + { + "epoch": 1.5396441906435316, + "grad_norm": 2.202254295349121, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7183926701545715, + "num_tokens": 357328930.0, + "step": 14020 + }, + { + "epoch": 1.5397540083461454, + "grad_norm": 2.3405864238739014, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.6999063491821289, + "num_tokens": 357354173.0, + "step": 14021 + }, + { + "epoch": 1.5398638260487592, + "grad_norm": 2.1465818881988525, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.71395343542099, + "num_tokens": 357380396.0, + "step": 14022 + }, + { + "epoch": 1.5399736437513727, + "grad_norm": 2.156196117401123, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7203776240348816, + "num_tokens": 357408171.0, + "step": 14023 + }, + { + "epoch": 1.5400834614539862, + "grad_norm": 2.6889989376068115, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7438133955001831, + "num_tokens": 357426698.0, + "step": 14024 + }, + { + "epoch": 1.5401932791566, + "grad_norm": 2.385432720184326, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7308118343353271, + "num_tokens": 357448767.0, + "step": 14025 + }, + { + "epoch": 1.5403030968592137, + "grad_norm": 1.8655197620391846, + "learning_rate": 1e-06, + "loss": 1.0468, + "mean_token_accuracy": 0.6841957569122314, + "num_tokens": 357484308.0, + "step": 14026 + }, + { + "epoch": 1.5404129145618275, + "grad_norm": 2.069713592529297, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.713905394077301, + "num_tokens": 357514756.0, + "step": 14027 + }, + { + "epoch": 1.540522732264441, + "grad_norm": 2.4347774982452393, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.721449613571167, + "num_tokens": 357536158.0, + "step": 14028 + }, + { + "epoch": 1.5406325499670546, + "grad_norm": 1.8968578577041626, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7169678211212158, + "num_tokens": 357570144.0, + "step": 14029 + }, + { + "epoch": 1.5407423676696683, + "grad_norm": 2.3133034706115723, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7011741399765015, + "num_tokens": 357593410.0, + "step": 14030 + }, + { + "epoch": 1.540852185372282, + "grad_norm": 2.0412380695343018, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7076300382614136, + "num_tokens": 357623470.0, + "step": 14031 + }, + { + "epoch": 1.5409620030748958, + "grad_norm": 2.268800973892212, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7315291166305542, + "num_tokens": 357647398.0, + "step": 14032 + }, + { + "epoch": 1.5410718207775094, + "grad_norm": 2.405130624771118, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7309558391571045, + "num_tokens": 357669947.0, + "step": 14033 + }, + { + "epoch": 1.541181638480123, + "grad_norm": 2.473069667816162, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7177475690841675, + "num_tokens": 357693124.0, + "step": 14034 + }, + { + "epoch": 1.5412914561827367, + "grad_norm": 2.3397092819213867, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7063268423080444, + "num_tokens": 357718378.0, + "step": 14035 + }, + { + "epoch": 1.5414012738853504, + "grad_norm": 2.300450325012207, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.726538360118866, + "num_tokens": 357742273.0, + "step": 14036 + }, + { + "epoch": 1.541511091587964, + "grad_norm": 1.9702105522155762, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7044996023178101, + "num_tokens": 357774949.0, + "step": 14037 + }, + { + "epoch": 1.5416209092905775, + "grad_norm": 2.5760886669158936, + "learning_rate": 1e-06, + "loss": 0.7936, + "mean_token_accuracy": 0.7426497936248779, + "num_tokens": 357794906.0, + "step": 14038 + }, + { + "epoch": 1.5417307269931912, + "grad_norm": 2.4982943534851074, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7072232365608215, + "num_tokens": 357816401.0, + "step": 14039 + }, + { + "epoch": 1.541840544695805, + "grad_norm": 2.7093863487243652, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7203436493873596, + "num_tokens": 357835786.0, + "step": 14040 + }, + { + "epoch": 1.5419503623984188, + "grad_norm": 2.0892462730407715, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7268227338790894, + "num_tokens": 357865378.0, + "step": 14041 + }, + { + "epoch": 1.5420601801010323, + "grad_norm": 2.6190412044525146, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7140853404998779, + "num_tokens": 357887498.0, + "step": 14042 + }, + { + "epoch": 1.5421699978036458, + "grad_norm": 2.331068992614746, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7318645715713501, + "num_tokens": 357910503.0, + "step": 14043 + }, + { + "epoch": 1.5422798155062596, + "grad_norm": 2.055790424346924, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7044377326965332, + "num_tokens": 357940177.0, + "step": 14044 + }, + { + "epoch": 1.5423896332088733, + "grad_norm": 2.503868818283081, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7080338001251221, + "num_tokens": 357961792.0, + "step": 14045 + }, + { + "epoch": 1.5424994509114869, + "grad_norm": 2.311920404434204, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.7004480361938477, + "num_tokens": 357987233.0, + "step": 14046 + }, + { + "epoch": 1.5426092686141006, + "grad_norm": 2.1044423580169678, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7150853872299194, + "num_tokens": 358014940.0, + "step": 14047 + }, + { + "epoch": 1.5427190863167142, + "grad_norm": 2.249793291091919, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7155844569206238, + "num_tokens": 358040893.0, + "step": 14048 + }, + { + "epoch": 1.542828904019328, + "grad_norm": 2.2592127323150635, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7062264680862427, + "num_tokens": 358067477.0, + "step": 14049 + }, + { + "epoch": 1.5429387217219417, + "grad_norm": 2.409377098083496, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7303932309150696, + "num_tokens": 358091601.0, + "step": 14050 + }, + { + "epoch": 1.5430485394245552, + "grad_norm": 2.02013897895813, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.684760570526123, + "num_tokens": 358124822.0, + "step": 14051 + }, + { + "epoch": 1.5431583571271688, + "grad_norm": 2.266773223876953, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7021760940551758, + "num_tokens": 358151020.0, + "step": 14052 + }, + { + "epoch": 1.5432681748297825, + "grad_norm": 2.2437987327575684, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.6942915320396423, + "num_tokens": 358178930.0, + "step": 14053 + }, + { + "epoch": 1.5433779925323963, + "grad_norm": 2.220991611480713, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7043889760971069, + "num_tokens": 358204534.0, + "step": 14054 + }, + { + "epoch": 1.54348781023501, + "grad_norm": 2.4437191486358643, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7410454154014587, + "num_tokens": 358225196.0, + "step": 14055 + }, + { + "epoch": 1.5435976279376236, + "grad_norm": 2.2117443084716797, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7079563736915588, + "num_tokens": 358251407.0, + "step": 14056 + }, + { + "epoch": 1.543707445640237, + "grad_norm": 2.2369256019592285, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7143765687942505, + "num_tokens": 358280039.0, + "step": 14057 + }, + { + "epoch": 1.5438172633428509, + "grad_norm": 1.8971199989318848, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.7026752233505249, + "num_tokens": 358313953.0, + "step": 14058 + }, + { + "epoch": 1.5439270810454646, + "grad_norm": 2.5023465156555176, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7118030786514282, + "num_tokens": 358336424.0, + "step": 14059 + }, + { + "epoch": 1.5440368987480781, + "grad_norm": 2.277104377746582, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7120944261550903, + "num_tokens": 358361334.0, + "step": 14060 + }, + { + "epoch": 1.544146716450692, + "grad_norm": 2.145374059677124, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7079722881317139, + "num_tokens": 358388333.0, + "step": 14061 + }, + { + "epoch": 1.5442565341533054, + "grad_norm": 2.3274500370025635, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7729596495628357, + "num_tokens": 358409950.0, + "step": 14062 + }, + { + "epoch": 1.5443663518559192, + "grad_norm": 2.3331871032714844, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7418581247329712, + "num_tokens": 358433633.0, + "step": 14063 + }, + { + "epoch": 1.544476169558533, + "grad_norm": 2.210461378097534, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7188540101051331, + "num_tokens": 358460525.0, + "step": 14064 + }, + { + "epoch": 1.5445859872611465, + "grad_norm": 2.3159101009368896, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7432639002799988, + "num_tokens": 358483590.0, + "step": 14065 + }, + { + "epoch": 1.54469580496376, + "grad_norm": 2.520016670227051, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7240211963653564, + "num_tokens": 358503852.0, + "step": 14066 + }, + { + "epoch": 1.5448056226663738, + "grad_norm": 2.156168222427368, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7513394355773926, + "num_tokens": 358528115.0, + "step": 14067 + }, + { + "epoch": 1.5449154403689875, + "grad_norm": 2.316601037979126, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7283211350440979, + "num_tokens": 358552504.0, + "step": 14068 + }, + { + "epoch": 1.5450252580716013, + "grad_norm": 2.3652596473693848, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7012432813644409, + "num_tokens": 358576121.0, + "step": 14069 + }, + { + "epoch": 1.5451350757742148, + "grad_norm": 2.0231196880340576, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6931377649307251, + "num_tokens": 358606858.0, + "step": 14070 + }, + { + "epoch": 1.5452448934768284, + "grad_norm": 2.4190785884857178, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7271895408630371, + "num_tokens": 358628636.0, + "step": 14071 + }, + { + "epoch": 1.5453547111794421, + "grad_norm": 1.9585487842559814, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7118980288505554, + "num_tokens": 358661829.0, + "step": 14072 + }, + { + "epoch": 1.5454645288820559, + "grad_norm": 2.2394309043884277, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.6987295150756836, + "num_tokens": 358690414.0, + "step": 14073 + }, + { + "epoch": 1.5455743465846694, + "grad_norm": 2.4040639400482178, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7320921421051025, + "num_tokens": 358711434.0, + "step": 14074 + }, + { + "epoch": 1.545684164287283, + "grad_norm": 2.4659252166748047, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7340748906135559, + "num_tokens": 358732429.0, + "step": 14075 + }, + { + "epoch": 1.5457939819898967, + "grad_norm": 2.1628918647766113, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7082926034927368, + "num_tokens": 358759009.0, + "step": 14076 + }, + { + "epoch": 1.5459037996925105, + "grad_norm": 2.415222644805908, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.7057823538780212, + "num_tokens": 358783221.0, + "step": 14077 + }, + { + "epoch": 1.5460136173951242, + "grad_norm": 2.392958879470825, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7107738256454468, + "num_tokens": 358806246.0, + "step": 14078 + }, + { + "epoch": 1.5461234350977378, + "grad_norm": 2.1604652404785156, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.6958770751953125, + "num_tokens": 358837198.0, + "step": 14079 + }, + { + "epoch": 1.5462332528003513, + "grad_norm": 2.4450109004974365, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7329996824264526, + "num_tokens": 358859752.0, + "step": 14080 + }, + { + "epoch": 1.546343070502965, + "grad_norm": 2.2982218265533447, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7218499183654785, + "num_tokens": 358885704.0, + "step": 14081 + }, + { + "epoch": 1.5464528882055788, + "grad_norm": 2.5239646434783936, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7143721580505371, + "num_tokens": 358907739.0, + "step": 14082 + }, + { + "epoch": 1.5465627059081926, + "grad_norm": 2.203979015350342, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7040972113609314, + "num_tokens": 358936459.0, + "step": 14083 + }, + { + "epoch": 1.546672523610806, + "grad_norm": 2.5143234729766846, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7173056602478027, + "num_tokens": 358959447.0, + "step": 14084 + }, + { + "epoch": 1.5467823413134196, + "grad_norm": 2.090874671936035, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7197628021240234, + "num_tokens": 358986741.0, + "step": 14085 + }, + { + "epoch": 1.5468921590160334, + "grad_norm": 2.2334587574005127, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7131770849227905, + "num_tokens": 359011324.0, + "step": 14086 + }, + { + "epoch": 1.5470019767186471, + "grad_norm": 2.20672869682312, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7258662581443787, + "num_tokens": 359037578.0, + "step": 14087 + }, + { + "epoch": 1.5471117944212607, + "grad_norm": 2.4873790740966797, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7208684086799622, + "num_tokens": 359060307.0, + "step": 14088 + }, + { + "epoch": 1.5472216121238742, + "grad_norm": 2.7623212337493896, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7293471097946167, + "num_tokens": 359079604.0, + "step": 14089 + }, + { + "epoch": 1.547331429826488, + "grad_norm": 2.4663538932800293, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7351618409156799, + "num_tokens": 359100647.0, + "step": 14090 + }, + { + "epoch": 1.5474412475291017, + "grad_norm": 2.232285499572754, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7124258279800415, + "num_tokens": 359126877.0, + "step": 14091 + }, + { + "epoch": 1.5475510652317155, + "grad_norm": 2.526175022125244, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7282384634017944, + "num_tokens": 359147750.0, + "step": 14092 + }, + { + "epoch": 1.547660882934329, + "grad_norm": 2.4279301166534424, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7329869270324707, + "num_tokens": 359170066.0, + "step": 14093 + }, + { + "epoch": 1.5477707006369426, + "grad_norm": 2.3635616302490234, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.72710120677948, + "num_tokens": 359192840.0, + "step": 14094 + }, + { + "epoch": 1.5478805183395563, + "grad_norm": 2.297027111053467, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7250234484672546, + "num_tokens": 359216079.0, + "step": 14095 + }, + { + "epoch": 1.54799033604217, + "grad_norm": 2.316211700439453, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7258554697036743, + "num_tokens": 359241873.0, + "step": 14096 + }, + { + "epoch": 1.5481001537447838, + "grad_norm": 2.083258628845215, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7305718660354614, + "num_tokens": 359269610.0, + "step": 14097 + }, + { + "epoch": 1.5482099714473974, + "grad_norm": 2.4595959186553955, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7153046131134033, + "num_tokens": 359292070.0, + "step": 14098 + }, + { + "epoch": 1.548319789150011, + "grad_norm": 2.106492042541504, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7245739698410034, + "num_tokens": 359320504.0, + "step": 14099 + }, + { + "epoch": 1.5484296068526247, + "grad_norm": 1.969415545463562, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7051375508308411, + "num_tokens": 359352776.0, + "step": 14100 + }, + { + "epoch": 1.5485394245552384, + "grad_norm": 2.3713226318359375, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7270153760910034, + "num_tokens": 359375567.0, + "step": 14101 + }, + { + "epoch": 1.548649242257852, + "grad_norm": 2.158592939376831, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6976333856582642, + "num_tokens": 359404256.0, + "step": 14102 + }, + { + "epoch": 1.5487590599604655, + "grad_norm": 2.194962739944458, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7320722341537476, + "num_tokens": 359430220.0, + "step": 14103 + }, + { + "epoch": 1.5488688776630792, + "grad_norm": 2.1470820903778076, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7183793783187866, + "num_tokens": 359457404.0, + "step": 14104 + }, + { + "epoch": 1.548978695365693, + "grad_norm": 2.0380709171295166, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.6987680792808533, + "num_tokens": 359488041.0, + "step": 14105 + }, + { + "epoch": 1.5490885130683067, + "grad_norm": 2.179800033569336, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.697838544845581, + "num_tokens": 359515810.0, + "step": 14106 + }, + { + "epoch": 1.5491983307709203, + "grad_norm": 2.261863946914673, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7212987542152405, + "num_tokens": 359539913.0, + "step": 14107 + }, + { + "epoch": 1.5493081484735338, + "grad_norm": 2.477961540222168, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.720251202583313, + "num_tokens": 359563200.0, + "step": 14108 + }, + { + "epoch": 1.5494179661761476, + "grad_norm": 2.378462314605713, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7196848392486572, + "num_tokens": 359584310.0, + "step": 14109 + }, + { + "epoch": 1.5495277838787613, + "grad_norm": 2.1560606956481934, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7154954671859741, + "num_tokens": 359613403.0, + "step": 14110 + }, + { + "epoch": 1.5496376015813749, + "grad_norm": 2.6628787517547607, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7283053398132324, + "num_tokens": 359633831.0, + "step": 14111 + }, + { + "epoch": 1.5497474192839886, + "grad_norm": 2.215656042098999, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6940951943397522, + "num_tokens": 359662108.0, + "step": 14112 + }, + { + "epoch": 1.5498572369866022, + "grad_norm": 1.9856934547424316, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7253698706626892, + "num_tokens": 359694848.0, + "step": 14113 + }, + { + "epoch": 1.549967054689216, + "grad_norm": 2.5261149406433105, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.6997888088226318, + "num_tokens": 359717024.0, + "step": 14114 + }, + { + "epoch": 1.5500768723918297, + "grad_norm": 2.400836944580078, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7316935062408447, + "num_tokens": 359738749.0, + "step": 14115 + }, + { + "epoch": 1.5501866900944432, + "grad_norm": 2.2584733963012695, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7208144664764404, + "num_tokens": 359764202.0, + "step": 14116 + }, + { + "epoch": 1.5502965077970567, + "grad_norm": 2.6601719856262207, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7290316820144653, + "num_tokens": 359784387.0, + "step": 14117 + }, + { + "epoch": 1.5504063254996705, + "grad_norm": 2.2328150272369385, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7433557510375977, + "num_tokens": 359808132.0, + "step": 14118 + }, + { + "epoch": 1.5505161432022843, + "grad_norm": 2.466050624847412, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7365341186523438, + "num_tokens": 359828112.0, + "step": 14119 + }, + { + "epoch": 1.550625960904898, + "grad_norm": 2.304980754852295, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7361952066421509, + "num_tokens": 359852532.0, + "step": 14120 + }, + { + "epoch": 1.5507357786075116, + "grad_norm": 2.4325027465820312, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7203458547592163, + "num_tokens": 359875271.0, + "step": 14121 + }, + { + "epoch": 1.550845596310125, + "grad_norm": 2.017866611480713, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7216890454292297, + "num_tokens": 359906542.0, + "step": 14122 + }, + { + "epoch": 1.5509554140127388, + "grad_norm": 2.209455966949463, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7070921063423157, + "num_tokens": 359933939.0, + "step": 14123 + }, + { + "epoch": 1.5510652317153526, + "grad_norm": 2.370177984237671, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.720848560333252, + "num_tokens": 359958626.0, + "step": 14124 + }, + { + "epoch": 1.5511750494179661, + "grad_norm": 2.1135060787200928, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7253010869026184, + "num_tokens": 359987217.0, + "step": 14125 + }, + { + "epoch": 1.55128486712058, + "grad_norm": 2.209110736846924, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7072501182556152, + "num_tokens": 360014410.0, + "step": 14126 + }, + { + "epoch": 1.5513946848231934, + "grad_norm": 2.3931241035461426, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7481499910354614, + "num_tokens": 360035840.0, + "step": 14127 + }, + { + "epoch": 1.5515045025258072, + "grad_norm": 2.260202646255493, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7427672147750854, + "num_tokens": 360060379.0, + "step": 14128 + }, + { + "epoch": 1.551614320228421, + "grad_norm": 2.362011671066284, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7151041626930237, + "num_tokens": 360085283.0, + "step": 14129 + }, + { + "epoch": 1.5517241379310345, + "grad_norm": 2.365896224975586, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7395415902137756, + "num_tokens": 360108502.0, + "step": 14130 + }, + { + "epoch": 1.551833955633648, + "grad_norm": 2.0867936611175537, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7276009321212769, + "num_tokens": 360135157.0, + "step": 14131 + }, + { + "epoch": 1.5519437733362618, + "grad_norm": 2.3480591773986816, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7217467427253723, + "num_tokens": 360162363.0, + "step": 14132 + }, + { + "epoch": 1.5520535910388755, + "grad_norm": 2.283029317855835, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7301892638206482, + "num_tokens": 360186965.0, + "step": 14133 + }, + { + "epoch": 1.5521634087414893, + "grad_norm": 2.699770927429199, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.74406898021698, + "num_tokens": 360206307.0, + "step": 14134 + }, + { + "epoch": 1.5522732264441028, + "grad_norm": 2.052185535430908, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7290458679199219, + "num_tokens": 360237442.0, + "step": 14135 + }, + { + "epoch": 1.5523830441467164, + "grad_norm": 2.1961545944213867, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7196558713912964, + "num_tokens": 360263771.0, + "step": 14136 + }, + { + "epoch": 1.55249286184933, + "grad_norm": 1.9584786891937256, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.737500786781311, + "num_tokens": 360293467.0, + "step": 14137 + }, + { + "epoch": 1.5526026795519439, + "grad_norm": 2.392062187194824, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.6974228620529175, + "num_tokens": 360317618.0, + "step": 14138 + }, + { + "epoch": 1.5527124972545574, + "grad_norm": 2.3165082931518555, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7103298902511597, + "num_tokens": 360342139.0, + "step": 14139 + }, + { + "epoch": 1.552822314957171, + "grad_norm": 2.428287982940674, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7235642671585083, + "num_tokens": 360366401.0, + "step": 14140 + }, + { + "epoch": 1.5529321326597847, + "grad_norm": 2.347604513168335, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7151970863342285, + "num_tokens": 360389681.0, + "step": 14141 + }, + { + "epoch": 1.5530419503623984, + "grad_norm": 2.0218491554260254, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7066160440444946, + "num_tokens": 360422379.0, + "step": 14142 + }, + { + "epoch": 1.5531517680650122, + "grad_norm": 2.4209272861480713, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7216233611106873, + "num_tokens": 360444759.0, + "step": 14143 + }, + { + "epoch": 1.5532615857676257, + "grad_norm": 2.5305583477020264, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7209040522575378, + "num_tokens": 360464863.0, + "step": 14144 + }, + { + "epoch": 1.5533714034702393, + "grad_norm": 2.0973849296569824, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7117561101913452, + "num_tokens": 360491418.0, + "step": 14145 + }, + { + "epoch": 1.553481221172853, + "grad_norm": 2.173814535140991, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7306079864501953, + "num_tokens": 360517489.0, + "step": 14146 + }, + { + "epoch": 1.5535910388754668, + "grad_norm": 2.3170166015625, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7110011577606201, + "num_tokens": 360542585.0, + "step": 14147 + }, + { + "epoch": 1.5537008565780805, + "grad_norm": 2.4258711338043213, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7380000948905945, + "num_tokens": 360565660.0, + "step": 14148 + }, + { + "epoch": 1.553810674280694, + "grad_norm": 2.304502010345459, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7259514927864075, + "num_tokens": 360588632.0, + "step": 14149 + }, + { + "epoch": 1.5539204919833076, + "grad_norm": 2.3023931980133057, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.7044439315795898, + "num_tokens": 360611755.0, + "step": 14150 + }, + { + "epoch": 1.5540303096859214, + "grad_norm": 2.4645791053771973, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7475036382675171, + "num_tokens": 360633156.0, + "step": 14151 + }, + { + "epoch": 1.5541401273885351, + "grad_norm": 2.2704761028289795, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7216100096702576, + "num_tokens": 360658815.0, + "step": 14152 + }, + { + "epoch": 1.5542499450911487, + "grad_norm": 2.0921847820281982, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7196476459503174, + "num_tokens": 360687526.0, + "step": 14153 + }, + { + "epoch": 1.5543597627937622, + "grad_norm": 2.418508768081665, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7204986810684204, + "num_tokens": 360710396.0, + "step": 14154 + }, + { + "epoch": 1.554469580496376, + "grad_norm": 2.1898152828216553, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7336299419403076, + "num_tokens": 360737681.0, + "step": 14155 + }, + { + "epoch": 1.5545793981989897, + "grad_norm": 2.1745784282684326, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.7013328075408936, + "num_tokens": 360769150.0, + "step": 14156 + }, + { + "epoch": 1.5546892159016035, + "grad_norm": 2.2559962272644043, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7287245988845825, + "num_tokens": 360797464.0, + "step": 14157 + }, + { + "epoch": 1.554799033604217, + "grad_norm": 2.0798513889312744, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.709599494934082, + "num_tokens": 360827978.0, + "step": 14158 + }, + { + "epoch": 1.5549088513068305, + "grad_norm": 2.01037335395813, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.705773651599884, + "num_tokens": 360857880.0, + "step": 14159 + }, + { + "epoch": 1.5550186690094443, + "grad_norm": 2.194225549697876, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.700515627861023, + "num_tokens": 360887912.0, + "step": 14160 + }, + { + "epoch": 1.555128486712058, + "grad_norm": 2.1671335697174072, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7210719585418701, + "num_tokens": 360916298.0, + "step": 14161 + }, + { + "epoch": 1.5552383044146716, + "grad_norm": 2.4733328819274902, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.6968367099761963, + "num_tokens": 360939844.0, + "step": 14162 + }, + { + "epoch": 1.5553481221172853, + "grad_norm": 2.29123592376709, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.6999857425689697, + "num_tokens": 360962989.0, + "step": 14163 + }, + { + "epoch": 1.5554579398198989, + "grad_norm": 2.561502695083618, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7335718870162964, + "num_tokens": 360983354.0, + "step": 14164 + }, + { + "epoch": 1.5555677575225126, + "grad_norm": 2.2495577335357666, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7169238328933716, + "num_tokens": 361008518.0, + "step": 14165 + }, + { + "epoch": 1.5556775752251264, + "grad_norm": 2.295128345489502, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7066769599914551, + "num_tokens": 361033855.0, + "step": 14166 + }, + { + "epoch": 1.55578739292774, + "grad_norm": 2.4001481533050537, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7129384279251099, + "num_tokens": 361062551.0, + "step": 14167 + }, + { + "epoch": 1.5558972106303535, + "grad_norm": 2.214505195617676, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7059798836708069, + "num_tokens": 361089068.0, + "step": 14168 + }, + { + "epoch": 1.5560070283329672, + "grad_norm": 2.263443946838379, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7211610674858093, + "num_tokens": 361113565.0, + "step": 14169 + }, + { + "epoch": 1.556116846035581, + "grad_norm": 2.1752519607543945, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7187116146087646, + "num_tokens": 361141592.0, + "step": 14170 + }, + { + "epoch": 1.5562266637381947, + "grad_norm": 2.2646379470825195, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7166593074798584, + "num_tokens": 361167081.0, + "step": 14171 + }, + { + "epoch": 1.5563364814408083, + "grad_norm": 2.5615878105163574, + "learning_rate": 1e-06, + "loss": 0.8118, + "mean_token_accuracy": 0.7436133027076721, + "num_tokens": 361187321.0, + "step": 14172 + }, + { + "epoch": 1.5564462991434218, + "grad_norm": 2.3605940341949463, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.714061975479126, + "num_tokens": 361211720.0, + "step": 14173 + }, + { + "epoch": 1.5565561168460356, + "grad_norm": 2.5089118480682373, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7256031036376953, + "num_tokens": 361233984.0, + "step": 14174 + }, + { + "epoch": 1.5566659345486493, + "grad_norm": 2.2026076316833496, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.705260157585144, + "num_tokens": 361262175.0, + "step": 14175 + }, + { + "epoch": 1.5567757522512629, + "grad_norm": 2.697277784347534, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7165560722351074, + "num_tokens": 361282228.0, + "step": 14176 + }, + { + "epoch": 1.5568855699538766, + "grad_norm": 2.547206401824951, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7338964939117432, + "num_tokens": 361303532.0, + "step": 14177 + }, + { + "epoch": 1.5569953876564901, + "grad_norm": 2.51409912109375, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7223536968231201, + "num_tokens": 361323439.0, + "step": 14178 + }, + { + "epoch": 1.557105205359104, + "grad_norm": 2.294109344482422, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7240117788314819, + "num_tokens": 361348424.0, + "step": 14179 + }, + { + "epoch": 1.5572150230617177, + "grad_norm": 2.114140033721924, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7073925733566284, + "num_tokens": 361374588.0, + "step": 14180 + }, + { + "epoch": 1.5573248407643312, + "grad_norm": 2.253643274307251, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7352444529533386, + "num_tokens": 361397978.0, + "step": 14181 + }, + { + "epoch": 1.5574346584669447, + "grad_norm": 2.226794719696045, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7245866060256958, + "num_tokens": 361422641.0, + "step": 14182 + }, + { + "epoch": 1.5575444761695585, + "grad_norm": 2.3385815620422363, + "learning_rate": 1e-06, + "loss": 0.7658, + "mean_token_accuracy": 0.7552101612091064, + "num_tokens": 361443593.0, + "step": 14183 + }, + { + "epoch": 1.5576542938721722, + "grad_norm": 2.4940991401672363, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7216536402702332, + "num_tokens": 361465847.0, + "step": 14184 + }, + { + "epoch": 1.557764111574786, + "grad_norm": 2.212310552597046, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7269684672355652, + "num_tokens": 361493180.0, + "step": 14185 + }, + { + "epoch": 1.5578739292773995, + "grad_norm": 2.286999464035034, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7227852940559387, + "num_tokens": 361517017.0, + "step": 14186 + }, + { + "epoch": 1.557983746980013, + "grad_norm": 2.396043300628662, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.704993724822998, + "num_tokens": 361541634.0, + "step": 14187 + }, + { + "epoch": 1.5580935646826268, + "grad_norm": 2.2161805629730225, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7290491461753845, + "num_tokens": 361571548.0, + "step": 14188 + }, + { + "epoch": 1.5582033823852406, + "grad_norm": 1.9221585988998413, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7041934728622437, + "num_tokens": 361603644.0, + "step": 14189 + }, + { + "epoch": 1.5583132000878541, + "grad_norm": 2.324753522872925, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.6984754204750061, + "num_tokens": 361630035.0, + "step": 14190 + }, + { + "epoch": 1.5584230177904677, + "grad_norm": 2.2814223766326904, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7257876396179199, + "num_tokens": 361657346.0, + "step": 14191 + }, + { + "epoch": 1.5585328354930814, + "grad_norm": 2.1613285541534424, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7146965861320496, + "num_tokens": 361686461.0, + "step": 14192 + }, + { + "epoch": 1.5586426531956952, + "grad_norm": 2.3229806423187256, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7381759285926819, + "num_tokens": 361708809.0, + "step": 14193 + }, + { + "epoch": 1.558752470898309, + "grad_norm": 2.125807046890259, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7345037460327148, + "num_tokens": 361733999.0, + "step": 14194 + }, + { + "epoch": 1.5588622886009225, + "grad_norm": 2.388190507888794, + "learning_rate": 1e-06, + "loss": 0.7606, + "mean_token_accuracy": 0.7530012726783752, + "num_tokens": 361757286.0, + "step": 14195 + }, + { + "epoch": 1.558972106303536, + "grad_norm": 2.776103973388672, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7313696146011353, + "num_tokens": 361776564.0, + "step": 14196 + }, + { + "epoch": 1.5590819240061498, + "grad_norm": 2.3471381664276123, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.7122557163238525, + "num_tokens": 361802097.0, + "step": 14197 + }, + { + "epoch": 1.5591917417087635, + "grad_norm": 2.210479259490967, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7338486909866333, + "num_tokens": 361828864.0, + "step": 14198 + }, + { + "epoch": 1.5593015594113773, + "grad_norm": 2.205868721008301, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7316309809684753, + "num_tokens": 361853974.0, + "step": 14199 + }, + { + "epoch": 1.5594113771139908, + "grad_norm": 2.030977964401245, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7192667722702026, + "num_tokens": 361882781.0, + "step": 14200 + }, + { + "epoch": 1.5595211948166043, + "grad_norm": 2.2632803916931152, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.68985915184021, + "num_tokens": 361911264.0, + "step": 14201 + }, + { + "epoch": 1.559631012519218, + "grad_norm": 2.115826368331909, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7229017019271851, + "num_tokens": 361938149.0, + "step": 14202 + }, + { + "epoch": 1.5597408302218319, + "grad_norm": 2.5202786922454834, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.7491652369499207, + "num_tokens": 361959655.0, + "step": 14203 + }, + { + "epoch": 1.5598506479244454, + "grad_norm": 2.3572728633880615, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7112295031547546, + "num_tokens": 361984619.0, + "step": 14204 + }, + { + "epoch": 1.559960465627059, + "grad_norm": 2.1542489528656006, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7250226140022278, + "num_tokens": 362011693.0, + "step": 14205 + }, + { + "epoch": 1.5600702833296727, + "grad_norm": 2.3439321517944336, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6987251043319702, + "num_tokens": 362036763.0, + "step": 14206 + }, + { + "epoch": 1.5601801010322864, + "grad_norm": 2.318173885345459, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7094663381576538, + "num_tokens": 362060548.0, + "step": 14207 + }, + { + "epoch": 1.5602899187349002, + "grad_norm": 2.281049966812134, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7359117269515991, + "num_tokens": 362084990.0, + "step": 14208 + }, + { + "epoch": 1.5603997364375137, + "grad_norm": 2.0449204444885254, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7050265073776245, + "num_tokens": 362116201.0, + "step": 14209 + }, + { + "epoch": 1.5605095541401273, + "grad_norm": 2.2254221439361572, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7204827070236206, + "num_tokens": 362140320.0, + "step": 14210 + }, + { + "epoch": 1.560619371842741, + "grad_norm": 2.496044635772705, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7311860918998718, + "num_tokens": 362159880.0, + "step": 14211 + }, + { + "epoch": 1.5607291895453548, + "grad_norm": 2.1002039909362793, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7260522842407227, + "num_tokens": 362187742.0, + "step": 14212 + }, + { + "epoch": 1.5608390072479685, + "grad_norm": 2.2123517990112305, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7270702123641968, + "num_tokens": 362214601.0, + "step": 14213 + }, + { + "epoch": 1.560948824950582, + "grad_norm": 2.1165735721588135, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7196402549743652, + "num_tokens": 362241278.0, + "step": 14214 + }, + { + "epoch": 1.5610586426531956, + "grad_norm": 2.06105375289917, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7159814834594727, + "num_tokens": 362271667.0, + "step": 14215 + }, + { + "epoch": 1.5611684603558094, + "grad_norm": 2.1226320266723633, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.7048341631889343, + "num_tokens": 362301635.0, + "step": 14216 + }, + { + "epoch": 1.5612782780584231, + "grad_norm": 2.4125354290008545, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7185811996459961, + "num_tokens": 362324164.0, + "step": 14217 + }, + { + "epoch": 1.5613880957610367, + "grad_norm": 1.993963360786438, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7274469137191772, + "num_tokens": 362356130.0, + "step": 14218 + }, + { + "epoch": 1.5614979134636502, + "grad_norm": 2.4975333213806152, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7356846928596497, + "num_tokens": 362378032.0, + "step": 14219 + }, + { + "epoch": 1.561607731166264, + "grad_norm": 2.641002893447876, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7241193652153015, + "num_tokens": 362397799.0, + "step": 14220 + }, + { + "epoch": 1.5617175488688777, + "grad_norm": 2.5592591762542725, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.73186194896698, + "num_tokens": 362419433.0, + "step": 14221 + }, + { + "epoch": 1.5618273665714915, + "grad_norm": 2.3954174518585205, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7225079536437988, + "num_tokens": 362444643.0, + "step": 14222 + }, + { + "epoch": 1.561937184274105, + "grad_norm": 2.4101784229278564, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7111766934394836, + "num_tokens": 362469427.0, + "step": 14223 + }, + { + "epoch": 1.5620470019767185, + "grad_norm": 2.183471441268921, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.733262836933136, + "num_tokens": 362496419.0, + "step": 14224 + }, + { + "epoch": 1.5621568196793323, + "grad_norm": 2.5092885494232178, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7114452123641968, + "num_tokens": 362516618.0, + "step": 14225 + }, + { + "epoch": 1.562266637381946, + "grad_norm": 2.3030989170074463, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7019686698913574, + "num_tokens": 362541939.0, + "step": 14226 + }, + { + "epoch": 1.5623764550845596, + "grad_norm": 2.2093889713287354, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7252262234687805, + "num_tokens": 362569940.0, + "step": 14227 + }, + { + "epoch": 1.5624862727871733, + "grad_norm": 2.56058931350708, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7245296239852905, + "num_tokens": 362589763.0, + "step": 14228 + }, + { + "epoch": 1.5625960904897869, + "grad_norm": 2.2777099609375, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7214035987854004, + "num_tokens": 362616355.0, + "step": 14229 + }, + { + "epoch": 1.5627059081924006, + "grad_norm": 2.2131011486053467, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6951341032981873, + "num_tokens": 362644045.0, + "step": 14230 + }, + { + "epoch": 1.5628157258950144, + "grad_norm": 2.318302631378174, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7227997779846191, + "num_tokens": 362669468.0, + "step": 14231 + }, + { + "epoch": 1.562925543597628, + "grad_norm": 2.3390824794769287, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7291079759597778, + "num_tokens": 362692500.0, + "step": 14232 + }, + { + "epoch": 1.5630353613002415, + "grad_norm": 2.2693026065826416, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7222687005996704, + "num_tokens": 362716571.0, + "step": 14233 + }, + { + "epoch": 1.5631451790028552, + "grad_norm": 2.210024118423462, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7034814953804016, + "num_tokens": 362742647.0, + "step": 14234 + }, + { + "epoch": 1.563254996705469, + "grad_norm": 2.6360716819763184, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7443364858627319, + "num_tokens": 362761130.0, + "step": 14235 + }, + { + "epoch": 1.5633648144080827, + "grad_norm": 2.0522890090942383, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7177728414535522, + "num_tokens": 362791345.0, + "step": 14236 + }, + { + "epoch": 1.5634746321106963, + "grad_norm": 2.1772046089172363, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7129243016242981, + "num_tokens": 362816516.0, + "step": 14237 + }, + { + "epoch": 1.5635844498133098, + "grad_norm": 2.141103982925415, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.724461019039154, + "num_tokens": 362844963.0, + "step": 14238 + }, + { + "epoch": 1.5636942675159236, + "grad_norm": 2.3211801052093506, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7016557455062866, + "num_tokens": 362872053.0, + "step": 14239 + }, + { + "epoch": 1.5638040852185373, + "grad_norm": 2.4478228092193604, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7351218461990356, + "num_tokens": 362892783.0, + "step": 14240 + }, + { + "epoch": 1.5639139029211508, + "grad_norm": 2.12451171875, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7254268527030945, + "num_tokens": 362920516.0, + "step": 14241 + }, + { + "epoch": 1.5640237206237646, + "grad_norm": 2.058540105819702, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7016350030899048, + "num_tokens": 362951488.0, + "step": 14242 + }, + { + "epoch": 1.5641335383263781, + "grad_norm": 2.5183186531066895, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7303460836410522, + "num_tokens": 362974040.0, + "step": 14243 + }, + { + "epoch": 1.564243356028992, + "grad_norm": 2.0609817504882812, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7007908821105957, + "num_tokens": 363004354.0, + "step": 14244 + }, + { + "epoch": 1.5643531737316057, + "grad_norm": 2.0285675525665283, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7100722789764404, + "num_tokens": 363034712.0, + "step": 14245 + }, + { + "epoch": 1.5644629914342192, + "grad_norm": 2.036968946456909, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7185829877853394, + "num_tokens": 363063615.0, + "step": 14246 + }, + { + "epoch": 1.5645728091368327, + "grad_norm": 2.1939585208892822, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7434662580490112, + "num_tokens": 363088782.0, + "step": 14247 + }, + { + "epoch": 1.5646826268394465, + "grad_norm": 2.3768253326416016, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7147748470306396, + "num_tokens": 363114103.0, + "step": 14248 + }, + { + "epoch": 1.5647924445420602, + "grad_norm": 2.268493890762329, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7289150953292847, + "num_tokens": 363139606.0, + "step": 14249 + }, + { + "epoch": 1.564902262244674, + "grad_norm": 2.1195290088653564, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7169134616851807, + "num_tokens": 363165857.0, + "step": 14250 + }, + { + "epoch": 1.5650120799472875, + "grad_norm": 2.5584423542022705, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7260539531707764, + "num_tokens": 363185726.0, + "step": 14251 + }, + { + "epoch": 1.565121897649901, + "grad_norm": 2.382077932357788, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7086187601089478, + "num_tokens": 363209048.0, + "step": 14252 + }, + { + "epoch": 1.5652317153525148, + "grad_norm": 2.1390953063964844, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7216199636459351, + "num_tokens": 363236863.0, + "step": 14253 + }, + { + "epoch": 1.5653415330551286, + "grad_norm": 2.1487038135528564, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.711347222328186, + "num_tokens": 363264707.0, + "step": 14254 + }, + { + "epoch": 1.565451350757742, + "grad_norm": 2.8005919456481934, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7459931373596191, + "num_tokens": 363282219.0, + "step": 14255 + }, + { + "epoch": 1.5655611684603556, + "grad_norm": 2.6152970790863037, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.6999837756156921, + "num_tokens": 363303808.0, + "step": 14256 + }, + { + "epoch": 1.5656709861629694, + "grad_norm": 2.003105878829956, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7186310887336731, + "num_tokens": 363338197.0, + "step": 14257 + }, + { + "epoch": 1.5657808038655832, + "grad_norm": 2.5168557167053223, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7435959577560425, + "num_tokens": 363360442.0, + "step": 14258 + }, + { + "epoch": 1.565890621568197, + "grad_norm": 2.0934979915618896, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7147220373153687, + "num_tokens": 363391999.0, + "step": 14259 + }, + { + "epoch": 1.5660004392708105, + "grad_norm": 2.241687059402466, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7358448505401611, + "num_tokens": 363414316.0, + "step": 14260 + }, + { + "epoch": 1.566110256973424, + "grad_norm": 2.112466812133789, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7106986045837402, + "num_tokens": 363444049.0, + "step": 14261 + }, + { + "epoch": 1.5662200746760377, + "grad_norm": 2.0916361808776855, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7040019035339355, + "num_tokens": 363473539.0, + "step": 14262 + }, + { + "epoch": 1.5663298923786515, + "grad_norm": 2.32882022857666, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7094268798828125, + "num_tokens": 363497813.0, + "step": 14263 + }, + { + "epoch": 1.5664397100812653, + "grad_norm": 2.210634469985962, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7098895311355591, + "num_tokens": 363523227.0, + "step": 14264 + }, + { + "epoch": 1.5665495277838788, + "grad_norm": 2.16182017326355, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7129480242729187, + "num_tokens": 363551408.0, + "step": 14265 + }, + { + "epoch": 1.5666593454864923, + "grad_norm": 2.163022756576538, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7105077505111694, + "num_tokens": 363579611.0, + "step": 14266 + }, + { + "epoch": 1.566769163189106, + "grad_norm": 1.9376747608184814, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6995521783828735, + "num_tokens": 363613810.0, + "step": 14267 + }, + { + "epoch": 1.5668789808917198, + "grad_norm": 2.189133882522583, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7207356691360474, + "num_tokens": 363639880.0, + "step": 14268 + }, + { + "epoch": 1.5669887985943334, + "grad_norm": 2.24565052986145, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7086751461029053, + "num_tokens": 363664614.0, + "step": 14269 + }, + { + "epoch": 1.567098616296947, + "grad_norm": 2.2600748538970947, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7294152975082397, + "num_tokens": 363689725.0, + "step": 14270 + }, + { + "epoch": 1.5672084339995607, + "grad_norm": 2.0596330165863037, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7244154810905457, + "num_tokens": 363721180.0, + "step": 14271 + }, + { + "epoch": 1.5673182517021744, + "grad_norm": 2.200340747833252, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7151870727539062, + "num_tokens": 363745718.0, + "step": 14272 + }, + { + "epoch": 1.5674280694047882, + "grad_norm": 2.2931363582611084, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7281658053398132, + "num_tokens": 363771016.0, + "step": 14273 + }, + { + "epoch": 1.5675378871074017, + "grad_norm": 2.2675211429595947, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.719876766204834, + "num_tokens": 363796357.0, + "step": 14274 + }, + { + "epoch": 1.5676477048100153, + "grad_norm": 2.3585081100463867, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.715907096862793, + "num_tokens": 363821582.0, + "step": 14275 + }, + { + "epoch": 1.567757522512629, + "grad_norm": 2.4446966648101807, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7366993427276611, + "num_tokens": 363843139.0, + "step": 14276 + }, + { + "epoch": 1.5678673402152428, + "grad_norm": 2.242044687271118, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7342604994773865, + "num_tokens": 363868412.0, + "step": 14277 + }, + { + "epoch": 1.5679771579178565, + "grad_norm": 2.0210442543029785, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.712975025177002, + "num_tokens": 363898942.0, + "step": 14278 + }, + { + "epoch": 1.56808697562047, + "grad_norm": 2.3976504802703857, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7242540717124939, + "num_tokens": 363922816.0, + "step": 14279 + }, + { + "epoch": 1.5681967933230836, + "grad_norm": 2.1376218795776367, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.6907303333282471, + "num_tokens": 363953467.0, + "step": 14280 + }, + { + "epoch": 1.5683066110256974, + "grad_norm": 2.15040922164917, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7247495651245117, + "num_tokens": 363980885.0, + "step": 14281 + }, + { + "epoch": 1.568416428728311, + "grad_norm": 2.3144469261169434, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7162439227104187, + "num_tokens": 364006820.0, + "step": 14282 + }, + { + "epoch": 1.5685262464309246, + "grad_norm": 1.9318854808807373, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7182943820953369, + "num_tokens": 364038913.0, + "step": 14283 + }, + { + "epoch": 1.5686360641335382, + "grad_norm": 2.0523340702056885, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7097548246383667, + "num_tokens": 364067238.0, + "step": 14284 + }, + { + "epoch": 1.568745881836152, + "grad_norm": 2.3480377197265625, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7346981763839722, + "num_tokens": 364090369.0, + "step": 14285 + }, + { + "epoch": 1.5688556995387657, + "grad_norm": 2.292224407196045, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7242006063461304, + "num_tokens": 364113260.0, + "step": 14286 + }, + { + "epoch": 1.5689655172413794, + "grad_norm": 2.2672815322875977, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.7537544369697571, + "num_tokens": 364135891.0, + "step": 14287 + }, + { + "epoch": 1.569075334943993, + "grad_norm": 2.2319281101226807, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7285342216491699, + "num_tokens": 364160288.0, + "step": 14288 + }, + { + "epoch": 1.5691851526466065, + "grad_norm": 2.1076173782348633, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7100272178649902, + "num_tokens": 364188102.0, + "step": 14289 + }, + { + "epoch": 1.5692949703492203, + "grad_norm": 2.1779074668884277, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.731208086013794, + "num_tokens": 364212653.0, + "step": 14290 + }, + { + "epoch": 1.569404788051834, + "grad_norm": 1.8771324157714844, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7135910987854004, + "num_tokens": 364247237.0, + "step": 14291 + }, + { + "epoch": 1.5695146057544476, + "grad_norm": 2.182915687561035, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7332562208175659, + "num_tokens": 364272978.0, + "step": 14292 + }, + { + "epoch": 1.5696244234570613, + "grad_norm": 2.3969814777374268, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7362887859344482, + "num_tokens": 364296082.0, + "step": 14293 + }, + { + "epoch": 1.5697342411596749, + "grad_norm": 2.048272132873535, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7208287119865417, + "num_tokens": 364325717.0, + "step": 14294 + }, + { + "epoch": 1.5698440588622886, + "grad_norm": 2.266975164413452, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7242168188095093, + "num_tokens": 364350747.0, + "step": 14295 + }, + { + "epoch": 1.5699538765649024, + "grad_norm": 2.3004398345947266, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7480499744415283, + "num_tokens": 364374072.0, + "step": 14296 + }, + { + "epoch": 1.570063694267516, + "grad_norm": 2.0544636249542236, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7000281810760498, + "num_tokens": 364405252.0, + "step": 14297 + }, + { + "epoch": 1.5701735119701294, + "grad_norm": 2.308213949203491, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7323603630065918, + "num_tokens": 364428186.0, + "step": 14298 + }, + { + "epoch": 1.5702833296727432, + "grad_norm": 2.1412768363952637, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7081027030944824, + "num_tokens": 364457138.0, + "step": 14299 + }, + { + "epoch": 1.570393147375357, + "grad_norm": 2.2267494201660156, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7061116695404053, + "num_tokens": 364486050.0, + "step": 14300 + }, + { + "epoch": 1.5705029650779707, + "grad_norm": 2.2175159454345703, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6969882845878601, + "num_tokens": 364513698.0, + "step": 14301 + }, + { + "epoch": 1.5706127827805842, + "grad_norm": 2.437838315963745, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7115883231163025, + "num_tokens": 364537009.0, + "step": 14302 + }, + { + "epoch": 1.5707226004831978, + "grad_norm": 2.3839993476867676, + "learning_rate": 1e-06, + "loss": 0.7979, + "mean_token_accuracy": 0.7415773868560791, + "num_tokens": 364559158.0, + "step": 14303 + }, + { + "epoch": 1.5708324181858115, + "grad_norm": 2.0335400104522705, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7103785276412964, + "num_tokens": 364589182.0, + "step": 14304 + }, + { + "epoch": 1.5709422358884253, + "grad_norm": 2.2472033500671387, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.741417646408081, + "num_tokens": 364612085.0, + "step": 14305 + }, + { + "epoch": 1.5710520535910388, + "grad_norm": 2.1292665004730225, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7063605189323425, + "num_tokens": 364641707.0, + "step": 14306 + }, + { + "epoch": 1.5711618712936526, + "grad_norm": 2.222435235977173, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7017601132392883, + "num_tokens": 364667557.0, + "step": 14307 + }, + { + "epoch": 1.5712716889962661, + "grad_norm": 2.139631509780884, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7439864277839661, + "num_tokens": 364691999.0, + "step": 14308 + }, + { + "epoch": 1.5713815066988799, + "grad_norm": 2.307628870010376, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7414898872375488, + "num_tokens": 364714808.0, + "step": 14309 + }, + { + "epoch": 1.5714913244014936, + "grad_norm": 2.1856918334960938, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7318357825279236, + "num_tokens": 364741441.0, + "step": 14310 + }, + { + "epoch": 1.5716011421041072, + "grad_norm": 1.989742636680603, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7097374200820923, + "num_tokens": 364772494.0, + "step": 14311 + }, + { + "epoch": 1.5717109598067207, + "grad_norm": 2.148925304412842, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7149296998977661, + "num_tokens": 364799409.0, + "step": 14312 + }, + { + "epoch": 1.5718207775093345, + "grad_norm": 2.358751058578491, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7205686569213867, + "num_tokens": 364821823.0, + "step": 14313 + }, + { + "epoch": 1.5719305952119482, + "grad_norm": 2.4913525581359863, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7435311079025269, + "num_tokens": 364843286.0, + "step": 14314 + }, + { + "epoch": 1.572040412914562, + "grad_norm": 2.34071946144104, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7444671392440796, + "num_tokens": 364866671.0, + "step": 14315 + }, + { + "epoch": 1.5721502306171755, + "grad_norm": 2.076995372772217, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.716916024684906, + "num_tokens": 364897373.0, + "step": 14316 + }, + { + "epoch": 1.572260048319789, + "grad_norm": 2.3192644119262695, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.694886326789856, + "num_tokens": 364923234.0, + "step": 14317 + }, + { + "epoch": 1.5723698660224028, + "grad_norm": 2.224379062652588, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7022674679756165, + "num_tokens": 364949618.0, + "step": 14318 + }, + { + "epoch": 1.5724796837250166, + "grad_norm": 2.172212600708008, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7114008665084839, + "num_tokens": 364978257.0, + "step": 14319 + }, + { + "epoch": 1.57258950142763, + "grad_norm": 2.2389845848083496, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7214624881744385, + "num_tokens": 365004372.0, + "step": 14320 + }, + { + "epoch": 1.5726993191302436, + "grad_norm": 2.0691728591918945, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7024003267288208, + "num_tokens": 365035376.0, + "step": 14321 + }, + { + "epoch": 1.5728091368328574, + "grad_norm": 2.5533266067504883, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.719423770904541, + "num_tokens": 365056502.0, + "step": 14322 + }, + { + "epoch": 1.5729189545354711, + "grad_norm": 2.19181227684021, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7151380777359009, + "num_tokens": 365085244.0, + "step": 14323 + }, + { + "epoch": 1.573028772238085, + "grad_norm": 2.1881093978881836, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7161514163017273, + "num_tokens": 365111979.0, + "step": 14324 + }, + { + "epoch": 1.5731385899406984, + "grad_norm": 2.495626926422119, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7209023237228394, + "num_tokens": 365133865.0, + "step": 14325 + }, + { + "epoch": 1.573248407643312, + "grad_norm": 2.187448501586914, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7224134206771851, + "num_tokens": 365159449.0, + "step": 14326 + }, + { + "epoch": 1.5733582253459257, + "grad_norm": 1.9676485061645508, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7029644846916199, + "num_tokens": 365191796.0, + "step": 14327 + }, + { + "epoch": 1.5734680430485395, + "grad_norm": 2.4091672897338867, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7389867305755615, + "num_tokens": 365214071.0, + "step": 14328 + }, + { + "epoch": 1.5735778607511532, + "grad_norm": 2.3041205406188965, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.698150634765625, + "num_tokens": 365238479.0, + "step": 14329 + }, + { + "epoch": 1.5736876784537668, + "grad_norm": 2.1924333572387695, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7035094499588013, + "num_tokens": 365264425.0, + "step": 14330 + }, + { + "epoch": 1.5737974961563803, + "grad_norm": 2.532769203186035, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7428849935531616, + "num_tokens": 365283809.0, + "step": 14331 + }, + { + "epoch": 1.573907313858994, + "grad_norm": 2.19317626953125, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7194229960441589, + "num_tokens": 365308837.0, + "step": 14332 + }, + { + "epoch": 1.5740171315616078, + "grad_norm": 2.2757678031921387, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7092684507369995, + "num_tokens": 365334168.0, + "step": 14333 + }, + { + "epoch": 1.5741269492642214, + "grad_norm": 2.221803665161133, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6860837936401367, + "num_tokens": 365361600.0, + "step": 14334 + }, + { + "epoch": 1.574236766966835, + "grad_norm": 2.166896104812622, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.7022998332977295, + "num_tokens": 365387947.0, + "step": 14335 + }, + { + "epoch": 1.5743465846694487, + "grad_norm": 2.194823980331421, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7238249778747559, + "num_tokens": 365414100.0, + "step": 14336 + }, + { + "epoch": 1.5744564023720624, + "grad_norm": 2.30644154548645, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7017748355865479, + "num_tokens": 365438283.0, + "step": 14337 + }, + { + "epoch": 1.5745662200746762, + "grad_norm": 2.6059467792510986, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7270169258117676, + "num_tokens": 365457893.0, + "step": 14338 + }, + { + "epoch": 1.5746760377772897, + "grad_norm": 2.2298965454101562, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7224158048629761, + "num_tokens": 365482477.0, + "step": 14339 + }, + { + "epoch": 1.5747858554799032, + "grad_norm": 2.1539690494537354, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7062520980834961, + "num_tokens": 365509770.0, + "step": 14340 + }, + { + "epoch": 1.574895673182517, + "grad_norm": 2.1476974487304688, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7306360602378845, + "num_tokens": 365537117.0, + "step": 14341 + }, + { + "epoch": 1.5750054908851308, + "grad_norm": 2.473416328430176, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7375376224517822, + "num_tokens": 365559926.0, + "step": 14342 + }, + { + "epoch": 1.5751153085877443, + "grad_norm": 2.5458641052246094, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7209393978118896, + "num_tokens": 365580784.0, + "step": 14343 + }, + { + "epoch": 1.575225126290358, + "grad_norm": 1.9750980138778687, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6958047151565552, + "num_tokens": 365614904.0, + "step": 14344 + }, + { + "epoch": 1.5753349439929716, + "grad_norm": 2.143738031387329, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7323314547538757, + "num_tokens": 365642188.0, + "step": 14345 + }, + { + "epoch": 1.5754447616955853, + "grad_norm": 2.288309335708618, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7137342691421509, + "num_tokens": 365666721.0, + "step": 14346 + }, + { + "epoch": 1.575554579398199, + "grad_norm": 2.206843614578247, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.701984167098999, + "num_tokens": 365693718.0, + "step": 14347 + }, + { + "epoch": 1.5756643971008126, + "grad_norm": 2.4317784309387207, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.729911744594574, + "num_tokens": 365716021.0, + "step": 14348 + }, + { + "epoch": 1.5757742148034262, + "grad_norm": 2.0825445652008057, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.6995286345481873, + "num_tokens": 365750532.0, + "step": 14349 + }, + { + "epoch": 1.57588403250604, + "grad_norm": 2.414687395095825, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7269641160964966, + "num_tokens": 365772595.0, + "step": 14350 + }, + { + "epoch": 1.5759938502086537, + "grad_norm": 2.259019136428833, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7343496084213257, + "num_tokens": 365796862.0, + "step": 14351 + }, + { + "epoch": 1.5761036679112674, + "grad_norm": 2.1953086853027344, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.708442211151123, + "num_tokens": 365824962.0, + "step": 14352 + }, + { + "epoch": 1.576213485613881, + "grad_norm": 2.413856029510498, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7225147485733032, + "num_tokens": 365847495.0, + "step": 14353 + }, + { + "epoch": 1.5763233033164945, + "grad_norm": 2.2530126571655273, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7170844078063965, + "num_tokens": 365871960.0, + "step": 14354 + }, + { + "epoch": 1.5764331210191083, + "grad_norm": 2.4052720069885254, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.742080807685852, + "num_tokens": 365893903.0, + "step": 14355 + }, + { + "epoch": 1.576542938721722, + "grad_norm": 2.1762783527374268, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.6909748315811157, + "num_tokens": 365923549.0, + "step": 14356 + }, + { + "epoch": 1.5766527564243356, + "grad_norm": 2.2965588569641113, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7437425851821899, + "num_tokens": 365946435.0, + "step": 14357 + }, + { + "epoch": 1.5767625741269493, + "grad_norm": 2.200836181640625, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.7412418723106384, + "num_tokens": 365972852.0, + "step": 14358 + }, + { + "epoch": 1.5768723918295628, + "grad_norm": 1.9863090515136719, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7240718007087708, + "num_tokens": 366003944.0, + "step": 14359 + }, + { + "epoch": 1.5769822095321766, + "grad_norm": 2.2767603397369385, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7017315626144409, + "num_tokens": 366029460.0, + "step": 14360 + }, + { + "epoch": 1.5770920272347904, + "grad_norm": 2.2128889560699463, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.711923360824585, + "num_tokens": 366056347.0, + "step": 14361 + }, + { + "epoch": 1.577201844937404, + "grad_norm": 2.305598735809326, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7209392189979553, + "num_tokens": 366081685.0, + "step": 14362 + }, + { + "epoch": 1.5773116626400174, + "grad_norm": 2.022528648376465, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7016918659210205, + "num_tokens": 366113587.0, + "step": 14363 + }, + { + "epoch": 1.5774214803426312, + "grad_norm": 2.322047233581543, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7370576858520508, + "num_tokens": 366139134.0, + "step": 14364 + }, + { + "epoch": 1.577531298045245, + "grad_norm": 2.1107139587402344, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.6989342570304871, + "num_tokens": 366167772.0, + "step": 14365 + }, + { + "epoch": 1.5776411157478587, + "grad_norm": 1.942662000656128, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7005629539489746, + "num_tokens": 366200771.0, + "step": 14366 + }, + { + "epoch": 1.5777509334504722, + "grad_norm": 2.229921817779541, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7137748003005981, + "num_tokens": 366227800.0, + "step": 14367 + }, + { + "epoch": 1.5778607511530858, + "grad_norm": 2.395536184310913, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7329074144363403, + "num_tokens": 366249339.0, + "step": 14368 + }, + { + "epoch": 1.5779705688556995, + "grad_norm": 1.9946517944335938, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7162458300590515, + "num_tokens": 366280435.0, + "step": 14369 + }, + { + "epoch": 1.5780803865583133, + "grad_norm": 2.3176345825195312, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7073891162872314, + "num_tokens": 366304979.0, + "step": 14370 + }, + { + "epoch": 1.5781902042609268, + "grad_norm": 2.210239887237549, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7100057005882263, + "num_tokens": 366332188.0, + "step": 14371 + }, + { + "epoch": 1.5783000219635406, + "grad_norm": 2.276944160461426, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7351680397987366, + "num_tokens": 366356426.0, + "step": 14372 + }, + { + "epoch": 1.5784098396661541, + "grad_norm": 2.3367388248443604, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7191107869148254, + "num_tokens": 366379298.0, + "step": 14373 + }, + { + "epoch": 1.5785196573687679, + "grad_norm": 2.17930006980896, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7086507081985474, + "num_tokens": 366405566.0, + "step": 14374 + }, + { + "epoch": 1.5786294750713816, + "grad_norm": 2.5569941997528076, + "learning_rate": 1e-06, + "loss": 0.805, + "mean_token_accuracy": 0.7454057335853577, + "num_tokens": 366425103.0, + "step": 14375 + }, + { + "epoch": 1.5787392927739952, + "grad_norm": 2.159754753112793, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7465530633926392, + "num_tokens": 366451729.0, + "step": 14376 + }, + { + "epoch": 1.5788491104766087, + "grad_norm": 2.3216490745544434, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.7101764678955078, + "num_tokens": 366474566.0, + "step": 14377 + }, + { + "epoch": 1.5789589281792225, + "grad_norm": 2.2086496353149414, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7027109861373901, + "num_tokens": 366500815.0, + "step": 14378 + }, + { + "epoch": 1.5790687458818362, + "grad_norm": 2.1206681728363037, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7113934755325317, + "num_tokens": 366528965.0, + "step": 14379 + }, + { + "epoch": 1.57917856358445, + "grad_norm": 2.2557568550109863, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7523066401481628, + "num_tokens": 366552041.0, + "step": 14380 + }, + { + "epoch": 1.5792883812870635, + "grad_norm": 2.1382808685302734, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7147051095962524, + "num_tokens": 366579229.0, + "step": 14381 + }, + { + "epoch": 1.579398198989677, + "grad_norm": 2.3737094402313232, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7265850901603699, + "num_tokens": 366600897.0, + "step": 14382 + }, + { + "epoch": 1.5795080166922908, + "grad_norm": 2.258849859237671, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7120352983474731, + "num_tokens": 366625956.0, + "step": 14383 + }, + { + "epoch": 1.5796178343949046, + "grad_norm": 2.056096076965332, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7253037691116333, + "num_tokens": 366653079.0, + "step": 14384 + }, + { + "epoch": 1.579727652097518, + "grad_norm": 2.202744483947754, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7202282547950745, + "num_tokens": 366678758.0, + "step": 14385 + }, + { + "epoch": 1.5798374698001316, + "grad_norm": 2.4006361961364746, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6822534203529358, + "num_tokens": 366704511.0, + "step": 14386 + }, + { + "epoch": 1.5799472875027454, + "grad_norm": 2.2453010082244873, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.713702917098999, + "num_tokens": 366729611.0, + "step": 14387 + }, + { + "epoch": 1.5800571052053591, + "grad_norm": 2.5345468521118164, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7306385040283203, + "num_tokens": 366750046.0, + "step": 14388 + }, + { + "epoch": 1.580166922907973, + "grad_norm": 2.1279776096343994, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7311619520187378, + "num_tokens": 366774877.0, + "step": 14389 + }, + { + "epoch": 1.5802767406105864, + "grad_norm": 2.1586592197418213, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7020288705825806, + "num_tokens": 366802863.0, + "step": 14390 + }, + { + "epoch": 1.5803865583132, + "grad_norm": 2.090890884399414, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7066079378128052, + "num_tokens": 366832432.0, + "step": 14391 + }, + { + "epoch": 1.5804963760158137, + "grad_norm": 2.1373212337493896, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6861777305603027, + "num_tokens": 366862228.0, + "step": 14392 + }, + { + "epoch": 1.5806061937184275, + "grad_norm": 2.1541147232055664, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7395790219306946, + "num_tokens": 366888892.0, + "step": 14393 + }, + { + "epoch": 1.5807160114210412, + "grad_norm": 2.6786763668060303, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7202581167221069, + "num_tokens": 366909337.0, + "step": 14394 + }, + { + "epoch": 1.5808258291236548, + "grad_norm": 2.817364454269409, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7389055490493774, + "num_tokens": 366927994.0, + "step": 14395 + }, + { + "epoch": 1.5809356468262683, + "grad_norm": 2.320957899093628, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.71284019947052, + "num_tokens": 366952012.0, + "step": 14396 + }, + { + "epoch": 1.581045464528882, + "grad_norm": 2.4427809715270996, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7188169956207275, + "num_tokens": 366976557.0, + "step": 14397 + }, + { + "epoch": 1.5811552822314958, + "grad_norm": 2.313992500305176, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7216391563415527, + "num_tokens": 367002038.0, + "step": 14398 + }, + { + "epoch": 1.5812650999341094, + "grad_norm": 2.1706578731536865, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7112050652503967, + "num_tokens": 367027153.0, + "step": 14399 + }, + { + "epoch": 1.5813749176367229, + "grad_norm": 2.4084484577178955, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.724353551864624, + "num_tokens": 367047965.0, + "step": 14400 + }, + { + "epoch": 1.5814847353393366, + "grad_norm": 2.189110279083252, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7332770228385925, + "num_tokens": 367072582.0, + "step": 14401 + }, + { + "epoch": 1.5815945530419504, + "grad_norm": 2.3225412368774414, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7243739366531372, + "num_tokens": 367097976.0, + "step": 14402 + }, + { + "epoch": 1.5817043707445642, + "grad_norm": 2.58141827583313, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7227895259857178, + "num_tokens": 367118980.0, + "step": 14403 + }, + { + "epoch": 1.5818141884471777, + "grad_norm": 2.0583622455596924, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7227709293365479, + "num_tokens": 367147902.0, + "step": 14404 + }, + { + "epoch": 1.5819240061497912, + "grad_norm": 2.0735833644866943, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6980154514312744, + "num_tokens": 367177337.0, + "step": 14405 + }, + { + "epoch": 1.582033823852405, + "grad_norm": 2.1788573265075684, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7268459796905518, + "num_tokens": 367202323.0, + "step": 14406 + }, + { + "epoch": 1.5821436415550187, + "grad_norm": 2.0449674129486084, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7241489887237549, + "num_tokens": 367231096.0, + "step": 14407 + }, + { + "epoch": 1.5822534592576323, + "grad_norm": 2.0699081420898438, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7485590577125549, + "num_tokens": 367258811.0, + "step": 14408 + }, + { + "epoch": 1.582363276960246, + "grad_norm": 2.1027181148529053, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7402371764183044, + "num_tokens": 367287943.0, + "step": 14409 + }, + { + "epoch": 1.5824730946628596, + "grad_norm": 2.080946207046509, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.703152060508728, + "num_tokens": 367319339.0, + "step": 14410 + }, + { + "epoch": 1.5825829123654733, + "grad_norm": 2.250704765319824, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.717320442199707, + "num_tokens": 367344402.0, + "step": 14411 + }, + { + "epoch": 1.582692730068087, + "grad_norm": 2.148191452026367, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7513022422790527, + "num_tokens": 367370494.0, + "step": 14412 + }, + { + "epoch": 1.5828025477707006, + "grad_norm": 2.3373804092407227, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7214555144309998, + "num_tokens": 367393656.0, + "step": 14413 + }, + { + "epoch": 1.5829123654733142, + "grad_norm": 2.060514211654663, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.72068852186203, + "num_tokens": 367422235.0, + "step": 14414 + }, + { + "epoch": 1.583022183175928, + "grad_norm": 2.185396432876587, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7114375829696655, + "num_tokens": 367450179.0, + "step": 14415 + }, + { + "epoch": 1.5831320008785417, + "grad_norm": 2.524839401245117, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7222035527229309, + "num_tokens": 367471864.0, + "step": 14416 + }, + { + "epoch": 1.5832418185811554, + "grad_norm": 2.285400390625, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7177115082740784, + "num_tokens": 367496624.0, + "step": 14417 + }, + { + "epoch": 1.583351636283769, + "grad_norm": 2.172804594039917, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7137758731842041, + "num_tokens": 367523706.0, + "step": 14418 + }, + { + "epoch": 1.5834614539863825, + "grad_norm": 2.6001064777374268, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7121410965919495, + "num_tokens": 367544846.0, + "step": 14419 + }, + { + "epoch": 1.5835712716889963, + "grad_norm": 2.227414608001709, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7020535469055176, + "num_tokens": 367574216.0, + "step": 14420 + }, + { + "epoch": 1.58368108939161, + "grad_norm": 2.666769027709961, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.7460154294967651, + "num_tokens": 367593119.0, + "step": 14421 + }, + { + "epoch": 1.5837909070942235, + "grad_norm": 2.3612334728240967, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7404946088790894, + "num_tokens": 367615556.0, + "step": 14422 + }, + { + "epoch": 1.5839007247968373, + "grad_norm": 2.591482162475586, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7212613821029663, + "num_tokens": 367636746.0, + "step": 14423 + }, + { + "epoch": 1.5840105424994508, + "grad_norm": 2.2266845703125, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7209009528160095, + "num_tokens": 367661156.0, + "step": 14424 + }, + { + "epoch": 1.5841203602020646, + "grad_norm": 2.328969717025757, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.72127366065979, + "num_tokens": 367683316.0, + "step": 14425 + }, + { + "epoch": 1.5842301779046783, + "grad_norm": 2.2310190200805664, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7211196422576904, + "num_tokens": 367708790.0, + "step": 14426 + }, + { + "epoch": 1.5843399956072919, + "grad_norm": 2.0457682609558105, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7137596607208252, + "num_tokens": 367739663.0, + "step": 14427 + }, + { + "epoch": 1.5844498133099054, + "grad_norm": 2.618770122528076, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7185533046722412, + "num_tokens": 367759399.0, + "step": 14428 + }, + { + "epoch": 1.5845596310125192, + "grad_norm": 1.8584612607955933, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.704197883605957, + "num_tokens": 367793154.0, + "step": 14429 + }, + { + "epoch": 1.584669448715133, + "grad_norm": 2.4049618244171143, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7235890030860901, + "num_tokens": 367816703.0, + "step": 14430 + }, + { + "epoch": 1.5847792664177467, + "grad_norm": 2.5025112628936768, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.6938729286193848, + "num_tokens": 367838696.0, + "step": 14431 + }, + { + "epoch": 1.5848890841203602, + "grad_norm": 2.221186876296997, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.697834849357605, + "num_tokens": 367868348.0, + "step": 14432 + }, + { + "epoch": 1.5849989018229738, + "grad_norm": 2.3345634937286377, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7167248725891113, + "num_tokens": 367892683.0, + "step": 14433 + }, + { + "epoch": 1.5851087195255875, + "grad_norm": 2.63503360748291, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7499759793281555, + "num_tokens": 367912035.0, + "step": 14434 + }, + { + "epoch": 1.5852185372282013, + "grad_norm": 2.1441359519958496, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7359355092048645, + "num_tokens": 367937170.0, + "step": 14435 + }, + { + "epoch": 1.5853283549308148, + "grad_norm": 2.393794059753418, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7237078547477722, + "num_tokens": 367959886.0, + "step": 14436 + }, + { + "epoch": 1.5854381726334283, + "grad_norm": 2.5634825229644775, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7382795810699463, + "num_tokens": 367983756.0, + "step": 14437 + }, + { + "epoch": 1.585547990336042, + "grad_norm": 2.212543487548828, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7269538640975952, + "num_tokens": 368011966.0, + "step": 14438 + }, + { + "epoch": 1.5856578080386559, + "grad_norm": 2.1384713649749756, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7376624345779419, + "num_tokens": 368038796.0, + "step": 14439 + }, + { + "epoch": 1.5857676257412696, + "grad_norm": 2.338469982147217, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.711514949798584, + "num_tokens": 368064585.0, + "step": 14440 + }, + { + "epoch": 1.5858774434438832, + "grad_norm": 2.370407819747925, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7204205393791199, + "num_tokens": 368088689.0, + "step": 14441 + }, + { + "epoch": 1.5859872611464967, + "grad_norm": 2.138042688369751, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7152361869812012, + "num_tokens": 368116187.0, + "step": 14442 + }, + { + "epoch": 1.5860970788491104, + "grad_norm": 2.196436643600464, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7190815210342407, + "num_tokens": 368142354.0, + "step": 14443 + }, + { + "epoch": 1.5862068965517242, + "grad_norm": 2.369872570037842, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7100721597671509, + "num_tokens": 368164977.0, + "step": 14444 + }, + { + "epoch": 1.586316714254338, + "grad_norm": 2.1486990451812744, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7245048880577087, + "num_tokens": 368191719.0, + "step": 14445 + }, + { + "epoch": 1.5864265319569515, + "grad_norm": 2.075307607650757, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7224320769309998, + "num_tokens": 368221632.0, + "step": 14446 + }, + { + "epoch": 1.586536349659565, + "grad_norm": 2.362349271774292, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7078473567962646, + "num_tokens": 368245760.0, + "step": 14447 + }, + { + "epoch": 1.5866461673621788, + "grad_norm": 2.4742515087127686, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7171611785888672, + "num_tokens": 368267405.0, + "step": 14448 + }, + { + "epoch": 1.5867559850647925, + "grad_norm": 2.28430438041687, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7245485782623291, + "num_tokens": 368291455.0, + "step": 14449 + }, + { + "epoch": 1.586865802767406, + "grad_norm": 2.3609774112701416, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7474568486213684, + "num_tokens": 368313471.0, + "step": 14450 + }, + { + "epoch": 1.5869756204700196, + "grad_norm": 2.3583121299743652, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7423712015151978, + "num_tokens": 368336306.0, + "step": 14451 + }, + { + "epoch": 1.5870854381726334, + "grad_norm": 2.494781494140625, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7271240949630737, + "num_tokens": 368356804.0, + "step": 14452 + }, + { + "epoch": 1.5871952558752471, + "grad_norm": 2.4085073471069336, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.712710440158844, + "num_tokens": 368380256.0, + "step": 14453 + }, + { + "epoch": 1.5873050735778609, + "grad_norm": 2.3760221004486084, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7184107303619385, + "num_tokens": 368405529.0, + "step": 14454 + }, + { + "epoch": 1.5874148912804744, + "grad_norm": 2.3069679737091064, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7357547283172607, + "num_tokens": 368428896.0, + "step": 14455 + }, + { + "epoch": 1.587524708983088, + "grad_norm": 2.2250027656555176, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7066390514373779, + "num_tokens": 368455338.0, + "step": 14456 + }, + { + "epoch": 1.5876345266857017, + "grad_norm": 2.1326944828033447, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7355091571807861, + "num_tokens": 368482802.0, + "step": 14457 + }, + { + "epoch": 1.5877443443883155, + "grad_norm": 2.2108681201934814, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7054873108863831, + "num_tokens": 368509127.0, + "step": 14458 + }, + { + "epoch": 1.5878541620909292, + "grad_norm": 2.15560245513916, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.7007551789283752, + "num_tokens": 368538027.0, + "step": 14459 + }, + { + "epoch": 1.5879639797935428, + "grad_norm": 2.256399154663086, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7205888032913208, + "num_tokens": 368564171.0, + "step": 14460 + }, + { + "epoch": 1.5880737974961563, + "grad_norm": 2.1689841747283936, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7102904915809631, + "num_tokens": 368590177.0, + "step": 14461 + }, + { + "epoch": 1.58818361519877, + "grad_norm": 2.2585065364837646, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6944800615310669, + "num_tokens": 368616932.0, + "step": 14462 + }, + { + "epoch": 1.5882934329013838, + "grad_norm": 2.174147129058838, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7195914387702942, + "num_tokens": 368644600.0, + "step": 14463 + }, + { + "epoch": 1.5884032506039973, + "grad_norm": 2.613485097885132, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7354657649993896, + "num_tokens": 368664514.0, + "step": 14464 + }, + { + "epoch": 1.5885130683066109, + "grad_norm": 2.2289528846740723, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7507142424583435, + "num_tokens": 368689302.0, + "step": 14465 + }, + { + "epoch": 1.5886228860092246, + "grad_norm": 2.253906726837158, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7104088664054871, + "num_tokens": 368715522.0, + "step": 14466 + }, + { + "epoch": 1.5887327037118384, + "grad_norm": 2.131295919418335, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7123796939849854, + "num_tokens": 368743281.0, + "step": 14467 + }, + { + "epoch": 1.5888425214144521, + "grad_norm": 2.224379539489746, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7345927953720093, + "num_tokens": 368768315.0, + "step": 14468 + }, + { + "epoch": 1.5889523391170657, + "grad_norm": 2.2909421920776367, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7215167880058289, + "num_tokens": 368790849.0, + "step": 14469 + }, + { + "epoch": 1.5890621568196792, + "grad_norm": 2.150789976119995, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7239348888397217, + "num_tokens": 368818017.0, + "step": 14470 + }, + { + "epoch": 1.589171974522293, + "grad_norm": 1.9643373489379883, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7181313037872314, + "num_tokens": 368850329.0, + "step": 14471 + }, + { + "epoch": 1.5892817922249067, + "grad_norm": 2.2743642330169678, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7355952262878418, + "num_tokens": 368873576.0, + "step": 14472 + }, + { + "epoch": 1.5893916099275203, + "grad_norm": 2.514923334121704, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7306311130523682, + "num_tokens": 368892410.0, + "step": 14473 + }, + { + "epoch": 1.589501427630134, + "grad_norm": 1.995460033416748, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7156205773353577, + "num_tokens": 368922258.0, + "step": 14474 + }, + { + "epoch": 1.5896112453327476, + "grad_norm": 2.2318382263183594, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7257881164550781, + "num_tokens": 368946815.0, + "step": 14475 + }, + { + "epoch": 1.5897210630353613, + "grad_norm": 2.194812536239624, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7119216322898865, + "num_tokens": 368974789.0, + "step": 14476 + }, + { + "epoch": 1.589830880737975, + "grad_norm": 1.9469276666641235, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7273589968681335, + "num_tokens": 369008929.0, + "step": 14477 + }, + { + "epoch": 1.5899406984405886, + "grad_norm": 2.081049919128418, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7219415903091431, + "num_tokens": 369038052.0, + "step": 14478 + }, + { + "epoch": 1.5900505161432021, + "grad_norm": 2.0658226013183594, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7245265245437622, + "num_tokens": 369065602.0, + "step": 14479 + }, + { + "epoch": 1.590160333845816, + "grad_norm": 2.353620767593384, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7069565653800964, + "num_tokens": 369091094.0, + "step": 14480 + }, + { + "epoch": 1.5902701515484297, + "grad_norm": 2.3034088611602783, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7236679196357727, + "num_tokens": 369115478.0, + "step": 14481 + }, + { + "epoch": 1.5903799692510434, + "grad_norm": 2.375441074371338, + "learning_rate": 1e-06, + "loss": 0.7871, + "mean_token_accuracy": 0.7546271085739136, + "num_tokens": 369137218.0, + "step": 14482 + }, + { + "epoch": 1.590489786953657, + "grad_norm": 2.3218255043029785, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7421331405639648, + "num_tokens": 369159094.0, + "step": 14483 + }, + { + "epoch": 1.5905996046562705, + "grad_norm": 2.0604958534240723, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7375634908676147, + "num_tokens": 369187968.0, + "step": 14484 + }, + { + "epoch": 1.5907094223588842, + "grad_norm": 2.3362176418304443, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.7003787159919739, + "num_tokens": 369213803.0, + "step": 14485 + }, + { + "epoch": 1.590819240061498, + "grad_norm": 2.09971284866333, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7200416922569275, + "num_tokens": 369242002.0, + "step": 14486 + }, + { + "epoch": 1.5909290577641115, + "grad_norm": 2.3366148471832275, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7330062985420227, + "num_tokens": 369265974.0, + "step": 14487 + }, + { + "epoch": 1.5910388754667253, + "grad_norm": 2.213297128677368, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7209694385528564, + "num_tokens": 369291571.0, + "step": 14488 + }, + { + "epoch": 1.5911486931693388, + "grad_norm": 2.463840961456299, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7186667919158936, + "num_tokens": 369314209.0, + "step": 14489 + }, + { + "epoch": 1.5912585108719526, + "grad_norm": 2.048319101333618, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7325996160507202, + "num_tokens": 369342572.0, + "step": 14490 + }, + { + "epoch": 1.5913683285745663, + "grad_norm": 1.9493826627731323, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6866549849510193, + "num_tokens": 369376870.0, + "step": 14491 + }, + { + "epoch": 1.5914781462771799, + "grad_norm": 1.9354621171951294, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.699174165725708, + "num_tokens": 369410089.0, + "step": 14492 + }, + { + "epoch": 1.5915879639797934, + "grad_norm": 2.009237289428711, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7025132775306702, + "num_tokens": 369441901.0, + "step": 14493 + }, + { + "epoch": 1.5916977816824072, + "grad_norm": 2.531423568725586, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7203722596168518, + "num_tokens": 369462907.0, + "step": 14494 + }, + { + "epoch": 1.591807599385021, + "grad_norm": 2.4111998081207275, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7337414026260376, + "num_tokens": 369485822.0, + "step": 14495 + }, + { + "epoch": 1.5919174170876347, + "grad_norm": 2.208580732345581, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.763679027557373, + "num_tokens": 369511106.0, + "step": 14496 + }, + { + "epoch": 1.5920272347902482, + "grad_norm": 1.8382903337478638, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7394579648971558, + "num_tokens": 369543591.0, + "step": 14497 + }, + { + "epoch": 1.5921370524928617, + "grad_norm": 1.9350746870040894, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.700510561466217, + "num_tokens": 369577887.0, + "step": 14498 + }, + { + "epoch": 1.5922468701954755, + "grad_norm": 2.062514305114746, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7050380110740662, + "num_tokens": 369609051.0, + "step": 14499 + }, + { + "epoch": 1.5923566878980893, + "grad_norm": 2.2933461666107178, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7427386045455933, + "num_tokens": 369634299.0, + "step": 14500 + }, + { + "epoch": 1.5924665056007028, + "grad_norm": 2.7339730262756348, + "learning_rate": 1e-06, + "loss": 0.754, + "mean_token_accuracy": 0.7562211751937866, + "num_tokens": 369651932.0, + "step": 14501 + }, + { + "epoch": 1.5925763233033163, + "grad_norm": 2.0519134998321533, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7186235189437866, + "num_tokens": 369681400.0, + "step": 14502 + }, + { + "epoch": 1.59268614100593, + "grad_norm": 2.386101722717285, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7087368965148926, + "num_tokens": 369707960.0, + "step": 14503 + }, + { + "epoch": 1.5927959587085438, + "grad_norm": 2.405695676803589, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6947194337844849, + "num_tokens": 369731956.0, + "step": 14504 + }, + { + "epoch": 1.5929057764111576, + "grad_norm": 2.5337305068969727, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7170605659484863, + "num_tokens": 369751837.0, + "step": 14505 + }, + { + "epoch": 1.5930155941137711, + "grad_norm": 2.350329637527466, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.72225022315979, + "num_tokens": 369775640.0, + "step": 14506 + }, + { + "epoch": 1.5931254118163847, + "grad_norm": 2.1323657035827637, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7244006395339966, + "num_tokens": 369802406.0, + "step": 14507 + }, + { + "epoch": 1.5932352295189984, + "grad_norm": 2.023996114730835, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7146469354629517, + "num_tokens": 369831733.0, + "step": 14508 + }, + { + "epoch": 1.5933450472216122, + "grad_norm": 2.469121217727661, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7214024662971497, + "num_tokens": 369852659.0, + "step": 14509 + }, + { + "epoch": 1.593454864924226, + "grad_norm": 2.2818939685821533, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7170051336288452, + "num_tokens": 369877297.0, + "step": 14510 + }, + { + "epoch": 1.5935646826268395, + "grad_norm": 2.129427194595337, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7156409621238708, + "num_tokens": 369904368.0, + "step": 14511 + }, + { + "epoch": 1.593674500329453, + "grad_norm": 2.203653335571289, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7269901037216187, + "num_tokens": 369930412.0, + "step": 14512 + }, + { + "epoch": 1.5937843180320668, + "grad_norm": 2.410820484161377, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7083295583724976, + "num_tokens": 369952826.0, + "step": 14513 + }, + { + "epoch": 1.5938941357346805, + "grad_norm": 2.353055953979492, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7072287201881409, + "num_tokens": 369978012.0, + "step": 14514 + }, + { + "epoch": 1.594003953437294, + "grad_norm": 2.2319753170013428, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7170446515083313, + "num_tokens": 370003538.0, + "step": 14515 + }, + { + "epoch": 1.5941137711399076, + "grad_norm": 2.6280059814453125, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7261573672294617, + "num_tokens": 370022923.0, + "step": 14516 + }, + { + "epoch": 1.5942235888425214, + "grad_norm": 2.3698980808258057, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7103474140167236, + "num_tokens": 370047093.0, + "step": 14517 + }, + { + "epoch": 1.5943334065451351, + "grad_norm": 2.0763540267944336, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6959011554718018, + "num_tokens": 370078321.0, + "step": 14518 + }, + { + "epoch": 1.5944432242477489, + "grad_norm": 2.241041660308838, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7373707890510559, + "num_tokens": 370102707.0, + "step": 14519 + }, + { + "epoch": 1.5945530419503624, + "grad_norm": 2.461801767349243, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.726220428943634, + "num_tokens": 370123980.0, + "step": 14520 + }, + { + "epoch": 1.594662859652976, + "grad_norm": 2.297908067703247, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7247635126113892, + "num_tokens": 370149615.0, + "step": 14521 + }, + { + "epoch": 1.5947726773555897, + "grad_norm": 2.365586757659912, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7113925814628601, + "num_tokens": 370173274.0, + "step": 14522 + }, + { + "epoch": 1.5948824950582035, + "grad_norm": 2.118403911590576, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6890573501586914, + "num_tokens": 370202792.0, + "step": 14523 + }, + { + "epoch": 1.5949923127608172, + "grad_norm": 2.074810028076172, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7329573631286621, + "num_tokens": 370232228.0, + "step": 14524 + }, + { + "epoch": 1.5951021304634307, + "grad_norm": 2.328096628189087, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7111250758171082, + "num_tokens": 370256177.0, + "step": 14525 + }, + { + "epoch": 1.5952119481660443, + "grad_norm": 2.3295998573303223, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7014397382736206, + "num_tokens": 370281198.0, + "step": 14526 + }, + { + "epoch": 1.595321765868658, + "grad_norm": 2.3421809673309326, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7182626724243164, + "num_tokens": 370305157.0, + "step": 14527 + }, + { + "epoch": 1.5954315835712718, + "grad_norm": 2.197310447692871, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.724907636642456, + "num_tokens": 370332038.0, + "step": 14528 + }, + { + "epoch": 1.5955414012738853, + "grad_norm": 2.624934673309326, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7211183309555054, + "num_tokens": 370353648.0, + "step": 14529 + }, + { + "epoch": 1.5956512189764989, + "grad_norm": 2.091181516647339, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.720485270023346, + "num_tokens": 370382324.0, + "step": 14530 + }, + { + "epoch": 1.5957610366791126, + "grad_norm": 2.4655661582946777, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7284626364707947, + "num_tokens": 370405278.0, + "step": 14531 + }, + { + "epoch": 1.5958708543817264, + "grad_norm": 2.5371780395507812, + "learning_rate": 1e-06, + "loss": 0.7724, + "mean_token_accuracy": 0.7656422853469849, + "num_tokens": 370424757.0, + "step": 14532 + }, + { + "epoch": 1.5959806720843401, + "grad_norm": 2.183852434158325, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7110459804534912, + "num_tokens": 370453344.0, + "step": 14533 + }, + { + "epoch": 1.5960904897869537, + "grad_norm": 2.549785614013672, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7360650300979614, + "num_tokens": 370473745.0, + "step": 14534 + }, + { + "epoch": 1.5962003074895672, + "grad_norm": 2.4282729625701904, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.7017842531204224, + "num_tokens": 370497851.0, + "step": 14535 + }, + { + "epoch": 1.596310125192181, + "grad_norm": 2.2794511318206787, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7171846628189087, + "num_tokens": 370522560.0, + "step": 14536 + }, + { + "epoch": 1.5964199428947947, + "grad_norm": 2.3231894969940186, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7009525299072266, + "num_tokens": 370549170.0, + "step": 14537 + }, + { + "epoch": 1.5965297605974083, + "grad_norm": 2.25335955619812, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7366495132446289, + "num_tokens": 370574908.0, + "step": 14538 + }, + { + "epoch": 1.596639578300022, + "grad_norm": 2.146005868911743, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7243248820304871, + "num_tokens": 370602131.0, + "step": 14539 + }, + { + "epoch": 1.5967493960026355, + "grad_norm": 2.203678607940674, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.714238166809082, + "num_tokens": 370629602.0, + "step": 14540 + }, + { + "epoch": 1.5968592137052493, + "grad_norm": 2.15531849861145, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7119170427322388, + "num_tokens": 370656648.0, + "step": 14541 + }, + { + "epoch": 1.596969031407863, + "grad_norm": 2.5278565883636475, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7214569449424744, + "num_tokens": 370677399.0, + "step": 14542 + }, + { + "epoch": 1.5970788491104766, + "grad_norm": 2.391603946685791, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7312247157096863, + "num_tokens": 370698631.0, + "step": 14543 + }, + { + "epoch": 1.5971886668130901, + "grad_norm": 2.341451406478882, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7265506982803345, + "num_tokens": 370720885.0, + "step": 14544 + }, + { + "epoch": 1.5972984845157039, + "grad_norm": 2.5027358531951904, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7232518196105957, + "num_tokens": 370741477.0, + "step": 14545 + }, + { + "epoch": 1.5974083022183176, + "grad_norm": 2.135091543197632, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7040277719497681, + "num_tokens": 370769315.0, + "step": 14546 + }, + { + "epoch": 1.5975181199209314, + "grad_norm": 2.4707553386688232, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7275137305259705, + "num_tokens": 370790849.0, + "step": 14547 + }, + { + "epoch": 1.597627937623545, + "grad_norm": 2.1149442195892334, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7398660182952881, + "num_tokens": 370819327.0, + "step": 14548 + }, + { + "epoch": 1.5977377553261585, + "grad_norm": 2.217258930206299, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.733153223991394, + "num_tokens": 370845219.0, + "step": 14549 + }, + { + "epoch": 1.5978475730287722, + "grad_norm": 2.1709671020507812, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.702894926071167, + "num_tokens": 370873012.0, + "step": 14550 + }, + { + "epoch": 1.597957390731386, + "grad_norm": 2.324272871017456, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.7400354743003845, + "num_tokens": 370895543.0, + "step": 14551 + }, + { + "epoch": 1.5980672084339995, + "grad_norm": 2.294753074645996, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7200528383255005, + "num_tokens": 370920487.0, + "step": 14552 + }, + { + "epoch": 1.5981770261366133, + "grad_norm": 2.2690200805664062, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.718978226184845, + "num_tokens": 370945319.0, + "step": 14553 + }, + { + "epoch": 1.5982868438392268, + "grad_norm": 2.2067785263061523, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7296038866043091, + "num_tokens": 370972034.0, + "step": 14554 + }, + { + "epoch": 1.5983966615418406, + "grad_norm": 2.813274621963501, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7457045316696167, + "num_tokens": 370990233.0, + "step": 14555 + }, + { + "epoch": 1.5985064792444543, + "grad_norm": 2.6767513751983643, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7202622294425964, + "num_tokens": 371009932.0, + "step": 14556 + }, + { + "epoch": 1.5986162969470679, + "grad_norm": 2.201925039291382, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7102211117744446, + "num_tokens": 371039572.0, + "step": 14557 + }, + { + "epoch": 1.5987261146496814, + "grad_norm": 2.7202653884887695, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7220745086669922, + "num_tokens": 371056860.0, + "step": 14558 + }, + { + "epoch": 1.5988359323522952, + "grad_norm": 2.186992883682251, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7263208627700806, + "num_tokens": 371083149.0, + "step": 14559 + }, + { + "epoch": 1.598945750054909, + "grad_norm": 2.307373285293579, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7243341207504272, + "num_tokens": 371107834.0, + "step": 14560 + }, + { + "epoch": 1.5990555677575227, + "grad_norm": 2.2443642616271973, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7174713611602783, + "num_tokens": 371134204.0, + "step": 14561 + }, + { + "epoch": 1.5991653854601362, + "grad_norm": 2.3478593826293945, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7162611484527588, + "num_tokens": 371159990.0, + "step": 14562 + }, + { + "epoch": 1.5992752031627497, + "grad_norm": 2.4645791053771973, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7250722050666809, + "num_tokens": 371181601.0, + "step": 14563 + }, + { + "epoch": 1.5993850208653635, + "grad_norm": 2.1621553897857666, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7113648653030396, + "num_tokens": 371209107.0, + "step": 14564 + }, + { + "epoch": 1.5994948385679773, + "grad_norm": 2.2489006519317627, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7223008871078491, + "num_tokens": 371235881.0, + "step": 14565 + }, + { + "epoch": 1.5996046562705908, + "grad_norm": 2.1640784740448, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7290201187133789, + "num_tokens": 371261023.0, + "step": 14566 + }, + { + "epoch": 1.5997144739732043, + "grad_norm": 2.2354466915130615, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7115525603294373, + "num_tokens": 371288034.0, + "step": 14567 + }, + { + "epoch": 1.599824291675818, + "grad_norm": 2.185425043106079, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.710373044013977, + "num_tokens": 371314579.0, + "step": 14568 + }, + { + "epoch": 1.5999341093784318, + "grad_norm": 2.173708915710449, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7340284585952759, + "num_tokens": 371340166.0, + "step": 14569 + }, + { + "epoch": 1.6000439270810456, + "grad_norm": 2.2240588665008545, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7152878642082214, + "num_tokens": 371366910.0, + "step": 14570 + }, + { + "epoch": 1.6001537447836591, + "grad_norm": 2.225412368774414, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7197481393814087, + "num_tokens": 371391889.0, + "step": 14571 + }, + { + "epoch": 1.6002635624862727, + "grad_norm": 2.0919783115386963, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7242788076400757, + "num_tokens": 371419386.0, + "step": 14572 + }, + { + "epoch": 1.6003733801888864, + "grad_norm": 2.384225368499756, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.6999216079711914, + "num_tokens": 371444723.0, + "step": 14573 + }, + { + "epoch": 1.6004831978915002, + "grad_norm": 2.1786468029022217, + "learning_rate": 1e-06, + "loss": 0.8264, + "mean_token_accuracy": 0.7393804788589478, + "num_tokens": 371471578.0, + "step": 14574 + }, + { + "epoch": 1.600593015594114, + "grad_norm": 2.5917530059814453, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7296832799911499, + "num_tokens": 371491925.0, + "step": 14575 + }, + { + "epoch": 1.6007028332967275, + "grad_norm": 2.15297532081604, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7083355784416199, + "num_tokens": 371518278.0, + "step": 14576 + }, + { + "epoch": 1.600812650999341, + "grad_norm": 2.358504056930542, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7037471532821655, + "num_tokens": 371541522.0, + "step": 14577 + }, + { + "epoch": 1.6009224687019548, + "grad_norm": 2.719938278198242, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7501457929611206, + "num_tokens": 371558492.0, + "step": 14578 + }, + { + "epoch": 1.6010322864045685, + "grad_norm": 2.61830735206604, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7355661392211914, + "num_tokens": 371578574.0, + "step": 14579 + }, + { + "epoch": 1.601142104107182, + "grad_norm": 2.0882644653320312, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7220621109008789, + "num_tokens": 371608038.0, + "step": 14580 + }, + { + "epoch": 1.6012519218097956, + "grad_norm": 1.8563717603683472, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7327361106872559, + "num_tokens": 371643267.0, + "step": 14581 + }, + { + "epoch": 1.6013617395124093, + "grad_norm": 2.269578456878662, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7094243764877319, + "num_tokens": 371672080.0, + "step": 14582 + }, + { + "epoch": 1.601471557215023, + "grad_norm": 2.5123536586761475, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7269002199172974, + "num_tokens": 371693613.0, + "step": 14583 + }, + { + "epoch": 1.6015813749176369, + "grad_norm": 2.1865432262420654, + "learning_rate": 1e-06, + "loss": 1.025, + "mean_token_accuracy": 0.6910984516143799, + "num_tokens": 371723541.0, + "step": 14584 + }, + { + "epoch": 1.6016911926202504, + "grad_norm": 2.2814712524414062, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7307014465332031, + "num_tokens": 371747317.0, + "step": 14585 + }, + { + "epoch": 1.601801010322864, + "grad_norm": 2.3079888820648193, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7150416374206543, + "num_tokens": 371773378.0, + "step": 14586 + }, + { + "epoch": 1.6019108280254777, + "grad_norm": 2.248171329498291, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.693534791469574, + "num_tokens": 371799270.0, + "step": 14587 + }, + { + "epoch": 1.6020206457280914, + "grad_norm": 2.4493720531463623, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.712932288646698, + "num_tokens": 371821577.0, + "step": 14588 + }, + { + "epoch": 1.602130463430705, + "grad_norm": 2.232848644256592, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6881458163261414, + "num_tokens": 371848591.0, + "step": 14589 + }, + { + "epoch": 1.6022402811333187, + "grad_norm": 2.0879883766174316, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7193659543991089, + "num_tokens": 371876168.0, + "step": 14590 + }, + { + "epoch": 1.6023500988359323, + "grad_norm": 2.034791946411133, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6978054046630859, + "num_tokens": 371907252.0, + "step": 14591 + }, + { + "epoch": 1.602459916538546, + "grad_norm": 2.795092821121216, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7279289960861206, + "num_tokens": 371926054.0, + "step": 14592 + }, + { + "epoch": 1.6025697342411598, + "grad_norm": 2.192131757736206, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7106473445892334, + "num_tokens": 371954845.0, + "step": 14593 + }, + { + "epoch": 1.6026795519437733, + "grad_norm": 2.1146633625030518, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7224487066268921, + "num_tokens": 371983448.0, + "step": 14594 + }, + { + "epoch": 1.6027893696463869, + "grad_norm": 2.2727303504943848, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.7001805305480957, + "num_tokens": 372011290.0, + "step": 14595 + }, + { + "epoch": 1.6028991873490006, + "grad_norm": 2.0811927318573, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7307683229446411, + "num_tokens": 372038494.0, + "step": 14596 + }, + { + "epoch": 1.6030090050516144, + "grad_norm": 2.041736602783203, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7258695960044861, + "num_tokens": 372067140.0, + "step": 14597 + }, + { + "epoch": 1.6031188227542281, + "grad_norm": 2.1207845211029053, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.711721658706665, + "num_tokens": 372092491.0, + "step": 14598 + }, + { + "epoch": 1.6032286404568417, + "grad_norm": 2.2423770427703857, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7112340927124023, + "num_tokens": 372119015.0, + "step": 14599 + }, + { + "epoch": 1.6033384581594552, + "grad_norm": 2.1073665618896484, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7025414705276489, + "num_tokens": 372150031.0, + "step": 14600 + }, + { + "epoch": 1.603448275862069, + "grad_norm": 2.0729169845581055, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7242469787597656, + "num_tokens": 372177211.0, + "step": 14601 + }, + { + "epoch": 1.6035580935646827, + "grad_norm": 2.3402671813964844, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7563728094100952, + "num_tokens": 372200744.0, + "step": 14602 + }, + { + "epoch": 1.6036679112672962, + "grad_norm": 2.155050277709961, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7254766225814819, + "num_tokens": 372227295.0, + "step": 14603 + }, + { + "epoch": 1.60377772896991, + "grad_norm": 2.3653132915496826, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7368803024291992, + "num_tokens": 372249231.0, + "step": 14604 + }, + { + "epoch": 1.6038875466725235, + "grad_norm": 2.3763301372528076, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7445992827415466, + "num_tokens": 372273662.0, + "step": 14605 + }, + { + "epoch": 1.6039973643751373, + "grad_norm": 2.1191773414611816, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7238770723342896, + "num_tokens": 372300891.0, + "step": 14606 + }, + { + "epoch": 1.604107182077751, + "grad_norm": 2.2035772800445557, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7173750400543213, + "num_tokens": 372327934.0, + "step": 14607 + }, + { + "epoch": 1.6042169997803646, + "grad_norm": 2.322173595428467, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7360827326774597, + "num_tokens": 372351489.0, + "step": 14608 + }, + { + "epoch": 1.6043268174829781, + "grad_norm": 2.495136022567749, + "learning_rate": 1e-06, + "loss": 0.7669, + "mean_token_accuracy": 0.757420003414154, + "num_tokens": 372371218.0, + "step": 14609 + }, + { + "epoch": 1.6044366351855919, + "grad_norm": 1.9868592023849487, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7270973920822144, + "num_tokens": 372403773.0, + "step": 14610 + }, + { + "epoch": 1.6045464528882056, + "grad_norm": 2.146754026412964, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7059847116470337, + "num_tokens": 372433909.0, + "step": 14611 + }, + { + "epoch": 1.6046562705908194, + "grad_norm": 2.1157751083374023, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7023065090179443, + "num_tokens": 372465186.0, + "step": 14612 + }, + { + "epoch": 1.604766088293433, + "grad_norm": 2.266866445541382, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7330707311630249, + "num_tokens": 372489430.0, + "step": 14613 + }, + { + "epoch": 1.6048759059960465, + "grad_norm": 2.278695821762085, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.730376124382019, + "num_tokens": 372513377.0, + "step": 14614 + }, + { + "epoch": 1.6049857236986602, + "grad_norm": 2.1113598346710205, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7159629464149475, + "num_tokens": 372542790.0, + "step": 14615 + }, + { + "epoch": 1.605095541401274, + "grad_norm": 2.3135342597961426, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7133564949035645, + "num_tokens": 372566745.0, + "step": 14616 + }, + { + "epoch": 1.6052053591038875, + "grad_norm": 2.203251838684082, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7085990905761719, + "num_tokens": 372592669.0, + "step": 14617 + }, + { + "epoch": 1.605315176806501, + "grad_norm": 1.984969139099121, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7178391218185425, + "num_tokens": 372623431.0, + "step": 14618 + }, + { + "epoch": 1.6054249945091148, + "grad_norm": 2.3773930072784424, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.728442907333374, + "num_tokens": 372645728.0, + "step": 14619 + }, + { + "epoch": 1.6055348122117286, + "grad_norm": 2.039579391479492, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7000383138656616, + "num_tokens": 372674574.0, + "step": 14620 + }, + { + "epoch": 1.6056446299143423, + "grad_norm": 2.1565794944763184, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7151554822921753, + "num_tokens": 372701855.0, + "step": 14621 + }, + { + "epoch": 1.6057544476169558, + "grad_norm": 2.1590394973754883, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7194223403930664, + "num_tokens": 372729737.0, + "step": 14622 + }, + { + "epoch": 1.6058642653195694, + "grad_norm": 2.179656744003296, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7468804121017456, + "num_tokens": 372755587.0, + "step": 14623 + }, + { + "epoch": 1.6059740830221831, + "grad_norm": 2.1909589767456055, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.6988166570663452, + "num_tokens": 372781776.0, + "step": 14624 + }, + { + "epoch": 1.606083900724797, + "grad_norm": 2.6824710369110107, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7310974597930908, + "num_tokens": 372800795.0, + "step": 14625 + }, + { + "epoch": 1.6061937184274107, + "grad_norm": 2.610957622528076, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7327718734741211, + "num_tokens": 372820181.0, + "step": 14626 + }, + { + "epoch": 1.6063035361300242, + "grad_norm": 2.551915407180786, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7475771903991699, + "num_tokens": 372839991.0, + "step": 14627 + }, + { + "epoch": 1.6064133538326377, + "grad_norm": 2.1466598510742188, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7140239477157593, + "num_tokens": 372867581.0, + "step": 14628 + }, + { + "epoch": 1.6065231715352515, + "grad_norm": 2.381089925765991, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7201298475265503, + "num_tokens": 372891309.0, + "step": 14629 + }, + { + "epoch": 1.6066329892378652, + "grad_norm": 2.2790491580963135, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7089352011680603, + "num_tokens": 372918569.0, + "step": 14630 + }, + { + "epoch": 1.6067428069404788, + "grad_norm": 2.156348466873169, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6985549330711365, + "num_tokens": 372948818.0, + "step": 14631 + }, + { + "epoch": 1.6068526246430923, + "grad_norm": 2.229942560195923, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7197897434234619, + "num_tokens": 372973812.0, + "step": 14632 + }, + { + "epoch": 1.606962442345706, + "grad_norm": 2.407308578491211, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7146577835083008, + "num_tokens": 372997028.0, + "step": 14633 + }, + { + "epoch": 1.6070722600483198, + "grad_norm": 2.2626230716705322, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7169645428657532, + "num_tokens": 373022971.0, + "step": 14634 + }, + { + "epoch": 1.6071820777509336, + "grad_norm": 2.000783920288086, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7154020071029663, + "num_tokens": 373055294.0, + "step": 14635 + }, + { + "epoch": 1.6072918954535471, + "grad_norm": 2.2802317142486572, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7082723379135132, + "num_tokens": 373081442.0, + "step": 14636 + }, + { + "epoch": 1.6074017131561606, + "grad_norm": 2.2355687618255615, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7230318188667297, + "num_tokens": 373107296.0, + "step": 14637 + }, + { + "epoch": 1.6075115308587744, + "grad_norm": 2.126136064529419, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7284330129623413, + "num_tokens": 373136234.0, + "step": 14638 + }, + { + "epoch": 1.6076213485613882, + "grad_norm": 2.0599286556243896, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.709161639213562, + "num_tokens": 373166241.0, + "step": 14639 + }, + { + "epoch": 1.607731166264002, + "grad_norm": 2.3574352264404297, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7128351330757141, + "num_tokens": 373190479.0, + "step": 14640 + }, + { + "epoch": 1.6078409839666155, + "grad_norm": 2.4511959552764893, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7339574098587036, + "num_tokens": 373212071.0, + "step": 14641 + }, + { + "epoch": 1.607950801669229, + "grad_norm": 1.9345905780792236, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7303791046142578, + "num_tokens": 373243847.0, + "step": 14642 + }, + { + "epoch": 1.6080606193718427, + "grad_norm": 1.9909205436706543, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.717416524887085, + "num_tokens": 373275736.0, + "step": 14643 + }, + { + "epoch": 1.6081704370744565, + "grad_norm": 2.2444915771484375, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7212933301925659, + "num_tokens": 373302014.0, + "step": 14644 + }, + { + "epoch": 1.60828025477707, + "grad_norm": 2.4805288314819336, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7075419425964355, + "num_tokens": 373326626.0, + "step": 14645 + }, + { + "epoch": 1.6083900724796836, + "grad_norm": 2.312619686126709, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7298227548599243, + "num_tokens": 373350298.0, + "step": 14646 + }, + { + "epoch": 1.6084998901822973, + "grad_norm": 2.2733895778656006, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7266401052474976, + "num_tokens": 373375151.0, + "step": 14647 + }, + { + "epoch": 1.608609707884911, + "grad_norm": 2.3260610103607178, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7316575050354004, + "num_tokens": 373398570.0, + "step": 14648 + }, + { + "epoch": 1.6087195255875248, + "grad_norm": 2.1822872161865234, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7204949855804443, + "num_tokens": 373424635.0, + "step": 14649 + }, + { + "epoch": 1.6088293432901384, + "grad_norm": 2.193392515182495, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.714573860168457, + "num_tokens": 373450948.0, + "step": 14650 + }, + { + "epoch": 1.608939160992752, + "grad_norm": 2.34828519821167, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7168290615081787, + "num_tokens": 373473478.0, + "step": 14651 + }, + { + "epoch": 1.6090489786953657, + "grad_norm": 2.400346040725708, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7232410907745361, + "num_tokens": 373497368.0, + "step": 14652 + }, + { + "epoch": 1.6091587963979794, + "grad_norm": 2.6368865966796875, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7119913101196289, + "num_tokens": 373518402.0, + "step": 14653 + }, + { + "epoch": 1.609268614100593, + "grad_norm": 2.326221227645874, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7137860059738159, + "num_tokens": 373542171.0, + "step": 14654 + }, + { + "epoch": 1.6093784318032067, + "grad_norm": 2.117105007171631, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.694617509841919, + "num_tokens": 373572270.0, + "step": 14655 + }, + { + "epoch": 1.6094882495058203, + "grad_norm": 2.2290685176849365, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7145733833312988, + "num_tokens": 373598084.0, + "step": 14656 + }, + { + "epoch": 1.609598067208434, + "grad_norm": 2.2238783836364746, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7077465057373047, + "num_tokens": 373625706.0, + "step": 14657 + }, + { + "epoch": 1.6097078849110478, + "grad_norm": 2.669043779373169, + "learning_rate": 1e-06, + "loss": 0.8015, + "mean_token_accuracy": 0.7407282590866089, + "num_tokens": 373644582.0, + "step": 14658 + }, + { + "epoch": 1.6098177026136613, + "grad_norm": 2.3298702239990234, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7233877182006836, + "num_tokens": 373667648.0, + "step": 14659 + }, + { + "epoch": 1.6099275203162748, + "grad_norm": 2.268439531326294, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7191055417060852, + "num_tokens": 373692286.0, + "step": 14660 + }, + { + "epoch": 1.6100373380188886, + "grad_norm": 1.9656634330749512, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7314155697822571, + "num_tokens": 373723962.0, + "step": 14661 + }, + { + "epoch": 1.6101471557215024, + "grad_norm": 2.3114025592803955, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7442255020141602, + "num_tokens": 373749233.0, + "step": 14662 + }, + { + "epoch": 1.6102569734241161, + "grad_norm": 2.1069939136505127, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7067338824272156, + "num_tokens": 373780065.0, + "step": 14663 + }, + { + "epoch": 1.6103667911267296, + "grad_norm": 2.1189661026000977, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7029346823692322, + "num_tokens": 373809330.0, + "step": 14664 + }, + { + "epoch": 1.6104766088293432, + "grad_norm": 2.7403671741485596, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7222052216529846, + "num_tokens": 373828792.0, + "step": 14665 + }, + { + "epoch": 1.610586426531957, + "grad_norm": 2.766235589981079, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.716558575630188, + "num_tokens": 373847100.0, + "step": 14666 + }, + { + "epoch": 1.6106962442345707, + "grad_norm": 2.1363022327423096, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7416707873344421, + "num_tokens": 373875363.0, + "step": 14667 + }, + { + "epoch": 1.6108060619371842, + "grad_norm": 2.4873969554901123, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7205837965011597, + "num_tokens": 373901733.0, + "step": 14668 + }, + { + "epoch": 1.610915879639798, + "grad_norm": 2.1060523986816406, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7011334300041199, + "num_tokens": 373930920.0, + "step": 14669 + }, + { + "epoch": 1.6110256973424115, + "grad_norm": 2.147819995880127, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7133199572563171, + "num_tokens": 373958239.0, + "step": 14670 + }, + { + "epoch": 1.6111355150450253, + "grad_norm": 2.282735824584961, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7085590958595276, + "num_tokens": 373984597.0, + "step": 14671 + }, + { + "epoch": 1.611245332747639, + "grad_norm": 1.9570279121398926, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.6972169876098633, + "num_tokens": 374016001.0, + "step": 14672 + }, + { + "epoch": 1.6113551504502526, + "grad_norm": 2.2699925899505615, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.6986194849014282, + "num_tokens": 374043137.0, + "step": 14673 + }, + { + "epoch": 1.611464968152866, + "grad_norm": 2.166559934616089, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7113645076751709, + "num_tokens": 374071373.0, + "step": 14674 + }, + { + "epoch": 1.6115747858554799, + "grad_norm": 2.2404279708862305, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7301993370056152, + "num_tokens": 374097656.0, + "step": 14675 + }, + { + "epoch": 1.6116846035580936, + "grad_norm": 2.4529614448547363, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7196148037910461, + "num_tokens": 374118997.0, + "step": 14676 + }, + { + "epoch": 1.6117944212607074, + "grad_norm": 2.0132665634155273, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7122476696968079, + "num_tokens": 374149393.0, + "step": 14677 + }, + { + "epoch": 1.611904238963321, + "grad_norm": 2.5523366928100586, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7141029834747314, + "num_tokens": 374170516.0, + "step": 14678 + }, + { + "epoch": 1.6120140566659344, + "grad_norm": 2.269667148590088, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7330018281936646, + "num_tokens": 374195495.0, + "step": 14679 + }, + { + "epoch": 1.6121238743685482, + "grad_norm": 2.6247735023498535, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7459732890129089, + "num_tokens": 374214297.0, + "step": 14680 + }, + { + "epoch": 1.612233692071162, + "grad_norm": 2.474334955215454, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7027724981307983, + "num_tokens": 374235358.0, + "step": 14681 + }, + { + "epoch": 1.6123435097737755, + "grad_norm": 2.5339319705963135, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7356775999069214, + "num_tokens": 374255212.0, + "step": 14682 + }, + { + "epoch": 1.612453327476389, + "grad_norm": 2.206674814224243, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7317125797271729, + "num_tokens": 374284878.0, + "step": 14683 + }, + { + "epoch": 1.6125631451790028, + "grad_norm": 2.264117479324341, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7178601026535034, + "num_tokens": 374311635.0, + "step": 14684 + }, + { + "epoch": 1.6126729628816165, + "grad_norm": 2.2497756481170654, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7325975894927979, + "num_tokens": 374337781.0, + "step": 14685 + }, + { + "epoch": 1.6127827805842303, + "grad_norm": 2.335785150527954, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7289692759513855, + "num_tokens": 374362381.0, + "step": 14686 + }, + { + "epoch": 1.6128925982868438, + "grad_norm": 2.110201835632324, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7183481454849243, + "num_tokens": 374392188.0, + "step": 14687 + }, + { + "epoch": 1.6130024159894574, + "grad_norm": 2.345635175704956, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7181974053382874, + "num_tokens": 374416229.0, + "step": 14688 + }, + { + "epoch": 1.6131122336920711, + "grad_norm": 2.475585460662842, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7250074744224548, + "num_tokens": 374437956.0, + "step": 14689 + }, + { + "epoch": 1.6132220513946849, + "grad_norm": 2.5021555423736572, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7151257395744324, + "num_tokens": 374459117.0, + "step": 14690 + }, + { + "epoch": 1.6133318690972986, + "grad_norm": 2.4609837532043457, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7269468307495117, + "num_tokens": 374481100.0, + "step": 14691 + }, + { + "epoch": 1.6134416867999122, + "grad_norm": 2.247375726699829, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7210276126861572, + "num_tokens": 374507322.0, + "step": 14692 + }, + { + "epoch": 1.6135515045025257, + "grad_norm": 2.308950424194336, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.718438982963562, + "num_tokens": 374531761.0, + "step": 14693 + }, + { + "epoch": 1.6136613222051395, + "grad_norm": 2.312391519546509, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7174948453903198, + "num_tokens": 374557123.0, + "step": 14694 + }, + { + "epoch": 1.6137711399077532, + "grad_norm": 2.7834644317626953, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7364118695259094, + "num_tokens": 374575163.0, + "step": 14695 + }, + { + "epoch": 1.6138809576103668, + "grad_norm": 2.3879473209381104, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7231995463371277, + "num_tokens": 374597037.0, + "step": 14696 + }, + { + "epoch": 1.6139907753129803, + "grad_norm": 2.1704213619232178, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.719297468662262, + "num_tokens": 374624068.0, + "step": 14697 + }, + { + "epoch": 1.614100593015594, + "grad_norm": 2.253063201904297, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7075644135475159, + "num_tokens": 374649165.0, + "step": 14698 + }, + { + "epoch": 1.6142104107182078, + "grad_norm": 2.7670910358428955, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7285609245300293, + "num_tokens": 374666388.0, + "step": 14699 + }, + { + "epoch": 1.6143202284208216, + "grad_norm": 2.2825205326080322, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7203015685081482, + "num_tokens": 374693513.0, + "step": 14700 + }, + { + "epoch": 1.614430046123435, + "grad_norm": 2.286506414413452, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7101812362670898, + "num_tokens": 374719587.0, + "step": 14701 + }, + { + "epoch": 1.6145398638260486, + "grad_norm": 2.3136186599731445, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7191760540008545, + "num_tokens": 374745308.0, + "step": 14702 + }, + { + "epoch": 1.6146496815286624, + "grad_norm": 2.332785129547119, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7193269729614258, + "num_tokens": 374769622.0, + "step": 14703 + }, + { + "epoch": 1.6147594992312762, + "grad_norm": 2.4250576496124268, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.721662163734436, + "num_tokens": 374790223.0, + "step": 14704 + }, + { + "epoch": 1.61486931693389, + "grad_norm": 2.290250778198242, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7099872827529907, + "num_tokens": 374816985.0, + "step": 14705 + }, + { + "epoch": 1.6149791346365034, + "grad_norm": 2.1613192558288574, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7234878540039062, + "num_tokens": 374843440.0, + "step": 14706 + }, + { + "epoch": 1.615088952339117, + "grad_norm": 2.3637454509735107, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7232992053031921, + "num_tokens": 374865673.0, + "step": 14707 + }, + { + "epoch": 1.6151987700417307, + "grad_norm": 2.329782009124756, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7119663953781128, + "num_tokens": 374890404.0, + "step": 14708 + }, + { + "epoch": 1.6153085877443445, + "grad_norm": 2.0904769897460938, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.735935389995575, + "num_tokens": 374919246.0, + "step": 14709 + }, + { + "epoch": 1.615418405446958, + "grad_norm": 1.9146862030029297, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7002511024475098, + "num_tokens": 374952809.0, + "step": 14710 + }, + { + "epoch": 1.6155282231495716, + "grad_norm": 2.011063814163208, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.6941964030265808, + "num_tokens": 374984513.0, + "step": 14711 + }, + { + "epoch": 1.6156380408521853, + "grad_norm": 2.2189197540283203, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7181403636932373, + "num_tokens": 375010116.0, + "step": 14712 + }, + { + "epoch": 1.615747858554799, + "grad_norm": 2.201927900314331, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7095417976379395, + "num_tokens": 375035278.0, + "step": 14713 + }, + { + "epoch": 1.6158576762574128, + "grad_norm": 2.366664409637451, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7250341773033142, + "num_tokens": 375060455.0, + "step": 14714 + }, + { + "epoch": 1.6159674939600264, + "grad_norm": 2.8694255352020264, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7444988489151001, + "num_tokens": 375077758.0, + "step": 14715 + }, + { + "epoch": 1.61607731166264, + "grad_norm": 2.3910818099975586, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7160691022872925, + "num_tokens": 375099862.0, + "step": 14716 + }, + { + "epoch": 1.6161871293652537, + "grad_norm": 2.3426623344421387, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7288225889205933, + "num_tokens": 375122307.0, + "step": 14717 + }, + { + "epoch": 1.6162969470678674, + "grad_norm": 2.3193442821502686, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7360252737998962, + "num_tokens": 375146259.0, + "step": 14718 + }, + { + "epoch": 1.616406764770481, + "grad_norm": 2.1339340209960938, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.713951587677002, + "num_tokens": 375174228.0, + "step": 14719 + }, + { + "epoch": 1.6165165824730947, + "grad_norm": 2.5350937843322754, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7244824767112732, + "num_tokens": 375196965.0, + "step": 14720 + }, + { + "epoch": 1.6166264001757082, + "grad_norm": 2.3677690029144287, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7093522548675537, + "num_tokens": 375221734.0, + "step": 14721 + }, + { + "epoch": 1.616736217878322, + "grad_norm": 2.4627764225006104, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7319799661636353, + "num_tokens": 375243378.0, + "step": 14722 + }, + { + "epoch": 1.6168460355809358, + "grad_norm": 2.1222081184387207, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7286361455917358, + "num_tokens": 375271839.0, + "step": 14723 + }, + { + "epoch": 1.6169558532835493, + "grad_norm": 2.0729904174804688, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7090739607810974, + "num_tokens": 375302749.0, + "step": 14724 + }, + { + "epoch": 1.6170656709861628, + "grad_norm": 2.221916437149048, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6873378157615662, + "num_tokens": 375328773.0, + "step": 14725 + }, + { + "epoch": 1.6171754886887766, + "grad_norm": 1.9193499088287354, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7171199917793274, + "num_tokens": 375361297.0, + "step": 14726 + }, + { + "epoch": 1.6172853063913903, + "grad_norm": 2.3039309978485107, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7180876135826111, + "num_tokens": 375386138.0, + "step": 14727 + }, + { + "epoch": 1.617395124094004, + "grad_norm": 2.4350340366363525, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7084720134735107, + "num_tokens": 375410777.0, + "step": 14728 + }, + { + "epoch": 1.6175049417966176, + "grad_norm": 1.9633917808532715, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7141950726509094, + "num_tokens": 375442959.0, + "step": 14729 + }, + { + "epoch": 1.6176147594992312, + "grad_norm": 2.257115602493286, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7049333453178406, + "num_tokens": 375471831.0, + "step": 14730 + }, + { + "epoch": 1.617724577201845, + "grad_norm": 2.1594789028167725, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7005881071090698, + "num_tokens": 375500535.0, + "step": 14731 + }, + { + "epoch": 1.6178343949044587, + "grad_norm": 2.323988199234009, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7326209545135498, + "num_tokens": 375524247.0, + "step": 14732 + }, + { + "epoch": 1.6179442126070722, + "grad_norm": 1.8649743795394897, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.697691798210144, + "num_tokens": 375562521.0, + "step": 14733 + }, + { + "epoch": 1.618054030309686, + "grad_norm": 2.504945755004883, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7111565470695496, + "num_tokens": 375586614.0, + "step": 14734 + }, + { + "epoch": 1.6181638480122995, + "grad_norm": 2.311176061630249, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7227845191955566, + "num_tokens": 375610219.0, + "step": 14735 + }, + { + "epoch": 1.6182736657149133, + "grad_norm": 2.6148791313171387, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.7455185651779175, + "num_tokens": 375630125.0, + "step": 14736 + }, + { + "epoch": 1.618383483417527, + "grad_norm": 2.5595626831054688, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.7216430306434631, + "num_tokens": 375655877.0, + "step": 14737 + }, + { + "epoch": 1.6184933011201406, + "grad_norm": 2.3232645988464355, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7193411588668823, + "num_tokens": 375682562.0, + "step": 14738 + }, + { + "epoch": 1.618603118822754, + "grad_norm": 2.2206342220306396, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7015269994735718, + "num_tokens": 375708902.0, + "step": 14739 + }, + { + "epoch": 1.6187129365253679, + "grad_norm": 2.472395896911621, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7551161050796509, + "num_tokens": 375729646.0, + "step": 14740 + }, + { + "epoch": 1.6188227542279816, + "grad_norm": 2.370910167694092, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7233600616455078, + "num_tokens": 375752384.0, + "step": 14741 + }, + { + "epoch": 1.6189325719305954, + "grad_norm": 2.102299213409424, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7226324081420898, + "num_tokens": 375782239.0, + "step": 14742 + }, + { + "epoch": 1.619042389633209, + "grad_norm": 1.955113172531128, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7241365909576416, + "num_tokens": 375814498.0, + "step": 14743 + }, + { + "epoch": 1.6191522073358224, + "grad_norm": 2.5211355686187744, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7450578212738037, + "num_tokens": 375834768.0, + "step": 14744 + }, + { + "epoch": 1.6192620250384362, + "grad_norm": 2.199416399002075, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7300996780395508, + "num_tokens": 375860382.0, + "step": 14745 + }, + { + "epoch": 1.61937184274105, + "grad_norm": 2.1994831562042236, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7156921625137329, + "num_tokens": 375885799.0, + "step": 14746 + }, + { + "epoch": 1.6194816604436635, + "grad_norm": 2.454650402069092, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7240978479385376, + "num_tokens": 375908489.0, + "step": 14747 + }, + { + "epoch": 1.619591478146277, + "grad_norm": 2.1125435829162598, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.6960657238960266, + "num_tokens": 375936625.0, + "step": 14748 + }, + { + "epoch": 1.6197012958488908, + "grad_norm": 2.0838966369628906, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7347935438156128, + "num_tokens": 375965762.0, + "step": 14749 + }, + { + "epoch": 1.6198111135515045, + "grad_norm": 2.3090994358062744, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.735054612159729, + "num_tokens": 375989673.0, + "step": 14750 + }, + { + "epoch": 1.6199209312541183, + "grad_norm": 2.2331228256225586, + "learning_rate": 1e-06, + "loss": 0.784, + "mean_token_accuracy": 0.752257764339447, + "num_tokens": 376011700.0, + "step": 14751 + }, + { + "epoch": 1.6200307489567318, + "grad_norm": 2.0775206089019775, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7249168157577515, + "num_tokens": 376039457.0, + "step": 14752 + }, + { + "epoch": 1.6201405666593454, + "grad_norm": 2.3080554008483887, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7346036434173584, + "num_tokens": 376062644.0, + "step": 14753 + }, + { + "epoch": 1.6202503843619591, + "grad_norm": 2.4192955493927, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7380601763725281, + "num_tokens": 376083823.0, + "step": 14754 + }, + { + "epoch": 1.6203602020645729, + "grad_norm": 2.162191390991211, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7245687246322632, + "num_tokens": 376112417.0, + "step": 14755 + }, + { + "epoch": 1.6204700197671866, + "grad_norm": 2.2924680709838867, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.717514157295227, + "num_tokens": 376137114.0, + "step": 14756 + }, + { + "epoch": 1.6205798374698002, + "grad_norm": 2.1031835079193115, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7139161825180054, + "num_tokens": 376165836.0, + "step": 14757 + }, + { + "epoch": 1.6206896551724137, + "grad_norm": 2.1204047203063965, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7147113084793091, + "num_tokens": 376195707.0, + "step": 14758 + }, + { + "epoch": 1.6207994728750275, + "grad_norm": 2.5239741802215576, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7233842015266418, + "num_tokens": 376216388.0, + "step": 14759 + }, + { + "epoch": 1.6209092905776412, + "grad_norm": 2.11175799369812, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7354847192764282, + "num_tokens": 376245433.0, + "step": 14760 + }, + { + "epoch": 1.6210191082802548, + "grad_norm": 2.3661704063415527, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7161308526992798, + "num_tokens": 376270139.0, + "step": 14761 + }, + { + "epoch": 1.6211289259828683, + "grad_norm": 2.19311785697937, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7226470708847046, + "num_tokens": 376297530.0, + "step": 14762 + }, + { + "epoch": 1.621238743685482, + "grad_norm": 2.0460331439971924, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7060738801956177, + "num_tokens": 376330166.0, + "step": 14763 + }, + { + "epoch": 1.6213485613880958, + "grad_norm": 2.2213289737701416, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7370704412460327, + "num_tokens": 376355414.0, + "step": 14764 + }, + { + "epoch": 1.6214583790907096, + "grad_norm": 2.1289024353027344, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7235469222068787, + "num_tokens": 376382710.0, + "step": 14765 + }, + { + "epoch": 1.621568196793323, + "grad_norm": 2.4645297527313232, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7118054628372192, + "num_tokens": 376405463.0, + "step": 14766 + }, + { + "epoch": 1.6216780144959366, + "grad_norm": 2.412705898284912, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.714529275894165, + "num_tokens": 376428769.0, + "step": 14767 + }, + { + "epoch": 1.6217878321985504, + "grad_norm": 2.422542095184326, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7390389442443848, + "num_tokens": 376451979.0, + "step": 14768 + }, + { + "epoch": 1.6218976499011641, + "grad_norm": 2.2540488243103027, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7148064374923706, + "num_tokens": 376478241.0, + "step": 14769 + }, + { + "epoch": 1.6220074676037777, + "grad_norm": 2.460461378097534, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7270931601524353, + "num_tokens": 376501141.0, + "step": 14770 + }, + { + "epoch": 1.6221172853063914, + "grad_norm": 2.0377955436706543, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6967207193374634, + "num_tokens": 376532273.0, + "step": 14771 + }, + { + "epoch": 1.622227103009005, + "grad_norm": 2.2106130123138428, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6914809942245483, + "num_tokens": 376561256.0, + "step": 14772 + }, + { + "epoch": 1.6223369207116187, + "grad_norm": 2.6442556381225586, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7100625038146973, + "num_tokens": 376581919.0, + "step": 14773 + }, + { + "epoch": 1.6224467384142325, + "grad_norm": 2.2026171684265137, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7105506062507629, + "num_tokens": 376608471.0, + "step": 14774 + }, + { + "epoch": 1.622556556116846, + "grad_norm": 2.1042134761810303, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7384997010231018, + "num_tokens": 376635143.0, + "step": 14775 + }, + { + "epoch": 1.6226663738194596, + "grad_norm": 2.351454257965088, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7340223789215088, + "num_tokens": 376658098.0, + "step": 14776 + }, + { + "epoch": 1.6227761915220733, + "grad_norm": 2.329012632369995, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7274897694587708, + "num_tokens": 376682399.0, + "step": 14777 + }, + { + "epoch": 1.622886009224687, + "grad_norm": 2.284235954284668, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7009062767028809, + "num_tokens": 376708986.0, + "step": 14778 + }, + { + "epoch": 1.6229958269273008, + "grad_norm": 2.0341835021972656, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7165464162826538, + "num_tokens": 376739440.0, + "step": 14779 + }, + { + "epoch": 1.6231056446299144, + "grad_norm": 2.2254881858825684, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7139890193939209, + "num_tokens": 376766244.0, + "step": 14780 + }, + { + "epoch": 1.623215462332528, + "grad_norm": 2.436882972717285, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7129968404769897, + "num_tokens": 376790095.0, + "step": 14781 + }, + { + "epoch": 1.6233252800351416, + "grad_norm": 2.0756022930145264, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7241346836090088, + "num_tokens": 376818805.0, + "step": 14782 + }, + { + "epoch": 1.6234350977377554, + "grad_norm": 2.2399284839630127, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7150598764419556, + "num_tokens": 376843050.0, + "step": 14783 + }, + { + "epoch": 1.623544915440369, + "grad_norm": 2.103858232498169, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7276133894920349, + "num_tokens": 376872109.0, + "step": 14784 + }, + { + "epoch": 1.6236547331429827, + "grad_norm": 2.494506597518921, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7262029647827148, + "num_tokens": 376892929.0, + "step": 14785 + }, + { + "epoch": 1.6237645508455962, + "grad_norm": 1.9333446025848389, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.715351402759552, + "num_tokens": 376924798.0, + "step": 14786 + }, + { + "epoch": 1.62387436854821, + "grad_norm": 2.0835702419281006, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7276884317398071, + "num_tokens": 376952694.0, + "step": 14787 + }, + { + "epoch": 1.6239841862508237, + "grad_norm": 2.299252986907959, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7331206798553467, + "num_tokens": 376976581.0, + "step": 14788 + }, + { + "epoch": 1.6240940039534373, + "grad_norm": 2.210042715072632, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.711219072341919, + "num_tokens": 377004056.0, + "step": 14789 + }, + { + "epoch": 1.6242038216560508, + "grad_norm": 2.279935121536255, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7095352411270142, + "num_tokens": 377031491.0, + "step": 14790 + }, + { + "epoch": 1.6243136393586646, + "grad_norm": 1.9334518909454346, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7001661062240601, + "num_tokens": 377065976.0, + "step": 14791 + }, + { + "epoch": 1.6244234570612783, + "grad_norm": 2.0767555236816406, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7017483115196228, + "num_tokens": 377097210.0, + "step": 14792 + }, + { + "epoch": 1.624533274763892, + "grad_norm": 2.0042507648468018, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7150869369506836, + "num_tokens": 377127839.0, + "step": 14793 + }, + { + "epoch": 1.6246430924665056, + "grad_norm": 2.2596652507781982, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7352529168128967, + "num_tokens": 377152886.0, + "step": 14794 + }, + { + "epoch": 1.6247529101691192, + "grad_norm": 2.311711072921753, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.723084568977356, + "num_tokens": 377176915.0, + "step": 14795 + }, + { + "epoch": 1.624862727871733, + "grad_norm": 2.1201767921447754, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.6973470449447632, + "num_tokens": 377207172.0, + "step": 14796 + }, + { + "epoch": 1.6249725455743467, + "grad_norm": 2.602461814880371, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7534210681915283, + "num_tokens": 377225414.0, + "step": 14797 + }, + { + "epoch": 1.6250823632769602, + "grad_norm": 2.238980293273926, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7054111957550049, + "num_tokens": 377251184.0, + "step": 14798 + }, + { + "epoch": 1.6251921809795737, + "grad_norm": 2.3166141510009766, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7351452708244324, + "num_tokens": 377274843.0, + "step": 14799 + }, + { + "epoch": 1.6253019986821875, + "grad_norm": 2.266153335571289, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7230401039123535, + "num_tokens": 377301585.0, + "step": 14800 + }, + { + "epoch": 1.6254118163848013, + "grad_norm": 2.084123373031616, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7113323211669922, + "num_tokens": 377332399.0, + "step": 14801 + }, + { + "epoch": 1.625521634087415, + "grad_norm": 2.1683316230773926, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7268664836883545, + "num_tokens": 377357620.0, + "step": 14802 + }, + { + "epoch": 1.6256314517900285, + "grad_norm": 2.3603668212890625, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7132810950279236, + "num_tokens": 377382154.0, + "step": 14803 + }, + { + "epoch": 1.625741269492642, + "grad_norm": 2.363527297973633, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7301898002624512, + "num_tokens": 377404753.0, + "step": 14804 + }, + { + "epoch": 1.6258510871952558, + "grad_norm": 2.2197892665863037, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7440747618675232, + "num_tokens": 377429950.0, + "step": 14805 + }, + { + "epoch": 1.6259609048978696, + "grad_norm": 1.979405403137207, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7050232887268066, + "num_tokens": 377461359.0, + "step": 14806 + }, + { + "epoch": 1.6260707226004834, + "grad_norm": 2.028005361557007, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7199064493179321, + "num_tokens": 377491774.0, + "step": 14807 + }, + { + "epoch": 1.6261805403030969, + "grad_norm": 2.2958598136901855, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7188373804092407, + "num_tokens": 377518446.0, + "step": 14808 + }, + { + "epoch": 1.6262903580057104, + "grad_norm": 2.1944334506988525, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.718565821647644, + "num_tokens": 377544730.0, + "step": 14809 + }, + { + "epoch": 1.6264001757083242, + "grad_norm": 2.2339425086975098, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.7429413795471191, + "num_tokens": 377568675.0, + "step": 14810 + }, + { + "epoch": 1.626509993410938, + "grad_norm": 2.677504301071167, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7142191529273987, + "num_tokens": 377588027.0, + "step": 14811 + }, + { + "epoch": 1.6266198111135515, + "grad_norm": 2.4734396934509277, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7153866291046143, + "num_tokens": 377609922.0, + "step": 14812 + }, + { + "epoch": 1.626729628816165, + "grad_norm": 2.3421270847320557, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7434171438217163, + "num_tokens": 377634726.0, + "step": 14813 + }, + { + "epoch": 1.6268394465187788, + "grad_norm": 2.1075944900512695, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7218201160430908, + "num_tokens": 377665250.0, + "step": 14814 + }, + { + "epoch": 1.6269492642213925, + "grad_norm": 2.3265953063964844, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.719955325126648, + "num_tokens": 377689030.0, + "step": 14815 + }, + { + "epoch": 1.6270590819240063, + "grad_norm": 2.126166343688965, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7244314551353455, + "num_tokens": 377715721.0, + "step": 14816 + }, + { + "epoch": 1.6271688996266198, + "grad_norm": 2.0453310012817383, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7188241481781006, + "num_tokens": 377746949.0, + "step": 14817 + }, + { + "epoch": 1.6272787173292333, + "grad_norm": 2.90578031539917, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7371609210968018, + "num_tokens": 377762643.0, + "step": 14818 + }, + { + "epoch": 1.627388535031847, + "grad_norm": 2.297027111053467, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7114709615707397, + "num_tokens": 377788195.0, + "step": 14819 + }, + { + "epoch": 1.6274983527344609, + "grad_norm": 2.013300657272339, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7185591459274292, + "num_tokens": 377817149.0, + "step": 14820 + }, + { + "epoch": 1.6276081704370746, + "grad_norm": 2.0760622024536133, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7212716937065125, + "num_tokens": 377845407.0, + "step": 14821 + }, + { + "epoch": 1.6277179881396882, + "grad_norm": 2.1670002937316895, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7396719455718994, + "num_tokens": 377870797.0, + "step": 14822 + }, + { + "epoch": 1.6278278058423017, + "grad_norm": 2.37487530708313, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7384436130523682, + "num_tokens": 377893705.0, + "step": 14823 + }, + { + "epoch": 1.6279376235449154, + "grad_norm": 2.3874926567077637, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7311696410179138, + "num_tokens": 377917651.0, + "step": 14824 + }, + { + "epoch": 1.6280474412475292, + "grad_norm": 2.1065056324005127, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7161718010902405, + "num_tokens": 377947527.0, + "step": 14825 + }, + { + "epoch": 1.6281572589501427, + "grad_norm": 2.3087410926818848, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7046732902526855, + "num_tokens": 377973044.0, + "step": 14826 + }, + { + "epoch": 1.6282670766527563, + "grad_norm": 2.0802066326141357, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7153461575508118, + "num_tokens": 378002782.0, + "step": 14827 + }, + { + "epoch": 1.62837689435537, + "grad_norm": 2.2091305255889893, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7173953652381897, + "num_tokens": 378029585.0, + "step": 14828 + }, + { + "epoch": 1.6284867120579838, + "grad_norm": 2.3083651065826416, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7018738985061646, + "num_tokens": 378055879.0, + "step": 14829 + }, + { + "epoch": 1.6285965297605975, + "grad_norm": 2.0958659648895264, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.703154444694519, + "num_tokens": 378084818.0, + "step": 14830 + }, + { + "epoch": 1.628706347463211, + "grad_norm": 2.156050443649292, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7223355770111084, + "num_tokens": 378112351.0, + "step": 14831 + }, + { + "epoch": 1.6288161651658246, + "grad_norm": 2.190817356109619, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7072172164916992, + "num_tokens": 378138536.0, + "step": 14832 + }, + { + "epoch": 1.6289259828684384, + "grad_norm": 2.3965342044830322, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7271856069564819, + "num_tokens": 378162493.0, + "step": 14833 + }, + { + "epoch": 1.6290358005710521, + "grad_norm": 2.370293617248535, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7145947217941284, + "num_tokens": 378184960.0, + "step": 14834 + }, + { + "epoch": 1.6291456182736657, + "grad_norm": 2.4278011322021484, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.738858699798584, + "num_tokens": 378207654.0, + "step": 14835 + }, + { + "epoch": 1.6292554359762794, + "grad_norm": 2.1294455528259277, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7013558149337769, + "num_tokens": 378235695.0, + "step": 14836 + }, + { + "epoch": 1.629365253678893, + "grad_norm": 1.8799978494644165, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.6899346113204956, + "num_tokens": 378272010.0, + "step": 14837 + }, + { + "epoch": 1.6294750713815067, + "grad_norm": 2.7332193851470947, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7338153719902039, + "num_tokens": 378289353.0, + "step": 14838 + }, + { + "epoch": 1.6295848890841205, + "grad_norm": 2.41508412361145, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.6965212821960449, + "num_tokens": 378314055.0, + "step": 14839 + }, + { + "epoch": 1.629694706786734, + "grad_norm": 1.9969127178192139, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.698707103729248, + "num_tokens": 378345642.0, + "step": 14840 + }, + { + "epoch": 1.6298045244893475, + "grad_norm": 2.447319984436035, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7366984486579895, + "num_tokens": 378367106.0, + "step": 14841 + }, + { + "epoch": 1.6299143421919613, + "grad_norm": 2.310084342956543, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7238731980323792, + "num_tokens": 378390644.0, + "step": 14842 + }, + { + "epoch": 1.630024159894575, + "grad_norm": 2.2538297176361084, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.6997373104095459, + "num_tokens": 378416745.0, + "step": 14843 + }, + { + "epoch": 1.6301339775971888, + "grad_norm": 2.2625017166137695, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.723006010055542, + "num_tokens": 378441796.0, + "step": 14844 + }, + { + "epoch": 1.6302437952998023, + "grad_norm": 2.1735825538635254, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.715754508972168, + "num_tokens": 378469617.0, + "step": 14845 + }, + { + "epoch": 1.6303536130024159, + "grad_norm": 2.2651233673095703, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7155872583389282, + "num_tokens": 378498829.0, + "step": 14846 + }, + { + "epoch": 1.6304634307050296, + "grad_norm": 2.2124593257904053, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7339046001434326, + "num_tokens": 378526260.0, + "step": 14847 + }, + { + "epoch": 1.6305732484076434, + "grad_norm": 2.262075185775757, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.711087703704834, + "num_tokens": 378551607.0, + "step": 14848 + }, + { + "epoch": 1.630683066110257, + "grad_norm": 2.3777854442596436, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7430591583251953, + "num_tokens": 378573845.0, + "step": 14849 + }, + { + "epoch": 1.6307928838128707, + "grad_norm": 2.3980507850646973, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7300410866737366, + "num_tokens": 378599073.0, + "step": 14850 + }, + { + "epoch": 1.6309027015154842, + "grad_norm": 2.4135520458221436, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7337592244148254, + "num_tokens": 378622705.0, + "step": 14851 + }, + { + "epoch": 1.631012519218098, + "grad_norm": 2.204975128173828, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7182832360267639, + "num_tokens": 378649811.0, + "step": 14852 + }, + { + "epoch": 1.6311223369207117, + "grad_norm": 2.3061885833740234, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.7536267042160034, + "num_tokens": 378671601.0, + "step": 14853 + }, + { + "epoch": 1.6312321546233253, + "grad_norm": 2.6086909770965576, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7249733209609985, + "num_tokens": 378691132.0, + "step": 14854 + }, + { + "epoch": 1.6313419723259388, + "grad_norm": 2.0331404209136963, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7314856052398682, + "num_tokens": 378721120.0, + "step": 14855 + }, + { + "epoch": 1.6314517900285526, + "grad_norm": 2.0433108806610107, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.7185274362564087, + "num_tokens": 378752132.0, + "step": 14856 + }, + { + "epoch": 1.6315616077311663, + "grad_norm": 2.0592637062072754, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7656288743019104, + "num_tokens": 378778538.0, + "step": 14857 + }, + { + "epoch": 1.63167142543378, + "grad_norm": 2.5691874027252197, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7038118839263916, + "num_tokens": 378800856.0, + "step": 14858 + }, + { + "epoch": 1.6317812431363936, + "grad_norm": 2.3074538707733154, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7276373505592346, + "num_tokens": 378825127.0, + "step": 14859 + }, + { + "epoch": 1.6318910608390071, + "grad_norm": 2.2445600032806396, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7087167501449585, + "num_tokens": 378853726.0, + "step": 14860 + }, + { + "epoch": 1.632000878541621, + "grad_norm": 2.4026763439178467, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7129664421081543, + "num_tokens": 378875307.0, + "step": 14861 + }, + { + "epoch": 1.6321106962442347, + "grad_norm": 2.278526782989502, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7455387115478516, + "num_tokens": 378899257.0, + "step": 14862 + }, + { + "epoch": 1.6322205139468482, + "grad_norm": 2.1080198287963867, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7277695536613464, + "num_tokens": 378928142.0, + "step": 14863 + }, + { + "epoch": 1.6323303316494617, + "grad_norm": 2.1268608570098877, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7119991779327393, + "num_tokens": 378955398.0, + "step": 14864 + }, + { + "epoch": 1.6324401493520755, + "grad_norm": 2.399876356124878, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7325586080551147, + "num_tokens": 378976960.0, + "step": 14865 + }, + { + "epoch": 1.6325499670546892, + "grad_norm": 2.2506744861602783, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.6987900733947754, + "num_tokens": 379002774.0, + "step": 14866 + }, + { + "epoch": 1.632659784757303, + "grad_norm": 2.2921717166900635, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7245522737503052, + "num_tokens": 379028098.0, + "step": 14867 + }, + { + "epoch": 1.6327696024599165, + "grad_norm": 2.457613945007324, + "learning_rate": 1e-06, + "loss": 0.8118, + "mean_token_accuracy": 0.7404265999794006, + "num_tokens": 379049814.0, + "step": 14868 + }, + { + "epoch": 1.63287942016253, + "grad_norm": 2.082756280899048, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.6986385583877563, + "num_tokens": 379080209.0, + "step": 14869 + }, + { + "epoch": 1.6329892378651438, + "grad_norm": 2.1446993350982666, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7259528040885925, + "num_tokens": 379106329.0, + "step": 14870 + }, + { + "epoch": 1.6330990555677576, + "grad_norm": 2.434398889541626, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.69225013256073, + "num_tokens": 379132003.0, + "step": 14871 + }, + { + "epoch": 1.6332088732703713, + "grad_norm": 2.407834768295288, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7240078449249268, + "num_tokens": 379155029.0, + "step": 14872 + }, + { + "epoch": 1.6333186909729849, + "grad_norm": 2.3793206214904785, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7071893811225891, + "num_tokens": 379181245.0, + "step": 14873 + }, + { + "epoch": 1.6334285086755984, + "grad_norm": 2.3385393619537354, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7322264909744263, + "num_tokens": 379203035.0, + "step": 14874 + }, + { + "epoch": 1.6335383263782122, + "grad_norm": 2.3094100952148438, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7315924763679504, + "num_tokens": 379228042.0, + "step": 14875 + }, + { + "epoch": 1.633648144080826, + "grad_norm": 2.1685690879821777, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7570456266403198, + "num_tokens": 379252755.0, + "step": 14876 + }, + { + "epoch": 1.6337579617834395, + "grad_norm": 2.003777027130127, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7099493145942688, + "num_tokens": 379285705.0, + "step": 14877 + }, + { + "epoch": 1.633867779486053, + "grad_norm": 2.516939401626587, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.738276481628418, + "num_tokens": 379306729.0, + "step": 14878 + }, + { + "epoch": 1.6339775971886668, + "grad_norm": 2.4832801818847656, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7223495244979858, + "num_tokens": 379330449.0, + "step": 14879 + }, + { + "epoch": 1.6340874148912805, + "grad_norm": 2.2289540767669678, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7073636651039124, + "num_tokens": 379356971.0, + "step": 14880 + }, + { + "epoch": 1.6341972325938943, + "grad_norm": 2.591357707977295, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7048370242118835, + "num_tokens": 379378491.0, + "step": 14881 + }, + { + "epoch": 1.6343070502965078, + "grad_norm": 2.5591087341308594, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7327210903167725, + "num_tokens": 379399492.0, + "step": 14882 + }, + { + "epoch": 1.6344168679991213, + "grad_norm": 2.2936668395996094, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7244004607200623, + "num_tokens": 379425373.0, + "step": 14883 + }, + { + "epoch": 1.634526685701735, + "grad_norm": 2.0654399394989014, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7304624319076538, + "num_tokens": 379454742.0, + "step": 14884 + }, + { + "epoch": 1.6346365034043489, + "grad_norm": 2.3610453605651855, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.728684663772583, + "num_tokens": 379478758.0, + "step": 14885 + }, + { + "epoch": 1.6347463211069626, + "grad_norm": 2.004948616027832, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7318606376647949, + "num_tokens": 379509598.0, + "step": 14886 + }, + { + "epoch": 1.6348561388095761, + "grad_norm": 2.382642984390259, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7295049428939819, + "num_tokens": 379532055.0, + "step": 14887 + }, + { + "epoch": 1.6349659565121897, + "grad_norm": 2.463238000869751, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7259843349456787, + "num_tokens": 379556347.0, + "step": 14888 + }, + { + "epoch": 1.6350757742148034, + "grad_norm": 2.6888511180877686, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7339633703231812, + "num_tokens": 379577552.0, + "step": 14889 + }, + { + "epoch": 1.6351855919174172, + "grad_norm": 2.885078191757202, + "learning_rate": 1e-06, + "loss": 0.799, + "mean_token_accuracy": 0.7487198710441589, + "num_tokens": 379594081.0, + "step": 14890 + }, + { + "epoch": 1.6352954096200307, + "grad_norm": 2.399158477783203, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7011752128601074, + "num_tokens": 379619033.0, + "step": 14891 + }, + { + "epoch": 1.6354052273226443, + "grad_norm": 2.431406021118164, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7247421145439148, + "num_tokens": 379640855.0, + "step": 14892 + }, + { + "epoch": 1.635515045025258, + "grad_norm": 2.1214749813079834, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.713707447052002, + "num_tokens": 379671932.0, + "step": 14893 + }, + { + "epoch": 1.6356248627278718, + "grad_norm": 2.2727293968200684, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6875214576721191, + "num_tokens": 379699384.0, + "step": 14894 + }, + { + "epoch": 1.6357346804304855, + "grad_norm": 2.4859015941619873, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7222686409950256, + "num_tokens": 379720163.0, + "step": 14895 + }, + { + "epoch": 1.635844498133099, + "grad_norm": 2.3150393962860107, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.7361518144607544, + "num_tokens": 379743456.0, + "step": 14896 + }, + { + "epoch": 1.6359543158357126, + "grad_norm": 2.084615707397461, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7123920321464539, + "num_tokens": 379771583.0, + "step": 14897 + }, + { + "epoch": 1.6360641335383264, + "grad_norm": 2.168504238128662, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.732268214225769, + "num_tokens": 379799241.0, + "step": 14898 + }, + { + "epoch": 1.6361739512409401, + "grad_norm": 2.3093416690826416, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.707830548286438, + "num_tokens": 379823690.0, + "step": 14899 + }, + { + "epoch": 1.6362837689435537, + "grad_norm": 2.553596019744873, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7256788611412048, + "num_tokens": 379844964.0, + "step": 14900 + }, + { + "epoch": 1.6363935866461674, + "grad_norm": 2.8641574382781982, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7520223259925842, + "num_tokens": 379861934.0, + "step": 14901 + }, + { + "epoch": 1.636503404348781, + "grad_norm": 2.3500428199768066, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7435894012451172, + "num_tokens": 379885856.0, + "step": 14902 + }, + { + "epoch": 1.6366132220513947, + "grad_norm": 2.0263710021972656, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7220242023468018, + "num_tokens": 379914513.0, + "step": 14903 + }, + { + "epoch": 1.6367230397540085, + "grad_norm": 2.2518720626831055, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7190749645233154, + "num_tokens": 379940478.0, + "step": 14904 + }, + { + "epoch": 1.636832857456622, + "grad_norm": 2.2511727809906006, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7222687005996704, + "num_tokens": 379969464.0, + "step": 14905 + }, + { + "epoch": 1.6369426751592355, + "grad_norm": 2.075608730316162, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7189689874649048, + "num_tokens": 379997139.0, + "step": 14906 + }, + { + "epoch": 1.6370524928618493, + "grad_norm": 2.733309745788574, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7306115031242371, + "num_tokens": 380017563.0, + "step": 14907 + }, + { + "epoch": 1.637162310564463, + "grad_norm": 2.0161705017089844, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7315787076950073, + "num_tokens": 380045885.0, + "step": 14908 + }, + { + "epoch": 1.6372721282670768, + "grad_norm": 2.088840961456299, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7006734013557434, + "num_tokens": 380075791.0, + "step": 14909 + }, + { + "epoch": 1.6373819459696903, + "grad_norm": 2.2181270122528076, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7253717184066772, + "num_tokens": 380100212.0, + "step": 14910 + }, + { + "epoch": 1.6374917636723039, + "grad_norm": 2.233407497406006, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7001984715461731, + "num_tokens": 380126060.0, + "step": 14911 + }, + { + "epoch": 1.6376015813749176, + "grad_norm": 2.353710412979126, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7371100187301636, + "num_tokens": 380149699.0, + "step": 14912 + }, + { + "epoch": 1.6377113990775314, + "grad_norm": 2.3360538482666016, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.6984726190567017, + "num_tokens": 380173763.0, + "step": 14913 + }, + { + "epoch": 1.637821216780145, + "grad_norm": 2.1056900024414062, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7341621518135071, + "num_tokens": 380202715.0, + "step": 14914 + }, + { + "epoch": 1.6379310344827587, + "grad_norm": 2.256108045578003, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7028789520263672, + "num_tokens": 380228428.0, + "step": 14915 + }, + { + "epoch": 1.6380408521853722, + "grad_norm": 2.549333095550537, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7119810581207275, + "num_tokens": 380248467.0, + "step": 14916 + }, + { + "epoch": 1.638150669887986, + "grad_norm": 2.2716293334960938, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7166627645492554, + "num_tokens": 380274104.0, + "step": 14917 + }, + { + "epoch": 1.6382604875905997, + "grad_norm": 2.5224740505218506, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7253546714782715, + "num_tokens": 380295534.0, + "step": 14918 + }, + { + "epoch": 1.6383703052932133, + "grad_norm": 2.4749274253845215, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.740575909614563, + "num_tokens": 380316288.0, + "step": 14919 + }, + { + "epoch": 1.6384801229958268, + "grad_norm": 2.2848472595214844, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7188714742660522, + "num_tokens": 380342156.0, + "step": 14920 + }, + { + "epoch": 1.6385899406984406, + "grad_norm": 2.2155323028564453, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.707857608795166, + "num_tokens": 380369619.0, + "step": 14921 + }, + { + "epoch": 1.6386997584010543, + "grad_norm": 2.1914262771606445, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7195018529891968, + "num_tokens": 380396060.0, + "step": 14922 + }, + { + "epoch": 1.638809576103668, + "grad_norm": 2.0321147441864014, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7389000058174133, + "num_tokens": 380423643.0, + "step": 14923 + }, + { + "epoch": 1.6389193938062816, + "grad_norm": 1.9474034309387207, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.687987744808197, + "num_tokens": 380456651.0, + "step": 14924 + }, + { + "epoch": 1.6390292115088951, + "grad_norm": 2.272655725479126, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7192320227622986, + "num_tokens": 380482335.0, + "step": 14925 + }, + { + "epoch": 1.639139029211509, + "grad_norm": 2.339190721511841, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.718964159488678, + "num_tokens": 380506321.0, + "step": 14926 + }, + { + "epoch": 1.6392488469141226, + "grad_norm": 2.342979907989502, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.7502900958061218, + "num_tokens": 380527792.0, + "step": 14927 + }, + { + "epoch": 1.6393586646167362, + "grad_norm": 2.1355583667755127, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.7023452520370483, + "num_tokens": 380558443.0, + "step": 14928 + }, + { + "epoch": 1.6394684823193497, + "grad_norm": 2.0880255699157715, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7317619919776917, + "num_tokens": 380586620.0, + "step": 14929 + }, + { + "epoch": 1.6395783000219635, + "grad_norm": 2.063631057739258, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7210993766784668, + "num_tokens": 380616237.0, + "step": 14930 + }, + { + "epoch": 1.6396881177245772, + "grad_norm": 1.907723069190979, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7149688005447388, + "num_tokens": 380649424.0, + "step": 14931 + }, + { + "epoch": 1.639797935427191, + "grad_norm": 2.2317583560943604, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7362567782402039, + "num_tokens": 380672668.0, + "step": 14932 + }, + { + "epoch": 1.6399077531298045, + "grad_norm": 2.0505294799804688, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7053214907646179, + "num_tokens": 380703217.0, + "step": 14933 + }, + { + "epoch": 1.640017570832418, + "grad_norm": 2.342528820037842, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7091128826141357, + "num_tokens": 380727841.0, + "step": 14934 + }, + { + "epoch": 1.6401273885350318, + "grad_norm": 2.281261682510376, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.743983268737793, + "num_tokens": 380752215.0, + "step": 14935 + }, + { + "epoch": 1.6402372062376456, + "grad_norm": 2.3380746841430664, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7224785089492798, + "num_tokens": 380775440.0, + "step": 14936 + }, + { + "epoch": 1.6403470239402593, + "grad_norm": 2.28589129447937, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7386388778686523, + "num_tokens": 380799788.0, + "step": 14937 + }, + { + "epoch": 1.6404568416428729, + "grad_norm": 2.204026222229004, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7162312269210815, + "num_tokens": 380825752.0, + "step": 14938 + }, + { + "epoch": 1.6405666593454864, + "grad_norm": 2.0903234481811523, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7130560278892517, + "num_tokens": 380856363.0, + "step": 14939 + }, + { + "epoch": 1.6406764770481002, + "grad_norm": 2.1744065284729004, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7137691974639893, + "num_tokens": 380882858.0, + "step": 14940 + }, + { + "epoch": 1.640786294750714, + "grad_norm": 2.3314592838287354, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7150408029556274, + "num_tokens": 380907443.0, + "step": 14941 + }, + { + "epoch": 1.6408961124533274, + "grad_norm": 2.331843137741089, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7210010290145874, + "num_tokens": 380930720.0, + "step": 14942 + }, + { + "epoch": 1.641005930155941, + "grad_norm": 2.1989684104919434, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7290045619010925, + "num_tokens": 380957058.0, + "step": 14943 + }, + { + "epoch": 1.6411157478585547, + "grad_norm": 2.3368613719940186, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7088331580162048, + "num_tokens": 380981149.0, + "step": 14944 + }, + { + "epoch": 1.6412255655611685, + "grad_norm": 2.2986295223236084, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.715003252029419, + "num_tokens": 381007004.0, + "step": 14945 + }, + { + "epoch": 1.6413353832637823, + "grad_norm": 2.5807454586029053, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.72406005859375, + "num_tokens": 381027833.0, + "step": 14946 + }, + { + "epoch": 1.6414452009663958, + "grad_norm": 2.2357168197631836, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7259417772293091, + "num_tokens": 381052562.0, + "step": 14947 + }, + { + "epoch": 1.6415550186690093, + "grad_norm": 2.020212411880493, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7217270135879517, + "num_tokens": 381082410.0, + "step": 14948 + }, + { + "epoch": 1.641664836371623, + "grad_norm": 2.2490172386169434, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7152543067932129, + "num_tokens": 381109111.0, + "step": 14949 + }, + { + "epoch": 1.6417746540742368, + "grad_norm": 2.3681342601776123, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7135542631149292, + "num_tokens": 381132714.0, + "step": 14950 + }, + { + "epoch": 1.6418844717768504, + "grad_norm": 2.4166860580444336, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7462265491485596, + "num_tokens": 381153786.0, + "step": 14951 + }, + { + "epoch": 1.6419942894794641, + "grad_norm": 2.005589246749878, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7049281597137451, + "num_tokens": 381186598.0, + "step": 14952 + }, + { + "epoch": 1.6421041071820777, + "grad_norm": 2.2677218914031982, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7126928567886353, + "num_tokens": 381211130.0, + "step": 14953 + }, + { + "epoch": 1.6422139248846914, + "grad_norm": 2.667602300643921, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7271369695663452, + "num_tokens": 381230824.0, + "step": 14954 + }, + { + "epoch": 1.6423237425873052, + "grad_norm": 2.1742849349975586, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6950826644897461, + "num_tokens": 381259545.0, + "step": 14955 + }, + { + "epoch": 1.6424335602899187, + "grad_norm": 2.220607042312622, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7235565781593323, + "num_tokens": 381286146.0, + "step": 14956 + }, + { + "epoch": 1.6425433779925322, + "grad_norm": 2.779082775115967, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7445909380912781, + "num_tokens": 381303156.0, + "step": 14957 + }, + { + "epoch": 1.642653195695146, + "grad_norm": 2.521785259246826, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7400428056716919, + "num_tokens": 381322820.0, + "step": 14958 + }, + { + "epoch": 1.6427630133977598, + "grad_norm": 2.254256010055542, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7318408489227295, + "num_tokens": 381349578.0, + "step": 14959 + }, + { + "epoch": 1.6428728311003735, + "grad_norm": 2.2635722160339355, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7118609547615051, + "num_tokens": 381378160.0, + "step": 14960 + }, + { + "epoch": 1.642982648802987, + "grad_norm": 2.644402265548706, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7335125207901001, + "num_tokens": 381397203.0, + "step": 14961 + }, + { + "epoch": 1.6430924665056006, + "grad_norm": 2.2388834953308105, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7169269323348999, + "num_tokens": 381421589.0, + "step": 14962 + }, + { + "epoch": 1.6432022842082143, + "grad_norm": 2.256148099899292, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7165170311927795, + "num_tokens": 381447586.0, + "step": 14963 + }, + { + "epoch": 1.643312101910828, + "grad_norm": 2.0823633670806885, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.686331033706665, + "num_tokens": 381479134.0, + "step": 14964 + }, + { + "epoch": 1.6434219196134416, + "grad_norm": 2.2924671173095703, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7032679319381714, + "num_tokens": 381505117.0, + "step": 14965 + }, + { + "epoch": 1.6435317373160554, + "grad_norm": 2.1359658241271973, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.710025429725647, + "num_tokens": 381531568.0, + "step": 14966 + }, + { + "epoch": 1.643641555018669, + "grad_norm": 2.2668182849884033, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.741881251335144, + "num_tokens": 381555370.0, + "step": 14967 + }, + { + "epoch": 1.6437513727212827, + "grad_norm": 2.525118827819824, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7356276512145996, + "num_tokens": 381575612.0, + "step": 14968 + }, + { + "epoch": 1.6438611904238964, + "grad_norm": 2.0572080612182617, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7192785143852234, + "num_tokens": 381608228.0, + "step": 14969 + }, + { + "epoch": 1.64397100812651, + "grad_norm": 2.305453062057495, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7292493581771851, + "num_tokens": 381631198.0, + "step": 14970 + }, + { + "epoch": 1.6440808258291235, + "grad_norm": 2.2039995193481445, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7035927176475525, + "num_tokens": 381660380.0, + "step": 14971 + }, + { + "epoch": 1.6441906435317373, + "grad_norm": 2.2528631687164307, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7262498140335083, + "num_tokens": 381686145.0, + "step": 14972 + }, + { + "epoch": 1.644300461234351, + "grad_norm": 2.484192371368408, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7421636581420898, + "num_tokens": 381707161.0, + "step": 14973 + }, + { + "epoch": 1.6444102789369648, + "grad_norm": 2.1585631370544434, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7204923629760742, + "num_tokens": 381733906.0, + "step": 14974 + }, + { + "epoch": 1.6445200966395783, + "grad_norm": 2.267275094985962, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7410534024238586, + "num_tokens": 381759643.0, + "step": 14975 + }, + { + "epoch": 1.6446299143421919, + "grad_norm": 2.625422239303589, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7220657467842102, + "num_tokens": 381779412.0, + "step": 14976 + }, + { + "epoch": 1.6447397320448056, + "grad_norm": 2.2782797813415527, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.705203652381897, + "num_tokens": 381804909.0, + "step": 14977 + }, + { + "epoch": 1.6448495497474194, + "grad_norm": 2.1901988983154297, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7150574326515198, + "num_tokens": 381831077.0, + "step": 14978 + }, + { + "epoch": 1.644959367450033, + "grad_norm": 2.0141379833221436, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7081186175346375, + "num_tokens": 381863536.0, + "step": 14979 + }, + { + "epoch": 1.6450691851526464, + "grad_norm": 2.2932655811309814, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7352274656295776, + "num_tokens": 381888436.0, + "step": 14980 + }, + { + "epoch": 1.6451790028552602, + "grad_norm": 2.4461894035339355, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7173213958740234, + "num_tokens": 381912151.0, + "step": 14981 + }, + { + "epoch": 1.645288820557874, + "grad_norm": 2.2508368492126465, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7223504781723022, + "num_tokens": 381938687.0, + "step": 14982 + }, + { + "epoch": 1.6453986382604877, + "grad_norm": 2.2799816131591797, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7218963503837585, + "num_tokens": 381964306.0, + "step": 14983 + }, + { + "epoch": 1.6455084559631012, + "grad_norm": 2.16266131401062, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.692535400390625, + "num_tokens": 381994178.0, + "step": 14984 + }, + { + "epoch": 1.6456182736657148, + "grad_norm": 2.3367388248443604, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7440693378448486, + "num_tokens": 382018367.0, + "step": 14985 + }, + { + "epoch": 1.6457280913683285, + "grad_norm": 1.9368500709533691, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7020936608314514, + "num_tokens": 382051033.0, + "step": 14986 + }, + { + "epoch": 1.6458379090709423, + "grad_norm": 2.223761796951294, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.6888415813446045, + "num_tokens": 382078217.0, + "step": 14987 + }, + { + "epoch": 1.645947726773556, + "grad_norm": 2.2441232204437256, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7208669781684875, + "num_tokens": 382103674.0, + "step": 14988 + }, + { + "epoch": 1.6460575444761696, + "grad_norm": 2.5237746238708496, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7271255850791931, + "num_tokens": 382126060.0, + "step": 14989 + }, + { + "epoch": 1.6461673621787831, + "grad_norm": 2.4213569164276123, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.6988881230354309, + "num_tokens": 382150352.0, + "step": 14990 + }, + { + "epoch": 1.6462771798813969, + "grad_norm": 2.2028563022613525, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7502862215042114, + "num_tokens": 382174485.0, + "step": 14991 + }, + { + "epoch": 1.6463869975840106, + "grad_norm": 2.1903634071350098, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7131204009056091, + "num_tokens": 382203536.0, + "step": 14992 + }, + { + "epoch": 1.6464968152866242, + "grad_norm": 2.4169037342071533, + "learning_rate": 1e-06, + "loss": 0.8015, + "mean_token_accuracy": 0.7422354817390442, + "num_tokens": 382224657.0, + "step": 14993 + }, + { + "epoch": 1.6466066329892377, + "grad_norm": 2.346663236618042, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7212188839912415, + "num_tokens": 382249869.0, + "step": 14994 + }, + { + "epoch": 1.6467164506918515, + "grad_norm": 2.0271782875061035, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7022654414176941, + "num_tokens": 382279852.0, + "step": 14995 + }, + { + "epoch": 1.6468262683944652, + "grad_norm": 2.4515960216522217, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.717678427696228, + "num_tokens": 382301634.0, + "step": 14996 + }, + { + "epoch": 1.646936086097079, + "grad_norm": 2.226151943206787, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7039936184883118, + "num_tokens": 382329289.0, + "step": 14997 + }, + { + "epoch": 1.6470459037996925, + "grad_norm": 2.1758670806884766, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6995154619216919, + "num_tokens": 382356315.0, + "step": 14998 + }, + { + "epoch": 1.647155721502306, + "grad_norm": 2.2988486289978027, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7281575798988342, + "num_tokens": 382382180.0, + "step": 14999 + }, + { + "epoch": 1.6472655392049198, + "grad_norm": 2.4993319511413574, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7045606374740601, + "num_tokens": 382405510.0, + "step": 15000 + }, + { + "epoch": 1.6473753569075336, + "grad_norm": 2.1537792682647705, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7642987966537476, + "num_tokens": 382428725.0, + "step": 15001 + }, + { + "epoch": 1.6474851746101473, + "grad_norm": 2.1188509464263916, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.6937119960784912, + "num_tokens": 382456424.0, + "step": 15002 + }, + { + "epoch": 1.6475949923127609, + "grad_norm": 2.2803165912628174, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7101892828941345, + "num_tokens": 382481426.0, + "step": 15003 + }, + { + "epoch": 1.6477048100153744, + "grad_norm": 2.5099189281463623, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7122262716293335, + "num_tokens": 382504143.0, + "step": 15004 + }, + { + "epoch": 1.6478146277179881, + "grad_norm": 2.218750238418579, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7171107530593872, + "num_tokens": 382530685.0, + "step": 15005 + }, + { + "epoch": 1.647924445420602, + "grad_norm": 2.3181519508361816, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7311108708381653, + "num_tokens": 382554773.0, + "step": 15006 + }, + { + "epoch": 1.6480342631232154, + "grad_norm": 2.2756247520446777, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.729434072971344, + "num_tokens": 382578882.0, + "step": 15007 + }, + { + "epoch": 1.648144080825829, + "grad_norm": 2.407599449157715, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.721726655960083, + "num_tokens": 382602783.0, + "step": 15008 + }, + { + "epoch": 1.6482538985284427, + "grad_norm": 2.1419622898101807, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7212347388267517, + "num_tokens": 382631747.0, + "step": 15009 + }, + { + "epoch": 1.6483637162310565, + "grad_norm": 2.1122307777404785, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6916956305503845, + "num_tokens": 382662125.0, + "step": 15010 + }, + { + "epoch": 1.6484735339336702, + "grad_norm": 2.486873149871826, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7090016007423401, + "num_tokens": 382684428.0, + "step": 15011 + }, + { + "epoch": 1.6485833516362838, + "grad_norm": 2.0128655433654785, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.6917320489883423, + "num_tokens": 382717491.0, + "step": 15012 + }, + { + "epoch": 1.6486931693388973, + "grad_norm": 2.0106847286224365, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7228676080703735, + "num_tokens": 382748034.0, + "step": 15013 + }, + { + "epoch": 1.648802987041511, + "grad_norm": 2.255901336669922, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7085628509521484, + "num_tokens": 382774492.0, + "step": 15014 + }, + { + "epoch": 1.6489128047441248, + "grad_norm": 2.2506680488586426, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7350497245788574, + "num_tokens": 382800641.0, + "step": 15015 + }, + { + "epoch": 1.6490226224467384, + "grad_norm": 2.3184118270874023, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7289680242538452, + "num_tokens": 382825371.0, + "step": 15016 + }, + { + "epoch": 1.6491324401493521, + "grad_norm": 2.249948501586914, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7175791263580322, + "num_tokens": 382851715.0, + "step": 15017 + }, + { + "epoch": 1.6492422578519657, + "grad_norm": 2.2176504135131836, + "learning_rate": 1e-06, + "loss": 0.7741, + "mean_token_accuracy": 0.7542529702186584, + "num_tokens": 382876223.0, + "step": 15018 + }, + { + "epoch": 1.6493520755545794, + "grad_norm": 2.0143675804138184, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7156065702438354, + "num_tokens": 382906046.0, + "step": 15019 + }, + { + "epoch": 1.6494618932571932, + "grad_norm": 2.14267897605896, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7053061723709106, + "num_tokens": 382935196.0, + "step": 15020 + }, + { + "epoch": 1.6495717109598067, + "grad_norm": 2.0996193885803223, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.719497799873352, + "num_tokens": 382965647.0, + "step": 15021 + }, + { + "epoch": 1.6496815286624202, + "grad_norm": 1.8306938409805298, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7014569640159607, + "num_tokens": 383002797.0, + "step": 15022 + }, + { + "epoch": 1.649791346365034, + "grad_norm": 2.4793570041656494, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.718514084815979, + "num_tokens": 383026186.0, + "step": 15023 + }, + { + "epoch": 1.6499011640676478, + "grad_norm": 2.1677486896514893, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7328469753265381, + "num_tokens": 383052335.0, + "step": 15024 + }, + { + "epoch": 1.6500109817702615, + "grad_norm": 2.2040834426879883, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7103667259216309, + "num_tokens": 383078956.0, + "step": 15025 + }, + { + "epoch": 1.650120799472875, + "grad_norm": 2.2272543907165527, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7219113111495972, + "num_tokens": 383107072.0, + "step": 15026 + }, + { + "epoch": 1.6502306171754886, + "grad_norm": 2.2234299182891846, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7027005553245544, + "num_tokens": 383133263.0, + "step": 15027 + }, + { + "epoch": 1.6503404348781023, + "grad_norm": 2.0294241905212402, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7078797817230225, + "num_tokens": 383165502.0, + "step": 15028 + }, + { + "epoch": 1.650450252580716, + "grad_norm": 2.3376498222351074, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7012109756469727, + "num_tokens": 383191233.0, + "step": 15029 + }, + { + "epoch": 1.6505600702833296, + "grad_norm": 2.2200660705566406, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7523821592330933, + "num_tokens": 383215804.0, + "step": 15030 + }, + { + "epoch": 1.6506698879859434, + "grad_norm": 2.2103259563446045, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7135573625564575, + "num_tokens": 383241858.0, + "step": 15031 + }, + { + "epoch": 1.650779705688557, + "grad_norm": 2.1898789405822754, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7178981900215149, + "num_tokens": 383269072.0, + "step": 15032 + }, + { + "epoch": 1.6508895233911707, + "grad_norm": 2.149893045425415, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7213689088821411, + "num_tokens": 383294571.0, + "step": 15033 + }, + { + "epoch": 1.6509993410937844, + "grad_norm": 2.3373618125915527, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7270195484161377, + "num_tokens": 383322246.0, + "step": 15034 + }, + { + "epoch": 1.651109158796398, + "grad_norm": 2.2807090282440186, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7324142456054688, + "num_tokens": 383346939.0, + "step": 15035 + }, + { + "epoch": 1.6512189764990115, + "grad_norm": 2.257810592651367, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7257449626922607, + "num_tokens": 383372221.0, + "step": 15036 + }, + { + "epoch": 1.6513287942016253, + "grad_norm": 2.2933340072631836, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7379408478736877, + "num_tokens": 383394992.0, + "step": 15037 + }, + { + "epoch": 1.651438611904239, + "grad_norm": 2.334256410598755, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7206040620803833, + "num_tokens": 383419731.0, + "step": 15038 + }, + { + "epoch": 1.6515484296068528, + "grad_norm": 2.3603382110595703, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.6971921920776367, + "num_tokens": 383444327.0, + "step": 15039 + }, + { + "epoch": 1.6516582473094663, + "grad_norm": 2.270751953125, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7216624021530151, + "num_tokens": 383470024.0, + "step": 15040 + }, + { + "epoch": 1.6517680650120798, + "grad_norm": 2.1435861587524414, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7160323262214661, + "num_tokens": 383498100.0, + "step": 15041 + }, + { + "epoch": 1.6518778827146936, + "grad_norm": 2.08506178855896, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7205439209938049, + "num_tokens": 383527325.0, + "step": 15042 + }, + { + "epoch": 1.6519877004173074, + "grad_norm": 1.952278733253479, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.6945146322250366, + "num_tokens": 383558538.0, + "step": 15043 + }, + { + "epoch": 1.652097518119921, + "grad_norm": 2.101445436477661, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7121644616127014, + "num_tokens": 383588861.0, + "step": 15044 + }, + { + "epoch": 1.6522073358225344, + "grad_norm": 2.4170968532562256, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7146235704421997, + "num_tokens": 383611242.0, + "step": 15045 + }, + { + "epoch": 1.6523171535251482, + "grad_norm": 2.497403144836426, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7169041633605957, + "num_tokens": 383633371.0, + "step": 15046 + }, + { + "epoch": 1.652426971227762, + "grad_norm": 2.3494653701782227, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.718087911605835, + "num_tokens": 383656626.0, + "step": 15047 + }, + { + "epoch": 1.6525367889303757, + "grad_norm": 2.147645950317383, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.724029541015625, + "num_tokens": 383684291.0, + "step": 15048 + }, + { + "epoch": 1.6526466066329892, + "grad_norm": 2.188835382461548, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.704828143119812, + "num_tokens": 383711452.0, + "step": 15049 + }, + { + "epoch": 1.6527564243356028, + "grad_norm": 2.2746832370758057, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7271462678909302, + "num_tokens": 383736944.0, + "step": 15050 + }, + { + "epoch": 1.6528662420382165, + "grad_norm": 2.5395655632019043, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7295029163360596, + "num_tokens": 383757242.0, + "step": 15051 + }, + { + "epoch": 1.6529760597408303, + "grad_norm": 2.3864755630493164, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7299893498420715, + "num_tokens": 383778418.0, + "step": 15052 + }, + { + "epoch": 1.653085877443444, + "grad_norm": 2.3184964656829834, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7276734709739685, + "num_tokens": 383803155.0, + "step": 15053 + }, + { + "epoch": 1.6531956951460576, + "grad_norm": 2.3937764167785645, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7105547189712524, + "num_tokens": 383825615.0, + "step": 15054 + }, + { + "epoch": 1.653305512848671, + "grad_norm": 2.609673500061035, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7383751273155212, + "num_tokens": 383844758.0, + "step": 15055 + }, + { + "epoch": 1.6534153305512849, + "grad_norm": 2.428577423095703, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.699124813079834, + "num_tokens": 383869376.0, + "step": 15056 + }, + { + "epoch": 1.6535251482538986, + "grad_norm": 2.132758617401123, + "learning_rate": 1e-06, + "loss": 0.7718, + "mean_token_accuracy": 0.7522673606872559, + "num_tokens": 383893989.0, + "step": 15057 + }, + { + "epoch": 1.6536349659565122, + "grad_norm": 2.349447727203369, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7243657112121582, + "num_tokens": 383917509.0, + "step": 15058 + }, + { + "epoch": 1.6537447836591257, + "grad_norm": 2.143456220626831, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7173111438751221, + "num_tokens": 383943128.0, + "step": 15059 + }, + { + "epoch": 1.6538546013617395, + "grad_norm": 2.4520761966705322, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7346044182777405, + "num_tokens": 383964034.0, + "step": 15060 + }, + { + "epoch": 1.6539644190643532, + "grad_norm": 2.261765718460083, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.750067949295044, + "num_tokens": 383985993.0, + "step": 15061 + }, + { + "epoch": 1.654074236766967, + "grad_norm": 2.3958775997161865, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7077701091766357, + "num_tokens": 384009781.0, + "step": 15062 + }, + { + "epoch": 1.6541840544695805, + "grad_norm": 2.0803959369659424, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7016013264656067, + "num_tokens": 384039208.0, + "step": 15063 + }, + { + "epoch": 1.654293872172194, + "grad_norm": 2.1219732761383057, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7160086631774902, + "num_tokens": 384067609.0, + "step": 15064 + }, + { + "epoch": 1.6544036898748078, + "grad_norm": 2.3450660705566406, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7275218367576599, + "num_tokens": 384092861.0, + "step": 15065 + }, + { + "epoch": 1.6545135075774215, + "grad_norm": 2.28711199760437, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7269107699394226, + "num_tokens": 384118010.0, + "step": 15066 + }, + { + "epoch": 1.6546233252800353, + "grad_norm": 2.325824737548828, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.726862907409668, + "num_tokens": 384143571.0, + "step": 15067 + }, + { + "epoch": 1.6547331429826488, + "grad_norm": 2.4445273876190186, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.738121509552002, + "num_tokens": 384167138.0, + "step": 15068 + }, + { + "epoch": 1.6548429606852624, + "grad_norm": 2.537796974182129, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.723814845085144, + "num_tokens": 384186906.0, + "step": 15069 + }, + { + "epoch": 1.6549527783878761, + "grad_norm": 2.4250285625457764, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7258442640304565, + "num_tokens": 384210812.0, + "step": 15070 + }, + { + "epoch": 1.65506259609049, + "grad_norm": 1.9988363981246948, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7273113131523132, + "num_tokens": 384240467.0, + "step": 15071 + }, + { + "epoch": 1.6551724137931034, + "grad_norm": 2.164888620376587, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7208479642868042, + "num_tokens": 384267070.0, + "step": 15072 + }, + { + "epoch": 1.655282231495717, + "grad_norm": 2.5707311630249023, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7201672792434692, + "num_tokens": 384287238.0, + "step": 15073 + }, + { + "epoch": 1.6553920491983307, + "grad_norm": 2.725186347961426, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.731907844543457, + "num_tokens": 384305063.0, + "step": 15074 + }, + { + "epoch": 1.6555018669009445, + "grad_norm": 1.991692304611206, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.701960563659668, + "num_tokens": 384335326.0, + "step": 15075 + }, + { + "epoch": 1.6556116846035582, + "grad_norm": 2.01631760597229, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7249631285667419, + "num_tokens": 384365961.0, + "step": 15076 + }, + { + "epoch": 1.6557215023061718, + "grad_norm": 2.384058713912964, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7278269529342651, + "num_tokens": 384389882.0, + "step": 15077 + }, + { + "epoch": 1.6558313200087853, + "grad_norm": 2.4745986461639404, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7183007001876831, + "num_tokens": 384412116.0, + "step": 15078 + }, + { + "epoch": 1.655941137711399, + "grad_norm": 2.624462604522705, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7184922695159912, + "num_tokens": 384434611.0, + "step": 15079 + }, + { + "epoch": 1.6560509554140128, + "grad_norm": 2.158957004547119, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7272952795028687, + "num_tokens": 384463912.0, + "step": 15080 + }, + { + "epoch": 1.6561607731166264, + "grad_norm": 2.3364098072052, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7125033140182495, + "num_tokens": 384487566.0, + "step": 15081 + }, + { + "epoch": 1.65627059081924, + "grad_norm": 2.0533688068389893, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7033530473709106, + "num_tokens": 384517981.0, + "step": 15082 + }, + { + "epoch": 1.6563804085218536, + "grad_norm": 2.1343727111816406, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.715944766998291, + "num_tokens": 384544751.0, + "step": 15083 + }, + { + "epoch": 1.6564902262244674, + "grad_norm": 2.4771530628204346, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7356483936309814, + "num_tokens": 384565425.0, + "step": 15084 + }, + { + "epoch": 1.6566000439270812, + "grad_norm": 2.2362420558929443, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.731708288192749, + "num_tokens": 384592339.0, + "step": 15085 + }, + { + "epoch": 1.6567098616296947, + "grad_norm": 2.035466194152832, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7213497161865234, + "num_tokens": 384621307.0, + "step": 15086 + }, + { + "epoch": 1.6568196793323082, + "grad_norm": 2.1044952869415283, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.723192036151886, + "num_tokens": 384651858.0, + "step": 15087 + }, + { + "epoch": 1.656929497034922, + "grad_norm": 2.4282994270324707, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7439780235290527, + "num_tokens": 384673085.0, + "step": 15088 + }, + { + "epoch": 1.6570393147375357, + "grad_norm": 2.2861011028289795, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.745855450630188, + "num_tokens": 384698088.0, + "step": 15089 + }, + { + "epoch": 1.6571491324401495, + "grad_norm": 2.196721315383911, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.720683217048645, + "num_tokens": 384724762.0, + "step": 15090 + }, + { + "epoch": 1.657258950142763, + "grad_norm": 1.981104850769043, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7305309176445007, + "num_tokens": 384754200.0, + "step": 15091 + }, + { + "epoch": 1.6573687678453766, + "grad_norm": 2.1815431118011475, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7057502865791321, + "num_tokens": 384781959.0, + "step": 15092 + }, + { + "epoch": 1.6574785855479903, + "grad_norm": 2.4136228561401367, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7404478788375854, + "num_tokens": 384803351.0, + "step": 15093 + }, + { + "epoch": 1.657588403250604, + "grad_norm": 2.3017122745513916, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.740815281867981, + "num_tokens": 384827414.0, + "step": 15094 + }, + { + "epoch": 1.6576982209532176, + "grad_norm": 2.5261645317077637, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7281068563461304, + "num_tokens": 384849162.0, + "step": 15095 + }, + { + "epoch": 1.6578080386558314, + "grad_norm": 2.225621461868286, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7119650840759277, + "num_tokens": 384878095.0, + "step": 15096 + }, + { + "epoch": 1.657917856358445, + "grad_norm": 2.2796146869659424, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7292131185531616, + "num_tokens": 384903169.0, + "step": 15097 + }, + { + "epoch": 1.6580276740610587, + "grad_norm": 2.061743974685669, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7135628461837769, + "num_tokens": 384932910.0, + "step": 15098 + }, + { + "epoch": 1.6581374917636724, + "grad_norm": 2.1089026927948, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7261431217193604, + "num_tokens": 384960863.0, + "step": 15099 + }, + { + "epoch": 1.658247309466286, + "grad_norm": 2.1779167652130127, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7395460605621338, + "num_tokens": 384985946.0, + "step": 15100 + }, + { + "epoch": 1.6583571271688995, + "grad_norm": 2.7338225841522217, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.74127197265625, + "num_tokens": 385004493.0, + "step": 15101 + }, + { + "epoch": 1.6584669448715132, + "grad_norm": 2.228618860244751, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7302051782608032, + "num_tokens": 385028590.0, + "step": 15102 + }, + { + "epoch": 1.658576762574127, + "grad_norm": 2.2400763034820557, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7182950973510742, + "num_tokens": 385054202.0, + "step": 15103 + }, + { + "epoch": 1.6586865802767408, + "grad_norm": 2.5134596824645996, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.7525737285614014, + "num_tokens": 385073383.0, + "step": 15104 + }, + { + "epoch": 1.6587963979793543, + "grad_norm": 2.325538158416748, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7174283266067505, + "num_tokens": 385099562.0, + "step": 15105 + }, + { + "epoch": 1.6589062156819678, + "grad_norm": 2.337517261505127, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7047866582870483, + "num_tokens": 385122649.0, + "step": 15106 + }, + { + "epoch": 1.6590160333845816, + "grad_norm": 2.0755462646484375, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.6960532069206238, + "num_tokens": 385152360.0, + "step": 15107 + }, + { + "epoch": 1.6591258510871953, + "grad_norm": 2.1874935626983643, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7251535654067993, + "num_tokens": 385178165.0, + "step": 15108 + }, + { + "epoch": 1.6592356687898089, + "grad_norm": 2.255492925643921, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7575340270996094, + "num_tokens": 385200546.0, + "step": 15109 + }, + { + "epoch": 1.6593454864924224, + "grad_norm": 2.3570003509521484, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7204104661941528, + "num_tokens": 385222346.0, + "step": 15110 + }, + { + "epoch": 1.6594553041950362, + "grad_norm": 2.1090269088745117, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7018976211547852, + "num_tokens": 385249184.0, + "step": 15111 + }, + { + "epoch": 1.65956512189765, + "grad_norm": 1.9804946184158325, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6973101496696472, + "num_tokens": 385280518.0, + "step": 15112 + }, + { + "epoch": 1.6596749396002637, + "grad_norm": 2.315913677215576, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7393509745597839, + "num_tokens": 385301298.0, + "step": 15113 + }, + { + "epoch": 1.6597847573028772, + "grad_norm": 2.318279981613159, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7215887904167175, + "num_tokens": 385325855.0, + "step": 15114 + }, + { + "epoch": 1.6598945750054908, + "grad_norm": 2.1884820461273193, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.700373649597168, + "num_tokens": 385352079.0, + "step": 15115 + }, + { + "epoch": 1.6600043927081045, + "grad_norm": 2.2300963401794434, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7146187424659729, + "num_tokens": 385379503.0, + "step": 15116 + }, + { + "epoch": 1.6601142104107183, + "grad_norm": 2.163477897644043, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.726902961730957, + "num_tokens": 385406790.0, + "step": 15117 + }, + { + "epoch": 1.660224028113332, + "grad_norm": 1.9878290891647339, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.711502194404602, + "num_tokens": 385437339.0, + "step": 15118 + }, + { + "epoch": 1.6603338458159456, + "grad_norm": 2.2443459033966064, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7179778218269348, + "num_tokens": 385464372.0, + "step": 15119 + }, + { + "epoch": 1.660443663518559, + "grad_norm": 2.572688341140747, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7311257123947144, + "num_tokens": 385485088.0, + "step": 15120 + }, + { + "epoch": 1.6605534812211729, + "grad_norm": 2.0999701023101807, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7362732887268066, + "num_tokens": 385510538.0, + "step": 15121 + }, + { + "epoch": 1.6606632989237866, + "grad_norm": 2.127237558364868, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7503663301467896, + "num_tokens": 385534869.0, + "step": 15122 + }, + { + "epoch": 1.6607731166264001, + "grad_norm": 2.035874843597412, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7150015830993652, + "num_tokens": 385565676.0, + "step": 15123 + }, + { + "epoch": 1.6608829343290137, + "grad_norm": 2.5653626918792725, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7236684560775757, + "num_tokens": 385587850.0, + "step": 15124 + }, + { + "epoch": 1.6609927520316274, + "grad_norm": 2.3088252544403076, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7220724821090698, + "num_tokens": 385613271.0, + "step": 15125 + }, + { + "epoch": 1.6611025697342412, + "grad_norm": 2.08879017829895, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7133485674858093, + "num_tokens": 385641972.0, + "step": 15126 + }, + { + "epoch": 1.661212387436855, + "grad_norm": 2.336467981338501, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7196229100227356, + "num_tokens": 385664935.0, + "step": 15127 + }, + { + "epoch": 1.6613222051394685, + "grad_norm": 2.04732608795166, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7115613222122192, + "num_tokens": 385694119.0, + "step": 15128 + }, + { + "epoch": 1.661432022842082, + "grad_norm": 2.2606492042541504, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7398662567138672, + "num_tokens": 385719598.0, + "step": 15129 + }, + { + "epoch": 1.6615418405446958, + "grad_norm": 2.1331992149353027, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7345360517501831, + "num_tokens": 385748601.0, + "step": 15130 + }, + { + "epoch": 1.6616516582473095, + "grad_norm": 2.1173055171966553, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7020042538642883, + "num_tokens": 385779093.0, + "step": 15131 + }, + { + "epoch": 1.661761475949923, + "grad_norm": 2.0805208683013916, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7134889960289001, + "num_tokens": 385806202.0, + "step": 15132 + }, + { + "epoch": 1.6618712936525368, + "grad_norm": 2.204853057861328, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7258332371711731, + "num_tokens": 385832696.0, + "step": 15133 + }, + { + "epoch": 1.6619811113551504, + "grad_norm": 2.052492141723633, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.725832462310791, + "num_tokens": 385861449.0, + "step": 15134 + }, + { + "epoch": 1.6620909290577641, + "grad_norm": 2.2614099979400635, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7021021246910095, + "num_tokens": 385887422.0, + "step": 15135 + }, + { + "epoch": 1.6622007467603779, + "grad_norm": 2.188385486602783, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7088280916213989, + "num_tokens": 385914385.0, + "step": 15136 + }, + { + "epoch": 1.6623105644629914, + "grad_norm": 2.326274871826172, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7369092702865601, + "num_tokens": 385937333.0, + "step": 15137 + }, + { + "epoch": 1.662420382165605, + "grad_norm": 2.071928024291992, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7245231866836548, + "num_tokens": 385969016.0, + "step": 15138 + }, + { + "epoch": 1.6625301998682187, + "grad_norm": 2.251505136489868, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7253743410110474, + "num_tokens": 385993283.0, + "step": 15139 + }, + { + "epoch": 1.6626400175708325, + "grad_norm": 1.7850134372711182, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6920466423034668, + "num_tokens": 386030922.0, + "step": 15140 + }, + { + "epoch": 1.6627498352734462, + "grad_norm": 2.742482900619507, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.745836615562439, + "num_tokens": 386048841.0, + "step": 15141 + }, + { + "epoch": 1.6628596529760598, + "grad_norm": 2.101223945617676, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7349609136581421, + "num_tokens": 386074613.0, + "step": 15142 + }, + { + "epoch": 1.6629694706786733, + "grad_norm": 2.5432944297790527, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7114952802658081, + "num_tokens": 386096452.0, + "step": 15143 + }, + { + "epoch": 1.663079288381287, + "grad_norm": 2.2689685821533203, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7355449199676514, + "num_tokens": 386119935.0, + "step": 15144 + }, + { + "epoch": 1.6631891060839008, + "grad_norm": 2.4973130226135254, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7394380569458008, + "num_tokens": 386143912.0, + "step": 15145 + }, + { + "epoch": 1.6632989237865143, + "grad_norm": 2.398613452911377, + "learning_rate": 1e-06, + "loss": 0.8, + "mean_token_accuracy": 0.7453091144561768, + "num_tokens": 386165365.0, + "step": 15146 + }, + { + "epoch": 1.663408741489128, + "grad_norm": 2.1740498542785645, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.705358624458313, + "num_tokens": 386192374.0, + "step": 15147 + }, + { + "epoch": 1.6635185591917416, + "grad_norm": 2.2885818481445312, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7370284795761108, + "num_tokens": 386214321.0, + "step": 15148 + }, + { + "epoch": 1.6636283768943554, + "grad_norm": 2.187206268310547, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.6943792104721069, + "num_tokens": 386242167.0, + "step": 15149 + }, + { + "epoch": 1.6637381945969691, + "grad_norm": 2.0499730110168457, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7321946024894714, + "num_tokens": 386269891.0, + "step": 15150 + }, + { + "epoch": 1.6638480122995827, + "grad_norm": 2.550217628479004, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7216644883155823, + "num_tokens": 386289551.0, + "step": 15151 + }, + { + "epoch": 1.6639578300021962, + "grad_norm": 2.2983055114746094, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7050963640213013, + "num_tokens": 386314688.0, + "step": 15152 + }, + { + "epoch": 1.66406764770481, + "grad_norm": 2.630366086959839, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7322230339050293, + "num_tokens": 386335047.0, + "step": 15153 + }, + { + "epoch": 1.6641774654074237, + "grad_norm": 2.2166032791137695, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7050091624259949, + "num_tokens": 386362159.0, + "step": 15154 + }, + { + "epoch": 1.6642872831100375, + "grad_norm": 2.0465641021728516, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7171172499656677, + "num_tokens": 386391931.0, + "step": 15155 + }, + { + "epoch": 1.664397100812651, + "grad_norm": 2.1883482933044434, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7160614728927612, + "num_tokens": 386419191.0, + "step": 15156 + }, + { + "epoch": 1.6645069185152646, + "grad_norm": 2.1098763942718506, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7434799075126648, + "num_tokens": 386448210.0, + "step": 15157 + }, + { + "epoch": 1.6646167362178783, + "grad_norm": 2.326007604598999, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7096490859985352, + "num_tokens": 386472891.0, + "step": 15158 + }, + { + "epoch": 1.664726553920492, + "grad_norm": 2.3508105278015137, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7083718776702881, + "num_tokens": 386499039.0, + "step": 15159 + }, + { + "epoch": 1.6648363716231056, + "grad_norm": 2.299088478088379, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7351453900337219, + "num_tokens": 386522475.0, + "step": 15160 + }, + { + "epoch": 1.6649461893257194, + "grad_norm": 2.2208759784698486, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7375794649124146, + "num_tokens": 386546930.0, + "step": 15161 + }, + { + "epoch": 1.665056007028333, + "grad_norm": 1.9617676734924316, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7066056728363037, + "num_tokens": 386581501.0, + "step": 15162 + }, + { + "epoch": 1.6651658247309467, + "grad_norm": 1.939502477645874, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7118866443634033, + "num_tokens": 386615128.0, + "step": 15163 + }, + { + "epoch": 1.6652756424335604, + "grad_norm": 2.2356090545654297, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.6935292482376099, + "num_tokens": 386641179.0, + "step": 15164 + }, + { + "epoch": 1.665385460136174, + "grad_norm": 2.364217758178711, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7176576852798462, + "num_tokens": 386663706.0, + "step": 15165 + }, + { + "epoch": 1.6654952778387875, + "grad_norm": 2.240687847137451, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.729499101638794, + "num_tokens": 386689878.0, + "step": 15166 + }, + { + "epoch": 1.6656050955414012, + "grad_norm": 2.3944647312164307, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7273378372192383, + "num_tokens": 386711343.0, + "step": 15167 + }, + { + "epoch": 1.665714913244015, + "grad_norm": 2.088205575942993, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7077146768569946, + "num_tokens": 386741075.0, + "step": 15168 + }, + { + "epoch": 1.6658247309466288, + "grad_norm": 2.3342158794403076, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7424922585487366, + "num_tokens": 386764014.0, + "step": 15169 + }, + { + "epoch": 1.6659345486492423, + "grad_norm": 2.0777106285095215, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7088047862052917, + "num_tokens": 386791790.0, + "step": 15170 + }, + { + "epoch": 1.6660443663518558, + "grad_norm": 2.1564323902130127, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.732376754283905, + "num_tokens": 386820117.0, + "step": 15171 + }, + { + "epoch": 1.6661541840544696, + "grad_norm": 2.092064380645752, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.6971070766448975, + "num_tokens": 386850590.0, + "step": 15172 + }, + { + "epoch": 1.6662640017570833, + "grad_norm": 2.1297450065612793, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7269107103347778, + "num_tokens": 386877775.0, + "step": 15173 + }, + { + "epoch": 1.6663738194596969, + "grad_norm": 2.0523524284362793, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7160289287567139, + "num_tokens": 386907579.0, + "step": 15174 + }, + { + "epoch": 1.6664836371623104, + "grad_norm": 2.1490018367767334, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7274829149246216, + "num_tokens": 386933243.0, + "step": 15175 + }, + { + "epoch": 1.6665934548649242, + "grad_norm": 2.020462989807129, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7334771156311035, + "num_tokens": 386964436.0, + "step": 15176 + }, + { + "epoch": 1.666703272567538, + "grad_norm": 2.325273036956787, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.6967970132827759, + "num_tokens": 386991304.0, + "step": 15177 + }, + { + "epoch": 1.6668130902701517, + "grad_norm": 2.032979965209961, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.728519082069397, + "num_tokens": 387019112.0, + "step": 15178 + }, + { + "epoch": 1.6669229079727652, + "grad_norm": 2.2264480590820312, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7217797040939331, + "num_tokens": 387044497.0, + "step": 15179 + }, + { + "epoch": 1.6670327256753787, + "grad_norm": 2.0102243423461914, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.6956658363342285, + "num_tokens": 387076098.0, + "step": 15180 + }, + { + "epoch": 1.6671425433779925, + "grad_norm": 2.4515206813812256, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7347903251647949, + "num_tokens": 387098390.0, + "step": 15181 + }, + { + "epoch": 1.6672523610806063, + "grad_norm": 2.348142147064209, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7389594316482544, + "num_tokens": 387121722.0, + "step": 15182 + }, + { + "epoch": 1.66736217878322, + "grad_norm": 2.270646810531616, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.739279568195343, + "num_tokens": 387144604.0, + "step": 15183 + }, + { + "epoch": 1.6674719964858336, + "grad_norm": 2.396010160446167, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7392368316650391, + "num_tokens": 387166717.0, + "step": 15184 + }, + { + "epoch": 1.667581814188447, + "grad_norm": 2.636812686920166, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.727760910987854, + "num_tokens": 387187033.0, + "step": 15185 + }, + { + "epoch": 1.6676916318910608, + "grad_norm": 2.1286354064941406, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7048752307891846, + "num_tokens": 387215645.0, + "step": 15186 + }, + { + "epoch": 1.6678014495936746, + "grad_norm": 2.450885534286499, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7173551321029663, + "num_tokens": 387237854.0, + "step": 15187 + }, + { + "epoch": 1.6679112672962881, + "grad_norm": 2.369746685028076, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7097259759902954, + "num_tokens": 387261823.0, + "step": 15188 + }, + { + "epoch": 1.6680210849989017, + "grad_norm": 2.2490956783294678, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7118082046508789, + "num_tokens": 387288239.0, + "step": 15189 + }, + { + "epoch": 1.6681309027015154, + "grad_norm": 2.4485840797424316, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7133936285972595, + "num_tokens": 387310582.0, + "step": 15190 + }, + { + "epoch": 1.6682407204041292, + "grad_norm": 2.0358593463897705, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7420656681060791, + "num_tokens": 387339426.0, + "step": 15191 + }, + { + "epoch": 1.668350538106743, + "grad_norm": 2.0132133960723877, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.728460431098938, + "num_tokens": 387370554.0, + "step": 15192 + }, + { + "epoch": 1.6684603558093565, + "grad_norm": 2.2139699459075928, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7000291347503662, + "num_tokens": 387398225.0, + "step": 15193 + }, + { + "epoch": 1.66857017351197, + "grad_norm": 2.478646993637085, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7407479286193848, + "num_tokens": 387418914.0, + "step": 15194 + }, + { + "epoch": 1.6686799912145838, + "grad_norm": 2.543476104736328, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7349162697792053, + "num_tokens": 387437796.0, + "step": 15195 + }, + { + "epoch": 1.6687898089171975, + "grad_norm": 2.20780086517334, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7266125679016113, + "num_tokens": 387463667.0, + "step": 15196 + }, + { + "epoch": 1.668899626619811, + "grad_norm": 2.2664668560028076, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7297148108482361, + "num_tokens": 387488314.0, + "step": 15197 + }, + { + "epoch": 1.6690094443224248, + "grad_norm": 2.293431282043457, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7127155661582947, + "num_tokens": 387514212.0, + "step": 15198 + }, + { + "epoch": 1.6691192620250384, + "grad_norm": 1.881626009941101, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.739904522895813, + "num_tokens": 387547019.0, + "step": 15199 + }, + { + "epoch": 1.669229079727652, + "grad_norm": 2.2386324405670166, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7242147922515869, + "num_tokens": 387573041.0, + "step": 15200 + }, + { + "epoch": 1.6693388974302659, + "grad_norm": 2.1978869438171387, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7197542190551758, + "num_tokens": 387600114.0, + "step": 15201 + }, + { + "epoch": 1.6694487151328794, + "grad_norm": 2.183133363723755, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.716032862663269, + "num_tokens": 387626526.0, + "step": 15202 + }, + { + "epoch": 1.669558532835493, + "grad_norm": 2.1836585998535156, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7196027040481567, + "num_tokens": 387653091.0, + "step": 15203 + }, + { + "epoch": 1.6696683505381067, + "grad_norm": 2.403177261352539, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7487598061561584, + "num_tokens": 387674163.0, + "step": 15204 + }, + { + "epoch": 1.6697781682407205, + "grad_norm": 2.436227321624756, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.709040641784668, + "num_tokens": 387698559.0, + "step": 15205 + }, + { + "epoch": 1.6698879859433342, + "grad_norm": 2.1394834518432617, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7204272747039795, + "num_tokens": 387726851.0, + "step": 15206 + }, + { + "epoch": 1.6699978036459477, + "grad_norm": 2.059992551803589, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7078197002410889, + "num_tokens": 387755839.0, + "step": 15207 + }, + { + "epoch": 1.6701076213485613, + "grad_norm": 2.089177131652832, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7185023427009583, + "num_tokens": 387784251.0, + "step": 15208 + }, + { + "epoch": 1.670217439051175, + "grad_norm": 2.410322904586792, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7117913961410522, + "num_tokens": 387806512.0, + "step": 15209 + }, + { + "epoch": 1.6703272567537888, + "grad_norm": 2.546910285949707, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7128770351409912, + "num_tokens": 387827699.0, + "step": 15210 + }, + { + "epoch": 1.6704370744564023, + "grad_norm": 2.4236700534820557, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.714424729347229, + "num_tokens": 387850195.0, + "step": 15211 + }, + { + "epoch": 1.670546892159016, + "grad_norm": 2.320099353790283, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7391767501831055, + "num_tokens": 387872815.0, + "step": 15212 + }, + { + "epoch": 1.6706567098616296, + "grad_norm": 2.7579398155212402, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.7368477582931519, + "num_tokens": 387890327.0, + "step": 15213 + }, + { + "epoch": 1.6707665275642434, + "grad_norm": 2.3534789085388184, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7287436723709106, + "num_tokens": 387911442.0, + "step": 15214 + }, + { + "epoch": 1.6708763452668571, + "grad_norm": 2.5719335079193115, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7167956233024597, + "num_tokens": 387933244.0, + "step": 15215 + }, + { + "epoch": 1.6709861629694707, + "grad_norm": 2.221707582473755, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7111386656761169, + "num_tokens": 387961042.0, + "step": 15216 + }, + { + "epoch": 1.6710959806720842, + "grad_norm": 2.1781206130981445, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7165882587432861, + "num_tokens": 387986864.0, + "step": 15217 + }, + { + "epoch": 1.671205798374698, + "grad_norm": 2.2724826335906982, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7203370332717896, + "num_tokens": 388012486.0, + "step": 15218 + }, + { + "epoch": 1.6713156160773117, + "grad_norm": 2.2156026363372803, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.736528754234314, + "num_tokens": 388037285.0, + "step": 15219 + }, + { + "epoch": 1.6714254337799255, + "grad_norm": 2.150629997253418, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.713096559047699, + "num_tokens": 388066708.0, + "step": 15220 + }, + { + "epoch": 1.671535251482539, + "grad_norm": 2.0635690689086914, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7284286618232727, + "num_tokens": 388093989.0, + "step": 15221 + }, + { + "epoch": 1.6716450691851525, + "grad_norm": 2.2435083389282227, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.6970964670181274, + "num_tokens": 388120848.0, + "step": 15222 + }, + { + "epoch": 1.6717548868877663, + "grad_norm": 2.4646451473236084, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7179834842681885, + "num_tokens": 388141881.0, + "step": 15223 + }, + { + "epoch": 1.67186470459038, + "grad_norm": 2.2325921058654785, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.725308895111084, + "num_tokens": 388168670.0, + "step": 15224 + }, + { + "epoch": 1.6719745222929936, + "grad_norm": 2.2317800521850586, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.716029167175293, + "num_tokens": 388194516.0, + "step": 15225 + }, + { + "epoch": 1.6720843399956071, + "grad_norm": 2.3576853275299072, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7116780281066895, + "num_tokens": 388217070.0, + "step": 15226 + }, + { + "epoch": 1.6721941576982209, + "grad_norm": 1.960622787475586, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7254565954208374, + "num_tokens": 388247836.0, + "step": 15227 + }, + { + "epoch": 1.6723039754008346, + "grad_norm": 2.6479313373565674, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7443910837173462, + "num_tokens": 388265535.0, + "step": 15228 + }, + { + "epoch": 1.6724137931034484, + "grad_norm": 2.070188283920288, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7049380540847778, + "num_tokens": 388295220.0, + "step": 15229 + }, + { + "epoch": 1.672523610806062, + "grad_norm": 2.191100597381592, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7226418852806091, + "num_tokens": 388322850.0, + "step": 15230 + }, + { + "epoch": 1.6726334285086755, + "grad_norm": 2.4917097091674805, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7348464131355286, + "num_tokens": 388343470.0, + "step": 15231 + }, + { + "epoch": 1.6727432462112892, + "grad_norm": 2.1606454849243164, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7124160528182983, + "num_tokens": 388370960.0, + "step": 15232 + }, + { + "epoch": 1.672853063913903, + "grad_norm": 2.607720375061035, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7345223426818848, + "num_tokens": 388391437.0, + "step": 15233 + }, + { + "epoch": 1.6729628816165167, + "grad_norm": 2.7364919185638428, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7307307124137878, + "num_tokens": 388409296.0, + "step": 15234 + }, + { + "epoch": 1.6730726993191303, + "grad_norm": 2.6431686878204346, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7302594184875488, + "num_tokens": 388429724.0, + "step": 15235 + }, + { + "epoch": 1.6731825170217438, + "grad_norm": 2.2943975925445557, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7367098927497864, + "num_tokens": 388453609.0, + "step": 15236 + }, + { + "epoch": 1.6732923347243576, + "grad_norm": 2.196729898452759, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7129117250442505, + "num_tokens": 388480663.0, + "step": 15237 + }, + { + "epoch": 1.6734021524269713, + "grad_norm": 2.5581352710723877, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7421643733978271, + "num_tokens": 388500260.0, + "step": 15238 + }, + { + "epoch": 1.6735119701295849, + "grad_norm": 2.8858556747436523, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.7441883087158203, + "num_tokens": 388517690.0, + "step": 15239 + }, + { + "epoch": 1.6736217878321984, + "grad_norm": 2.263908863067627, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7310917377471924, + "num_tokens": 388542058.0, + "step": 15240 + }, + { + "epoch": 1.6737316055348122, + "grad_norm": 2.4124324321746826, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.7376558184623718, + "num_tokens": 388563857.0, + "step": 15241 + }, + { + "epoch": 1.673841423237426, + "grad_norm": 2.2430527210235596, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7263230681419373, + "num_tokens": 388588038.0, + "step": 15242 + }, + { + "epoch": 1.6739512409400397, + "grad_norm": 2.277432918548584, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.723885178565979, + "num_tokens": 388614568.0, + "step": 15243 + }, + { + "epoch": 1.6740610586426532, + "grad_norm": 2.218721628189087, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7493302822113037, + "num_tokens": 388638924.0, + "step": 15244 + }, + { + "epoch": 1.6741708763452667, + "grad_norm": 2.237916946411133, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.6962332725524902, + "num_tokens": 388665959.0, + "step": 15245 + }, + { + "epoch": 1.6742806940478805, + "grad_norm": 2.4867074489593506, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7268683314323425, + "num_tokens": 388686885.0, + "step": 15246 + }, + { + "epoch": 1.6743905117504942, + "grad_norm": 2.0689897537231445, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7185890078544617, + "num_tokens": 388715690.0, + "step": 15247 + }, + { + "epoch": 1.674500329453108, + "grad_norm": 1.95343816280365, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7267427444458008, + "num_tokens": 388749595.0, + "step": 15248 + }, + { + "epoch": 1.6746101471557215, + "grad_norm": 2.3125555515289307, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7137936949729919, + "num_tokens": 388772925.0, + "step": 15249 + }, + { + "epoch": 1.674719964858335, + "grad_norm": 2.151496171951294, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7268377542495728, + "num_tokens": 388800603.0, + "step": 15250 + }, + { + "epoch": 1.6748297825609488, + "grad_norm": 1.965448260307312, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7216444611549377, + "num_tokens": 388832464.0, + "step": 15251 + }, + { + "epoch": 1.6749396002635626, + "grad_norm": 2.1834988594055176, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7110938429832458, + "num_tokens": 388858521.0, + "step": 15252 + }, + { + "epoch": 1.6750494179661761, + "grad_norm": 2.238055944442749, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.69530189037323, + "num_tokens": 388885550.0, + "step": 15253 + }, + { + "epoch": 1.6751592356687897, + "grad_norm": 2.214707136154175, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.710625171661377, + "num_tokens": 388912120.0, + "step": 15254 + }, + { + "epoch": 1.6752690533714034, + "grad_norm": 2.7832207679748535, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7356890439987183, + "num_tokens": 388931082.0, + "step": 15255 + }, + { + "epoch": 1.6753788710740172, + "grad_norm": 2.3499345779418945, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7290216088294983, + "num_tokens": 388954251.0, + "step": 15256 + }, + { + "epoch": 1.675488688776631, + "grad_norm": 2.3494138717651367, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7124578952789307, + "num_tokens": 388980244.0, + "step": 15257 + }, + { + "epoch": 1.6755985064792445, + "grad_norm": 2.136657476425171, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7048364877700806, + "num_tokens": 389007800.0, + "step": 15258 + }, + { + "epoch": 1.675708324181858, + "grad_norm": 2.154266357421875, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7179068326950073, + "num_tokens": 389034938.0, + "step": 15259 + }, + { + "epoch": 1.6758181418844718, + "grad_norm": 2.0241737365722656, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7192108631134033, + "num_tokens": 389064294.0, + "step": 15260 + }, + { + "epoch": 1.6759279595870855, + "grad_norm": 2.323554039001465, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7079517841339111, + "num_tokens": 389089789.0, + "step": 15261 + }, + { + "epoch": 1.676037777289699, + "grad_norm": 1.99447500705719, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7146739959716797, + "num_tokens": 389121114.0, + "step": 15262 + }, + { + "epoch": 1.6761475949923128, + "grad_norm": 2.0365686416625977, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7139277458190918, + "num_tokens": 389153473.0, + "step": 15263 + }, + { + "epoch": 1.6762574126949263, + "grad_norm": 2.4057066440582275, + "learning_rate": 1e-06, + "loss": 0.7799, + "mean_token_accuracy": 0.754037618637085, + "num_tokens": 389175625.0, + "step": 15264 + }, + { + "epoch": 1.67636723039754, + "grad_norm": 2.0167202949523926, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.703438401222229, + "num_tokens": 389205550.0, + "step": 15265 + }, + { + "epoch": 1.6764770481001539, + "grad_norm": 2.3760733604431152, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7362436056137085, + "num_tokens": 389230295.0, + "step": 15266 + }, + { + "epoch": 1.6765868658027674, + "grad_norm": 2.0751113891601562, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7134529948234558, + "num_tokens": 389260249.0, + "step": 15267 + }, + { + "epoch": 1.676696683505381, + "grad_norm": 2.289043426513672, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7319514155387878, + "num_tokens": 389285183.0, + "step": 15268 + }, + { + "epoch": 1.6768065012079947, + "grad_norm": 2.1573898792266846, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7111623883247375, + "num_tokens": 389312646.0, + "step": 15269 + }, + { + "epoch": 1.6769163189106084, + "grad_norm": 2.030409574508667, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7323395013809204, + "num_tokens": 389340522.0, + "step": 15270 + }, + { + "epoch": 1.6770261366132222, + "grad_norm": 2.2078731060028076, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.729344367980957, + "num_tokens": 389367206.0, + "step": 15271 + }, + { + "epoch": 1.6771359543158357, + "grad_norm": 1.9975687265396118, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7100682258605957, + "num_tokens": 389399193.0, + "step": 15272 + }, + { + "epoch": 1.6772457720184493, + "grad_norm": 2.1858456134796143, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7095123529434204, + "num_tokens": 389427490.0, + "step": 15273 + }, + { + "epoch": 1.677355589721063, + "grad_norm": 2.1115527153015137, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7083839178085327, + "num_tokens": 389456737.0, + "step": 15274 + }, + { + "epoch": 1.6774654074236768, + "grad_norm": 2.0174195766448975, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7077864408493042, + "num_tokens": 389486916.0, + "step": 15275 + }, + { + "epoch": 1.6775752251262903, + "grad_norm": 2.4287073612213135, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7171084880828857, + "num_tokens": 389509235.0, + "step": 15276 + }, + { + "epoch": 1.677685042828904, + "grad_norm": 2.1336891651153564, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7200887203216553, + "num_tokens": 389537967.0, + "step": 15277 + }, + { + "epoch": 1.6777948605315176, + "grad_norm": 2.2431349754333496, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7484756708145142, + "num_tokens": 389562926.0, + "step": 15278 + }, + { + "epoch": 1.6779046782341314, + "grad_norm": 2.4523019790649414, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7184487581253052, + "num_tokens": 389584496.0, + "step": 15279 + }, + { + "epoch": 1.6780144959367451, + "grad_norm": 2.2788009643554688, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7145794630050659, + "num_tokens": 389609737.0, + "step": 15280 + }, + { + "epoch": 1.6781243136393587, + "grad_norm": 2.055088758468628, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7038683891296387, + "num_tokens": 389640407.0, + "step": 15281 + }, + { + "epoch": 1.6782341313419722, + "grad_norm": 2.312807321548462, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7204089760780334, + "num_tokens": 389664350.0, + "step": 15282 + }, + { + "epoch": 1.678343949044586, + "grad_norm": 2.3271498680114746, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7333495616912842, + "num_tokens": 389688045.0, + "step": 15283 + }, + { + "epoch": 1.6784537667471997, + "grad_norm": 2.3463873863220215, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7146505117416382, + "num_tokens": 389712607.0, + "step": 15284 + }, + { + "epoch": 1.6785635844498135, + "grad_norm": 2.404317617416382, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.7660673260688782, + "num_tokens": 389733963.0, + "step": 15285 + }, + { + "epoch": 1.678673402152427, + "grad_norm": 2.379589080810547, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7452706098556519, + "num_tokens": 389756048.0, + "step": 15286 + }, + { + "epoch": 1.6787832198550405, + "grad_norm": 2.1499555110931396, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7094038724899292, + "num_tokens": 389785621.0, + "step": 15287 + }, + { + "epoch": 1.6788930375576543, + "grad_norm": 2.4707493782043457, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.731024980545044, + "num_tokens": 389805795.0, + "step": 15288 + }, + { + "epoch": 1.679002855260268, + "grad_norm": 2.339277744293213, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7061232328414917, + "num_tokens": 389830513.0, + "step": 15289 + }, + { + "epoch": 1.6791126729628816, + "grad_norm": 2.4218997955322266, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7322748303413391, + "num_tokens": 389852646.0, + "step": 15290 + }, + { + "epoch": 1.6792224906654951, + "grad_norm": 2.5251355171203613, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7142969369888306, + "num_tokens": 389874198.0, + "step": 15291 + }, + { + "epoch": 1.6793323083681089, + "grad_norm": 2.016141176223755, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7435564398765564, + "num_tokens": 389902426.0, + "step": 15292 + }, + { + "epoch": 1.6794421260707226, + "grad_norm": 2.3527774810791016, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7102203369140625, + "num_tokens": 389928412.0, + "step": 15293 + }, + { + "epoch": 1.6795519437733364, + "grad_norm": 2.235297679901123, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7140269875526428, + "num_tokens": 389954730.0, + "step": 15294 + }, + { + "epoch": 1.67966176147595, + "grad_norm": 2.24678897857666, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7158143520355225, + "num_tokens": 389980091.0, + "step": 15295 + }, + { + "epoch": 1.6797715791785635, + "grad_norm": 2.270723342895508, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7101323008537292, + "num_tokens": 390004773.0, + "step": 15296 + }, + { + "epoch": 1.6798813968811772, + "grad_norm": 2.3851869106292725, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7483268976211548, + "num_tokens": 390025984.0, + "step": 15297 + }, + { + "epoch": 1.679991214583791, + "grad_norm": 2.1625311374664307, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.6975622177124023, + "num_tokens": 390053425.0, + "step": 15298 + }, + { + "epoch": 1.6801010322864047, + "grad_norm": 2.392183780670166, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7109202146530151, + "num_tokens": 390077203.0, + "step": 15299 + }, + { + "epoch": 1.6802108499890183, + "grad_norm": 2.0457236766815186, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7434592247009277, + "num_tokens": 390106103.0, + "step": 15300 + }, + { + "epoch": 1.6803206676916318, + "grad_norm": 2.3268136978149414, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6982837319374084, + "num_tokens": 390134022.0, + "step": 15301 + }, + { + "epoch": 1.6804304853942456, + "grad_norm": 2.0543625354766846, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6891248226165771, + "num_tokens": 390169832.0, + "step": 15302 + }, + { + "epoch": 1.6805403030968593, + "grad_norm": 2.142764091491699, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7209802269935608, + "num_tokens": 390198132.0, + "step": 15303 + }, + { + "epoch": 1.6806501207994728, + "grad_norm": 2.2134954929351807, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.732418954372406, + "num_tokens": 390226421.0, + "step": 15304 + }, + { + "epoch": 1.6807599385020864, + "grad_norm": 2.345597505569458, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7148059010505676, + "num_tokens": 390248765.0, + "step": 15305 + }, + { + "epoch": 1.6808697562047001, + "grad_norm": 2.073594331741333, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.6997789144515991, + "num_tokens": 390278556.0, + "step": 15306 + }, + { + "epoch": 1.680979573907314, + "grad_norm": 1.9481216669082642, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7113348245620728, + "num_tokens": 390310840.0, + "step": 15307 + }, + { + "epoch": 1.6810893916099277, + "grad_norm": 2.386829137802124, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7303478121757507, + "num_tokens": 390331942.0, + "step": 15308 + }, + { + "epoch": 1.6811992093125412, + "grad_norm": 2.5745255947113037, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7380558252334595, + "num_tokens": 390351581.0, + "step": 15309 + }, + { + "epoch": 1.6813090270151547, + "grad_norm": 2.138066291809082, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.69781494140625, + "num_tokens": 390382003.0, + "step": 15310 + }, + { + "epoch": 1.6814188447177685, + "grad_norm": 2.1582484245300293, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7019859552383423, + "num_tokens": 390408280.0, + "step": 15311 + }, + { + "epoch": 1.6815286624203822, + "grad_norm": 2.1214969158172607, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7097320556640625, + "num_tokens": 390438148.0, + "step": 15312 + }, + { + "epoch": 1.681638480122996, + "grad_norm": 2.541727066040039, + "learning_rate": 1e-06, + "loss": 0.7671, + "mean_token_accuracy": 0.7508569359779358, + "num_tokens": 390459459.0, + "step": 15313 + }, + { + "epoch": 1.6817482978256095, + "grad_norm": 2.8164222240448, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7260910272598267, + "num_tokens": 390476669.0, + "step": 15314 + }, + { + "epoch": 1.681858115528223, + "grad_norm": 2.614933967590332, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7348228693008423, + "num_tokens": 390495842.0, + "step": 15315 + }, + { + "epoch": 1.6819679332308368, + "grad_norm": 2.556500196456909, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7303276658058167, + "num_tokens": 390518173.0, + "step": 15316 + }, + { + "epoch": 1.6820777509334506, + "grad_norm": 2.3899431228637695, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7352769374847412, + "num_tokens": 390541491.0, + "step": 15317 + }, + { + "epoch": 1.6821875686360641, + "grad_norm": 2.0185840129852295, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7126705646514893, + "num_tokens": 390573403.0, + "step": 15318 + }, + { + "epoch": 1.6822973863386776, + "grad_norm": 2.2286479473114014, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6818375587463379, + "num_tokens": 390601562.0, + "step": 15319 + }, + { + "epoch": 1.6824072040412914, + "grad_norm": 2.47841739654541, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7084751129150391, + "num_tokens": 390624864.0, + "step": 15320 + }, + { + "epoch": 1.6825170217439052, + "grad_norm": 2.103376865386963, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7326860427856445, + "num_tokens": 390652985.0, + "step": 15321 + }, + { + "epoch": 1.682626839446519, + "grad_norm": 2.113013744354248, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7163535356521606, + "num_tokens": 390683241.0, + "step": 15322 + }, + { + "epoch": 1.6827366571491325, + "grad_norm": 2.490539312362671, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.7378685474395752, + "num_tokens": 390703752.0, + "step": 15323 + }, + { + "epoch": 1.682846474851746, + "grad_norm": 2.1490917205810547, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7144049406051636, + "num_tokens": 390729953.0, + "step": 15324 + }, + { + "epoch": 1.6829562925543597, + "grad_norm": 2.6178109645843506, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7666887044906616, + "num_tokens": 390749054.0, + "step": 15325 + }, + { + "epoch": 1.6830661102569735, + "grad_norm": 2.3854217529296875, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.731658399105072, + "num_tokens": 390771512.0, + "step": 15326 + }, + { + "epoch": 1.683175927959587, + "grad_norm": 2.1625583171844482, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.694579541683197, + "num_tokens": 390801907.0, + "step": 15327 + }, + { + "epoch": 1.6832857456622008, + "grad_norm": 2.6089060306549072, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7205076217651367, + "num_tokens": 390821587.0, + "step": 15328 + }, + { + "epoch": 1.6833955633648143, + "grad_norm": 2.4750566482543945, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7279573678970337, + "num_tokens": 390842243.0, + "step": 15329 + }, + { + "epoch": 1.683505381067428, + "grad_norm": 2.425187349319458, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7462108731269836, + "num_tokens": 390862838.0, + "step": 15330 + }, + { + "epoch": 1.6836151987700418, + "grad_norm": 2.323848247528076, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.733025312423706, + "num_tokens": 390885650.0, + "step": 15331 + }, + { + "epoch": 1.6837250164726554, + "grad_norm": 2.061490058898926, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7258008718490601, + "num_tokens": 390913798.0, + "step": 15332 + }, + { + "epoch": 1.683834834175269, + "grad_norm": 2.267887592315674, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7088617086410522, + "num_tokens": 390939542.0, + "step": 15333 + }, + { + "epoch": 1.6839446518778827, + "grad_norm": 2.235548496246338, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.735615611076355, + "num_tokens": 390962891.0, + "step": 15334 + }, + { + "epoch": 1.6840544695804964, + "grad_norm": 1.8057684898376465, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7204053401947021, + "num_tokens": 391000166.0, + "step": 15335 + }, + { + "epoch": 1.6841642872831102, + "grad_norm": 2.4440956115722656, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7174073457717896, + "num_tokens": 391024266.0, + "step": 15336 + }, + { + "epoch": 1.6842741049857237, + "grad_norm": 2.091187000274658, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7249779105186462, + "num_tokens": 391053103.0, + "step": 15337 + }, + { + "epoch": 1.6843839226883373, + "grad_norm": 2.062293767929077, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7301080822944641, + "num_tokens": 391081157.0, + "step": 15338 + }, + { + "epoch": 1.684493740390951, + "grad_norm": 2.180325508117676, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.710181474685669, + "num_tokens": 391109094.0, + "step": 15339 + }, + { + "epoch": 1.6846035580935648, + "grad_norm": 2.251322031021118, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7601812481880188, + "num_tokens": 391133595.0, + "step": 15340 + }, + { + "epoch": 1.6847133757961783, + "grad_norm": 2.4937000274658203, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7561070919036865, + "num_tokens": 391154912.0, + "step": 15341 + }, + { + "epoch": 1.684823193498792, + "grad_norm": 2.4026427268981934, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.717899739742279, + "num_tokens": 391179218.0, + "step": 15342 + }, + { + "epoch": 1.6849330112014056, + "grad_norm": 2.199432373046875, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7381153702735901, + "num_tokens": 391204796.0, + "step": 15343 + }, + { + "epoch": 1.6850428289040194, + "grad_norm": 2.6087329387664795, + "learning_rate": 1e-06, + "loss": 0.7478, + "mean_token_accuracy": 0.7530015110969543, + "num_tokens": 391223143.0, + "step": 15344 + }, + { + "epoch": 1.685152646606633, + "grad_norm": 2.440805673599243, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.755776047706604, + "num_tokens": 391243994.0, + "step": 15345 + }, + { + "epoch": 1.6852624643092466, + "grad_norm": 2.1580662727355957, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7104772329330444, + "num_tokens": 391272905.0, + "step": 15346 + }, + { + "epoch": 1.6853722820118602, + "grad_norm": 2.1890687942504883, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.743527889251709, + "num_tokens": 391298701.0, + "step": 15347 + }, + { + "epoch": 1.685482099714474, + "grad_norm": 2.151374340057373, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.726834237575531, + "num_tokens": 391326067.0, + "step": 15348 + }, + { + "epoch": 1.6855919174170877, + "grad_norm": 2.4564216136932373, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7041143178939819, + "num_tokens": 391349167.0, + "step": 15349 + }, + { + "epoch": 1.6857017351197014, + "grad_norm": 2.2109017372131348, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7404118776321411, + "num_tokens": 391375158.0, + "step": 15350 + }, + { + "epoch": 1.685811552822315, + "grad_norm": 2.120522975921631, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7124472856521606, + "num_tokens": 391403124.0, + "step": 15351 + }, + { + "epoch": 1.6859213705249285, + "grad_norm": 2.2968945503234863, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7148669958114624, + "num_tokens": 391428135.0, + "step": 15352 + }, + { + "epoch": 1.6860311882275423, + "grad_norm": 2.103463888168335, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7075579762458801, + "num_tokens": 391457214.0, + "step": 15353 + }, + { + "epoch": 1.686141005930156, + "grad_norm": 2.0196774005889893, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7009227275848389, + "num_tokens": 391487634.0, + "step": 15354 + }, + { + "epoch": 1.6862508236327696, + "grad_norm": 2.2120282649993896, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7211638689041138, + "num_tokens": 391515392.0, + "step": 15355 + }, + { + "epoch": 1.686360641335383, + "grad_norm": 2.405059337615967, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7231723666191101, + "num_tokens": 391536961.0, + "step": 15356 + }, + { + "epoch": 1.6864704590379969, + "grad_norm": 2.8561298847198486, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7395563721656799, + "num_tokens": 391554241.0, + "step": 15357 + }, + { + "epoch": 1.6865802767406106, + "grad_norm": 2.0653505325317383, + "learning_rate": 1e-06, + "loss": 1.0379, + "mean_token_accuracy": 0.6827419996261597, + "num_tokens": 391586294.0, + "step": 15358 + }, + { + "epoch": 1.6866900944432244, + "grad_norm": 2.395815849304199, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7320713996887207, + "num_tokens": 391608138.0, + "step": 15359 + }, + { + "epoch": 1.686799912145838, + "grad_norm": 2.9006948471069336, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7239409685134888, + "num_tokens": 391626384.0, + "step": 15360 + }, + { + "epoch": 1.6869097298484514, + "grad_norm": 2.3938674926757812, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7245824337005615, + "num_tokens": 391649823.0, + "step": 15361 + }, + { + "epoch": 1.6870195475510652, + "grad_norm": 2.1769332885742188, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7139905095100403, + "num_tokens": 391676927.0, + "step": 15362 + }, + { + "epoch": 1.687129365253679, + "grad_norm": 2.3035097122192383, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.704871416091919, + "num_tokens": 391702718.0, + "step": 15363 + }, + { + "epoch": 1.6872391829562927, + "grad_norm": 2.135864734649658, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7216432690620422, + "num_tokens": 391730449.0, + "step": 15364 + }, + { + "epoch": 1.6873490006589063, + "grad_norm": 2.4254534244537354, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7338266372680664, + "num_tokens": 391754884.0, + "step": 15365 + }, + { + "epoch": 1.6874588183615198, + "grad_norm": 2.0688862800598145, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.7038483619689941, + "num_tokens": 391786763.0, + "step": 15366 + }, + { + "epoch": 1.6875686360641335, + "grad_norm": 2.2967827320098877, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7051666378974915, + "num_tokens": 391811921.0, + "step": 15367 + }, + { + "epoch": 1.6876784537667473, + "grad_norm": 2.01094126701355, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7271121740341187, + "num_tokens": 391840825.0, + "step": 15368 + }, + { + "epoch": 1.6877882714693608, + "grad_norm": 2.3741302490234375, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7641544342041016, + "num_tokens": 391862393.0, + "step": 15369 + }, + { + "epoch": 1.6878980891719744, + "grad_norm": 2.3773326873779297, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7186295986175537, + "num_tokens": 391885623.0, + "step": 15370 + }, + { + "epoch": 1.6880079068745881, + "grad_norm": 2.2861669063568115, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7050995230674744, + "num_tokens": 391909747.0, + "step": 15371 + }, + { + "epoch": 1.6881177245772019, + "grad_norm": 2.626711130142212, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7369985580444336, + "num_tokens": 391929453.0, + "step": 15372 + }, + { + "epoch": 1.6882275422798156, + "grad_norm": 2.004160165786743, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7137835025787354, + "num_tokens": 391960339.0, + "step": 15373 + }, + { + "epoch": 1.6883373599824292, + "grad_norm": 2.0559098720550537, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7177435159683228, + "num_tokens": 391990122.0, + "step": 15374 + }, + { + "epoch": 1.6884471776850427, + "grad_norm": 2.168264389038086, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7049952745437622, + "num_tokens": 392017102.0, + "step": 15375 + }, + { + "epoch": 1.6885569953876565, + "grad_norm": 2.554107904434204, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7228250503540039, + "num_tokens": 392038864.0, + "step": 15376 + }, + { + "epoch": 1.6886668130902702, + "grad_norm": 2.251492500305176, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7149043679237366, + "num_tokens": 392065455.0, + "step": 15377 + }, + { + "epoch": 1.6887766307928838, + "grad_norm": 2.123567819595337, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7032877802848816, + "num_tokens": 392096443.0, + "step": 15378 + }, + { + "epoch": 1.6888864484954975, + "grad_norm": 2.3658173084259033, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.719395637512207, + "num_tokens": 392121848.0, + "step": 15379 + }, + { + "epoch": 1.688996266198111, + "grad_norm": 2.5844509601593018, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7336656451225281, + "num_tokens": 392142212.0, + "step": 15380 + }, + { + "epoch": 1.6891060839007248, + "grad_norm": 2.186570882797241, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7245814204216003, + "num_tokens": 392170684.0, + "step": 15381 + }, + { + "epoch": 1.6892159016033386, + "grad_norm": 2.098212480545044, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7126262784004211, + "num_tokens": 392197822.0, + "step": 15382 + }, + { + "epoch": 1.689325719305952, + "grad_norm": 1.8911024332046509, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7017332315444946, + "num_tokens": 392234571.0, + "step": 15383 + }, + { + "epoch": 1.6894355370085656, + "grad_norm": 2.1265604496002197, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7138246297836304, + "num_tokens": 392262165.0, + "step": 15384 + }, + { + "epoch": 1.6895453547111794, + "grad_norm": 2.3192875385284424, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7058231830596924, + "num_tokens": 392289433.0, + "step": 15385 + }, + { + "epoch": 1.6896551724137931, + "grad_norm": 2.361168622970581, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7012379169464111, + "num_tokens": 392313284.0, + "step": 15386 + }, + { + "epoch": 1.689764990116407, + "grad_norm": 2.185551643371582, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7101022005081177, + "num_tokens": 392341668.0, + "step": 15387 + }, + { + "epoch": 1.6898748078190204, + "grad_norm": 2.5261402130126953, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7048792839050293, + "num_tokens": 392363324.0, + "step": 15388 + }, + { + "epoch": 1.689984625521634, + "grad_norm": 2.2034623622894287, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7293990850448608, + "num_tokens": 392389012.0, + "step": 15389 + }, + { + "epoch": 1.6900944432242477, + "grad_norm": 2.36151385307312, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7129015326499939, + "num_tokens": 392412483.0, + "step": 15390 + }, + { + "epoch": 1.6902042609268615, + "grad_norm": 2.2352259159088135, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7271386981010437, + "num_tokens": 392438588.0, + "step": 15391 + }, + { + "epoch": 1.690314078629475, + "grad_norm": 1.8873592615127563, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7244745492935181, + "num_tokens": 392470832.0, + "step": 15392 + }, + { + "epoch": 1.6904238963320888, + "grad_norm": 2.084474802017212, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7150090336799622, + "num_tokens": 392501050.0, + "step": 15393 + }, + { + "epoch": 1.6905337140347023, + "grad_norm": 2.494229793548584, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7309681177139282, + "num_tokens": 392523924.0, + "step": 15394 + }, + { + "epoch": 1.690643531737316, + "grad_norm": 2.471611499786377, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7121446132659912, + "num_tokens": 392546210.0, + "step": 15395 + }, + { + "epoch": 1.6907533494399298, + "grad_norm": 2.322417736053467, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7275655269622803, + "num_tokens": 392570164.0, + "step": 15396 + }, + { + "epoch": 1.6908631671425434, + "grad_norm": 2.1953461170196533, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7037022113800049, + "num_tokens": 392595772.0, + "step": 15397 + }, + { + "epoch": 1.690972984845157, + "grad_norm": 2.0489461421966553, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.694489598274231, + "num_tokens": 392630115.0, + "step": 15398 + }, + { + "epoch": 1.6910828025477707, + "grad_norm": 2.353264570236206, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7207074165344238, + "num_tokens": 392654538.0, + "step": 15399 + }, + { + "epoch": 1.6911926202503844, + "grad_norm": 2.1451356410980225, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7088045477867126, + "num_tokens": 392681803.0, + "step": 15400 + }, + { + "epoch": 1.6913024379529982, + "grad_norm": 2.2469828128814697, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7118019461631775, + "num_tokens": 392708059.0, + "step": 15401 + }, + { + "epoch": 1.6914122556556117, + "grad_norm": 2.211726427078247, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.725095808506012, + "num_tokens": 392734268.0, + "step": 15402 + }, + { + "epoch": 1.6915220733582252, + "grad_norm": 2.2356269359588623, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7188798189163208, + "num_tokens": 392760645.0, + "step": 15403 + }, + { + "epoch": 1.691631891060839, + "grad_norm": 1.9480258226394653, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7011314034461975, + "num_tokens": 392795127.0, + "step": 15404 + }, + { + "epoch": 1.6917417087634528, + "grad_norm": 2.149726390838623, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.71903395652771, + "num_tokens": 392822704.0, + "step": 15405 + }, + { + "epoch": 1.6918515264660663, + "grad_norm": 2.200610876083374, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7181245684623718, + "num_tokens": 392850005.0, + "step": 15406 + }, + { + "epoch": 1.6919613441686798, + "grad_norm": 1.8887375593185425, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.6996957063674927, + "num_tokens": 392884031.0, + "step": 15407 + }, + { + "epoch": 1.6920711618712936, + "grad_norm": 2.41483211517334, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7268195748329163, + "num_tokens": 392906112.0, + "step": 15408 + }, + { + "epoch": 1.6921809795739073, + "grad_norm": 2.2316744327545166, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7099562883377075, + "num_tokens": 392937262.0, + "step": 15409 + }, + { + "epoch": 1.692290797276521, + "grad_norm": 1.9827519655227661, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7244817614555359, + "num_tokens": 392967756.0, + "step": 15410 + }, + { + "epoch": 1.6924006149791346, + "grad_norm": 1.9113259315490723, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7001578211784363, + "num_tokens": 393001763.0, + "step": 15411 + }, + { + "epoch": 1.6925104326817482, + "grad_norm": 2.323063611984253, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7276832461357117, + "num_tokens": 393024821.0, + "step": 15412 + }, + { + "epoch": 1.692620250384362, + "grad_norm": 2.239564895629883, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6998920440673828, + "num_tokens": 393052267.0, + "step": 15413 + }, + { + "epoch": 1.6927300680869757, + "grad_norm": 1.9249135255813599, + "learning_rate": 1e-06, + "loss": 0.8114, + "mean_token_accuracy": 0.745153546333313, + "num_tokens": 393082767.0, + "step": 15414 + }, + { + "epoch": 1.6928398857895894, + "grad_norm": 2.4271509647369385, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7334972620010376, + "num_tokens": 393104565.0, + "step": 15415 + }, + { + "epoch": 1.692949703492203, + "grad_norm": 2.0569636821746826, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7191306948661804, + "num_tokens": 393131168.0, + "step": 15416 + }, + { + "epoch": 1.6930595211948165, + "grad_norm": 2.388265371322632, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7344493865966797, + "num_tokens": 393152386.0, + "step": 15417 + }, + { + "epoch": 1.6931693388974303, + "grad_norm": 2.192574977874756, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7288230061531067, + "num_tokens": 393177449.0, + "step": 15418 + }, + { + "epoch": 1.693279156600044, + "grad_norm": 2.1955747604370117, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7372347116470337, + "num_tokens": 393202697.0, + "step": 15419 + }, + { + "epoch": 1.6933889743026576, + "grad_norm": 2.386420249938965, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7323101758956909, + "num_tokens": 393223704.0, + "step": 15420 + }, + { + "epoch": 1.693498792005271, + "grad_norm": 2.8540961742401123, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7432894706726074, + "num_tokens": 393241055.0, + "step": 15421 + }, + { + "epoch": 1.6936086097078848, + "grad_norm": 1.8692103624343872, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.6999921202659607, + "num_tokens": 393277489.0, + "step": 15422 + }, + { + "epoch": 1.6937184274104986, + "grad_norm": 2.3775932788848877, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7138111591339111, + "num_tokens": 393300671.0, + "step": 15423 + }, + { + "epoch": 1.6938282451131124, + "grad_norm": 2.264711380004883, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7014837861061096, + "num_tokens": 393325860.0, + "step": 15424 + }, + { + "epoch": 1.693938062815726, + "grad_norm": 2.1540262699127197, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7152450680732727, + "num_tokens": 393354734.0, + "step": 15425 + }, + { + "epoch": 1.6940478805183394, + "grad_norm": 2.1827635765075684, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.704552412033081, + "num_tokens": 393381017.0, + "step": 15426 + }, + { + "epoch": 1.6941576982209532, + "grad_norm": 2.3966195583343506, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7349580526351929, + "num_tokens": 393402909.0, + "step": 15427 + }, + { + "epoch": 1.694267515923567, + "grad_norm": 2.460557222366333, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7455153465270996, + "num_tokens": 393426117.0, + "step": 15428 + }, + { + "epoch": 1.6943773336261807, + "grad_norm": 2.46724271774292, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.729697585105896, + "num_tokens": 393448575.0, + "step": 15429 + }, + { + "epoch": 1.6944871513287942, + "grad_norm": 2.1765754222869873, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7178654670715332, + "num_tokens": 393475225.0, + "step": 15430 + }, + { + "epoch": 1.6945969690314078, + "grad_norm": 2.2919211387634277, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.748278796672821, + "num_tokens": 393500217.0, + "step": 15431 + }, + { + "epoch": 1.6947067867340215, + "grad_norm": 2.1816749572753906, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7238147854804993, + "num_tokens": 393528373.0, + "step": 15432 + }, + { + "epoch": 1.6948166044366353, + "grad_norm": 2.1171281337738037, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7213262319564819, + "num_tokens": 393554839.0, + "step": 15433 + }, + { + "epoch": 1.6949264221392488, + "grad_norm": 2.193207025527954, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7269394397735596, + "num_tokens": 393580027.0, + "step": 15434 + }, + { + "epoch": 1.6950362398418624, + "grad_norm": 2.076617956161499, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6982262134552002, + "num_tokens": 393609994.0, + "step": 15435 + }, + { + "epoch": 1.6951460575444761, + "grad_norm": 2.309375047683716, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7103502750396729, + "num_tokens": 393634120.0, + "step": 15436 + }, + { + "epoch": 1.6952558752470899, + "grad_norm": 2.4826061725616455, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7119242548942566, + "num_tokens": 393656606.0, + "step": 15437 + }, + { + "epoch": 1.6953656929497036, + "grad_norm": 2.117077112197876, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.705420732498169, + "num_tokens": 393681592.0, + "step": 15438 + }, + { + "epoch": 1.6954755106523172, + "grad_norm": 2.0608551502227783, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7060750126838684, + "num_tokens": 393712857.0, + "step": 15439 + }, + { + "epoch": 1.6955853283549307, + "grad_norm": 2.1273422241210938, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7193586826324463, + "num_tokens": 393738425.0, + "step": 15440 + }, + { + "epoch": 1.6956951460575445, + "grad_norm": 2.193923234939575, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7143141031265259, + "num_tokens": 393764911.0, + "step": 15441 + }, + { + "epoch": 1.6958049637601582, + "grad_norm": 2.651219367980957, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7374557256698608, + "num_tokens": 393785311.0, + "step": 15442 + }, + { + "epoch": 1.6959147814627717, + "grad_norm": 2.278960704803467, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7247483730316162, + "num_tokens": 393809950.0, + "step": 15443 + }, + { + "epoch": 1.6960245991653855, + "grad_norm": 2.389695405960083, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7180865406990051, + "num_tokens": 393833667.0, + "step": 15444 + }, + { + "epoch": 1.696134416867999, + "grad_norm": 2.542219877243042, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7223953008651733, + "num_tokens": 393854745.0, + "step": 15445 + }, + { + "epoch": 1.6962442345706128, + "grad_norm": 2.2391045093536377, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7593942284584045, + "num_tokens": 393879074.0, + "step": 15446 + }, + { + "epoch": 1.6963540522732266, + "grad_norm": 2.099236011505127, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7206718921661377, + "num_tokens": 393905067.0, + "step": 15447 + }, + { + "epoch": 1.69646386997584, + "grad_norm": 2.288336753845215, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7375931739807129, + "num_tokens": 393928450.0, + "step": 15448 + }, + { + "epoch": 1.6965736876784536, + "grad_norm": 2.255115509033203, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.717023491859436, + "num_tokens": 393953258.0, + "step": 15449 + }, + { + "epoch": 1.6966835053810674, + "grad_norm": 2.0298256874084473, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.736817479133606, + "num_tokens": 393983018.0, + "step": 15450 + }, + { + "epoch": 1.6967933230836811, + "grad_norm": 2.12947416305542, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7295469641685486, + "num_tokens": 394008701.0, + "step": 15451 + }, + { + "epoch": 1.696903140786295, + "grad_norm": 2.4654111862182617, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.7547175288200378, + "num_tokens": 394029097.0, + "step": 15452 + }, + { + "epoch": 1.6970129584889084, + "grad_norm": 1.9975489377975464, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7050708532333374, + "num_tokens": 394059991.0, + "step": 15453 + }, + { + "epoch": 1.697122776191522, + "grad_norm": 2.6313085556030273, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7281224727630615, + "num_tokens": 394084678.0, + "step": 15454 + }, + { + "epoch": 1.6972325938941357, + "grad_norm": 2.292205572128296, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7257269024848938, + "num_tokens": 394111159.0, + "step": 15455 + }, + { + "epoch": 1.6973424115967495, + "grad_norm": 2.439392566680908, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7209188938140869, + "num_tokens": 394133576.0, + "step": 15456 + }, + { + "epoch": 1.697452229299363, + "grad_norm": 2.6103360652923584, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7388792634010315, + "num_tokens": 394153437.0, + "step": 15457 + }, + { + "epoch": 1.6975620470019768, + "grad_norm": 2.0855355262756348, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7079737186431885, + "num_tokens": 394182155.0, + "step": 15458 + }, + { + "epoch": 1.6976718647045903, + "grad_norm": 2.419379472732544, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7033671140670776, + "num_tokens": 394204777.0, + "step": 15459 + }, + { + "epoch": 1.697781682407204, + "grad_norm": 2.4214980602264404, + "learning_rate": 1e-06, + "loss": 0.8176, + "mean_token_accuracy": 0.7390142679214478, + "num_tokens": 394226777.0, + "step": 15460 + }, + { + "epoch": 1.6978915001098178, + "grad_norm": 2.4823787212371826, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7312387824058533, + "num_tokens": 394249415.0, + "step": 15461 + }, + { + "epoch": 1.6980013178124314, + "grad_norm": 2.239159107208252, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7136808633804321, + "num_tokens": 394274714.0, + "step": 15462 + }, + { + "epoch": 1.698111135515045, + "grad_norm": 2.5732107162475586, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7241038680076599, + "num_tokens": 394297614.0, + "step": 15463 + }, + { + "epoch": 1.6982209532176586, + "grad_norm": 1.8745806217193604, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7125579714775085, + "num_tokens": 394329629.0, + "step": 15464 + }, + { + "epoch": 1.6983307709202724, + "grad_norm": 2.3223111629486084, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7213465571403503, + "num_tokens": 394352112.0, + "step": 15465 + }, + { + "epoch": 1.6984405886228862, + "grad_norm": 2.29343843460083, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7126774787902832, + "num_tokens": 394377131.0, + "step": 15466 + }, + { + "epoch": 1.6985504063254997, + "grad_norm": 2.316495418548584, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7089737057685852, + "num_tokens": 394402129.0, + "step": 15467 + }, + { + "epoch": 1.6986602240281132, + "grad_norm": 2.2959699630737305, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7265350818634033, + "num_tokens": 394428895.0, + "step": 15468 + }, + { + "epoch": 1.698770041730727, + "grad_norm": 1.9090145826339722, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.719704270362854, + "num_tokens": 394462638.0, + "step": 15469 + }, + { + "epoch": 1.6988798594333407, + "grad_norm": 1.9401848316192627, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7110098004341125, + "num_tokens": 394493660.0, + "step": 15470 + }, + { + "epoch": 1.6989896771359543, + "grad_norm": 2.364473581314087, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7115775942802429, + "num_tokens": 394517358.0, + "step": 15471 + }, + { + "epoch": 1.6990994948385678, + "grad_norm": 2.0198426246643066, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7075937986373901, + "num_tokens": 394546909.0, + "step": 15472 + }, + { + "epoch": 1.6992093125411816, + "grad_norm": 2.20578670501709, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.70671546459198, + "num_tokens": 394574372.0, + "step": 15473 + }, + { + "epoch": 1.6993191302437953, + "grad_norm": 2.781735420227051, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.714012622833252, + "num_tokens": 394592295.0, + "step": 15474 + }, + { + "epoch": 1.699428947946409, + "grad_norm": 2.349038600921631, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7410708665847778, + "num_tokens": 394614604.0, + "step": 15475 + }, + { + "epoch": 1.6995387656490226, + "grad_norm": 2.1643190383911133, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7333216667175293, + "num_tokens": 394639790.0, + "step": 15476 + }, + { + "epoch": 1.6996485833516362, + "grad_norm": 2.3954265117645264, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7056130170822144, + "num_tokens": 394664382.0, + "step": 15477 + }, + { + "epoch": 1.69975840105425, + "grad_norm": 2.353299617767334, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7220175862312317, + "num_tokens": 394685775.0, + "step": 15478 + }, + { + "epoch": 1.6998682187568637, + "grad_norm": 2.5054714679718018, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7386542558670044, + "num_tokens": 394706010.0, + "step": 15479 + }, + { + "epoch": 1.6999780364594774, + "grad_norm": 2.425405502319336, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7190280556678772, + "num_tokens": 394729150.0, + "step": 15480 + }, + { + "epoch": 1.700087854162091, + "grad_norm": 2.614220380783081, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.7478287220001221, + "num_tokens": 394748115.0, + "step": 15481 + }, + { + "epoch": 1.7001976718647045, + "grad_norm": 2.188737392425537, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7161329388618469, + "num_tokens": 394774833.0, + "step": 15482 + }, + { + "epoch": 1.7003074895673183, + "grad_norm": 2.3710672855377197, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7374987006187439, + "num_tokens": 394797136.0, + "step": 15483 + }, + { + "epoch": 1.700417307269932, + "grad_norm": 2.145289659500122, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.709392786026001, + "num_tokens": 394824768.0, + "step": 15484 + }, + { + "epoch": 1.7005271249725455, + "grad_norm": 1.9307405948638916, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7182966470718384, + "num_tokens": 394855240.0, + "step": 15485 + }, + { + "epoch": 1.700636942675159, + "grad_norm": 1.9518156051635742, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7053848505020142, + "num_tokens": 394888758.0, + "step": 15486 + }, + { + "epoch": 1.7007467603777728, + "grad_norm": 2.301748514175415, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7241441011428833, + "num_tokens": 394913979.0, + "step": 15487 + }, + { + "epoch": 1.7008565780803866, + "grad_norm": 2.4270100593566895, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7261700630187988, + "num_tokens": 394936713.0, + "step": 15488 + }, + { + "epoch": 1.7009663957830004, + "grad_norm": 2.340200662612915, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.731347918510437, + "num_tokens": 394959976.0, + "step": 15489 + }, + { + "epoch": 1.7010762134856139, + "grad_norm": 2.4161272048950195, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7311882972717285, + "num_tokens": 394983097.0, + "step": 15490 + }, + { + "epoch": 1.7011860311882274, + "grad_norm": 2.0261311531066895, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6950790882110596, + "num_tokens": 395016713.0, + "step": 15491 + }, + { + "epoch": 1.7012958488908412, + "grad_norm": 2.157975196838379, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7267223000526428, + "num_tokens": 395040987.0, + "step": 15492 + }, + { + "epoch": 1.701405666593455, + "grad_norm": 2.166776418685913, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7353442907333374, + "num_tokens": 395065506.0, + "step": 15493 + }, + { + "epoch": 1.7015154842960687, + "grad_norm": 2.1954147815704346, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7330437302589417, + "num_tokens": 395093426.0, + "step": 15494 + }, + { + "epoch": 1.7016253019986822, + "grad_norm": 2.186267137527466, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7110030651092529, + "num_tokens": 395120940.0, + "step": 15495 + }, + { + "epoch": 1.7017351197012958, + "grad_norm": 2.35871958732605, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7136930227279663, + "num_tokens": 395145671.0, + "step": 15496 + }, + { + "epoch": 1.7018449374039095, + "grad_norm": 2.2207565307617188, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7278380393981934, + "num_tokens": 395170833.0, + "step": 15497 + }, + { + "epoch": 1.7019547551065233, + "grad_norm": 2.1692378520965576, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7208521366119385, + "num_tokens": 395196928.0, + "step": 15498 + }, + { + "epoch": 1.7020645728091368, + "grad_norm": 2.2949371337890625, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7538906335830688, + "num_tokens": 395220930.0, + "step": 15499 + }, + { + "epoch": 1.7021743905117503, + "grad_norm": 2.0894110202789307, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.6984094977378845, + "num_tokens": 395248264.0, + "step": 15500 + }, + { + "epoch": 1.702284208214364, + "grad_norm": 2.2437140941619873, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7171359062194824, + "num_tokens": 395273277.0, + "step": 15501 + }, + { + "epoch": 1.7023940259169779, + "grad_norm": 2.2758569717407227, + "learning_rate": 1e-06, + "loss": 0.8275, + "mean_token_accuracy": 0.744720458984375, + "num_tokens": 395296828.0, + "step": 15502 + }, + { + "epoch": 1.7025038436195916, + "grad_norm": 2.0435245037078857, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6885887384414673, + "num_tokens": 395328202.0, + "step": 15503 + }, + { + "epoch": 1.7026136613222052, + "grad_norm": 2.3619115352630615, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7140846848487854, + "num_tokens": 395353513.0, + "step": 15504 + }, + { + "epoch": 1.7027234790248187, + "grad_norm": 2.6971914768218994, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7294197082519531, + "num_tokens": 395372913.0, + "step": 15505 + }, + { + "epoch": 1.7028332967274324, + "grad_norm": 2.6121394634246826, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7272518277168274, + "num_tokens": 395393975.0, + "step": 15506 + }, + { + "epoch": 1.7029431144300462, + "grad_norm": 2.0659372806549072, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7139496803283691, + "num_tokens": 395423439.0, + "step": 15507 + }, + { + "epoch": 1.7030529321326597, + "grad_norm": 2.2761001586914062, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7165390253067017, + "num_tokens": 395450013.0, + "step": 15508 + }, + { + "epoch": 1.7031627498352735, + "grad_norm": 2.416844606399536, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7477030158042908, + "num_tokens": 395472794.0, + "step": 15509 + }, + { + "epoch": 1.703272567537887, + "grad_norm": 2.251946210861206, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.743682324886322, + "num_tokens": 395496563.0, + "step": 15510 + }, + { + "epoch": 1.7033823852405008, + "grad_norm": 1.8591140508651733, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.71502685546875, + "num_tokens": 395532019.0, + "step": 15511 + }, + { + "epoch": 1.7034922029431145, + "grad_norm": 2.468613386154175, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7340474128723145, + "num_tokens": 395555022.0, + "step": 15512 + }, + { + "epoch": 1.703602020645728, + "grad_norm": 2.6822996139526367, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7366505861282349, + "num_tokens": 395572718.0, + "step": 15513 + }, + { + "epoch": 1.7037118383483416, + "grad_norm": 2.3461289405822754, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.725823700428009, + "num_tokens": 395598117.0, + "step": 15514 + }, + { + "epoch": 1.7038216560509554, + "grad_norm": 2.271151304244995, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7343809008598328, + "num_tokens": 395621820.0, + "step": 15515 + }, + { + "epoch": 1.7039314737535691, + "grad_norm": 2.2087666988372803, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7062748074531555, + "num_tokens": 395647320.0, + "step": 15516 + }, + { + "epoch": 1.7040412914561829, + "grad_norm": 2.342489242553711, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7223303914070129, + "num_tokens": 395669011.0, + "step": 15517 + }, + { + "epoch": 1.7041511091587964, + "grad_norm": 2.3290023803710938, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7153564095497131, + "num_tokens": 395693803.0, + "step": 15518 + }, + { + "epoch": 1.70426092686141, + "grad_norm": 2.6065454483032227, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7337382435798645, + "num_tokens": 395712814.0, + "step": 15519 + }, + { + "epoch": 1.7043707445640237, + "grad_norm": 2.2839460372924805, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7136166095733643, + "num_tokens": 395739062.0, + "step": 15520 + }, + { + "epoch": 1.7044805622666375, + "grad_norm": 2.301872491836548, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7181886434555054, + "num_tokens": 395762851.0, + "step": 15521 + }, + { + "epoch": 1.704590379969251, + "grad_norm": 2.1699397563934326, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.724990963935852, + "num_tokens": 395787880.0, + "step": 15522 + }, + { + "epoch": 1.7047001976718648, + "grad_norm": 2.1393659114837646, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7109959125518799, + "num_tokens": 395817230.0, + "step": 15523 + }, + { + "epoch": 1.7048100153744783, + "grad_norm": 2.377678155899048, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7104690670967102, + "num_tokens": 395842049.0, + "step": 15524 + }, + { + "epoch": 1.704919833077092, + "grad_norm": 2.017066240310669, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.718549370765686, + "num_tokens": 395874478.0, + "step": 15525 + }, + { + "epoch": 1.7050296507797058, + "grad_norm": 2.499854326248169, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7242186069488525, + "num_tokens": 395895225.0, + "step": 15526 + }, + { + "epoch": 1.7051394684823193, + "grad_norm": 2.3133039474487305, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7152228355407715, + "num_tokens": 395920166.0, + "step": 15527 + }, + { + "epoch": 1.7052492861849329, + "grad_norm": 2.3052659034729004, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.731348991394043, + "num_tokens": 395942671.0, + "step": 15528 + }, + { + "epoch": 1.7053591038875466, + "grad_norm": 2.402425527572632, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7127437591552734, + "num_tokens": 395964059.0, + "step": 15529 + }, + { + "epoch": 1.7054689215901604, + "grad_norm": 2.2079310417175293, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7176190614700317, + "num_tokens": 395990450.0, + "step": 15530 + }, + { + "epoch": 1.7055787392927741, + "grad_norm": 2.368744373321533, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7267366647720337, + "num_tokens": 396013621.0, + "step": 15531 + }, + { + "epoch": 1.7056885569953877, + "grad_norm": 1.961429476737976, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7111844420433044, + "num_tokens": 396047691.0, + "step": 15532 + }, + { + "epoch": 1.7057983746980012, + "grad_norm": 2.2729837894439697, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7184500694274902, + "num_tokens": 396072716.0, + "step": 15533 + }, + { + "epoch": 1.705908192400615, + "grad_norm": 2.2083330154418945, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7140206098556519, + "num_tokens": 396099227.0, + "step": 15534 + }, + { + "epoch": 1.7060180101032287, + "grad_norm": 2.367835283279419, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7337465286254883, + "num_tokens": 396123860.0, + "step": 15535 + }, + { + "epoch": 1.7061278278058423, + "grad_norm": 2.087040662765503, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.6920053958892822, + "num_tokens": 396155832.0, + "step": 15536 + }, + { + "epoch": 1.7062376455084558, + "grad_norm": 2.5422375202178955, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7313601970672607, + "num_tokens": 396175709.0, + "step": 15537 + }, + { + "epoch": 1.7063474632110696, + "grad_norm": 2.377368688583374, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.724320650100708, + "num_tokens": 396199633.0, + "step": 15538 + }, + { + "epoch": 1.7064572809136833, + "grad_norm": 2.3340537548065186, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7211683988571167, + "num_tokens": 396223561.0, + "step": 15539 + }, + { + "epoch": 1.706567098616297, + "grad_norm": 2.616274118423462, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7194344401359558, + "num_tokens": 396243723.0, + "step": 15540 + }, + { + "epoch": 1.7066769163189106, + "grad_norm": 2.052572011947632, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7343268990516663, + "num_tokens": 396275334.0, + "step": 15541 + }, + { + "epoch": 1.7067867340215241, + "grad_norm": 2.09372878074646, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7224318981170654, + "num_tokens": 396305266.0, + "step": 15542 + }, + { + "epoch": 1.706896551724138, + "grad_norm": 2.3830862045288086, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7325730919837952, + "num_tokens": 396327417.0, + "step": 15543 + }, + { + "epoch": 1.7070063694267517, + "grad_norm": 2.325469970703125, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.704959511756897, + "num_tokens": 396352874.0, + "step": 15544 + }, + { + "epoch": 1.7071161871293654, + "grad_norm": 2.4899399280548096, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7215093374252319, + "num_tokens": 396375202.0, + "step": 15545 + }, + { + "epoch": 1.707226004831979, + "grad_norm": 2.436993360519409, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.722443699836731, + "num_tokens": 396396432.0, + "step": 15546 + }, + { + "epoch": 1.7073358225345925, + "grad_norm": 2.017003059387207, + "learning_rate": 1e-06, + "loss": 0.8128, + "mean_token_accuracy": 0.7414494156837463, + "num_tokens": 396422284.0, + "step": 15547 + }, + { + "epoch": 1.7074456402372062, + "grad_norm": 2.2801876068115234, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7292869091033936, + "num_tokens": 396446532.0, + "step": 15548 + }, + { + "epoch": 1.70755545793982, + "grad_norm": 2.1684775352478027, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7024299502372742, + "num_tokens": 396474332.0, + "step": 15549 + }, + { + "epoch": 1.7076652756424335, + "grad_norm": 2.437518835067749, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7394412755966187, + "num_tokens": 396493980.0, + "step": 15550 + }, + { + "epoch": 1.707775093345047, + "grad_norm": 2.5343029499053955, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7349039316177368, + "num_tokens": 396514714.0, + "step": 15551 + }, + { + "epoch": 1.7078849110476608, + "grad_norm": 2.138249397277832, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7186301946640015, + "num_tokens": 396540435.0, + "step": 15552 + }, + { + "epoch": 1.7079947287502746, + "grad_norm": 2.172081470489502, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7118209600448608, + "num_tokens": 396567182.0, + "step": 15553 + }, + { + "epoch": 1.7081045464528883, + "grad_norm": 2.4624481201171875, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7342621684074402, + "num_tokens": 396589289.0, + "step": 15554 + }, + { + "epoch": 1.7082143641555019, + "grad_norm": 2.4128198623657227, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7383084893226624, + "num_tokens": 396613841.0, + "step": 15555 + }, + { + "epoch": 1.7083241818581154, + "grad_norm": 2.4137394428253174, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7272109389305115, + "num_tokens": 396635271.0, + "step": 15556 + }, + { + "epoch": 1.7084339995607292, + "grad_norm": 2.5925440788269043, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7352620363235474, + "num_tokens": 396654097.0, + "step": 15557 + }, + { + "epoch": 1.708543817263343, + "grad_norm": 2.2531380653381348, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.7025483846664429, + "num_tokens": 396679021.0, + "step": 15558 + }, + { + "epoch": 1.7086536349659565, + "grad_norm": 2.24552583694458, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7185406684875488, + "num_tokens": 396705580.0, + "step": 15559 + }, + { + "epoch": 1.7087634526685702, + "grad_norm": 2.3323497772216797, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7343093752861023, + "num_tokens": 396729502.0, + "step": 15560 + }, + { + "epoch": 1.7088732703711838, + "grad_norm": 2.29290771484375, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7217596173286438, + "num_tokens": 396753934.0, + "step": 15561 + }, + { + "epoch": 1.7089830880737975, + "grad_norm": 2.3902108669281006, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7170575261116028, + "num_tokens": 396777163.0, + "step": 15562 + }, + { + "epoch": 1.7090929057764113, + "grad_norm": 2.154128074645996, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7154330015182495, + "num_tokens": 396802552.0, + "step": 15563 + }, + { + "epoch": 1.7092027234790248, + "grad_norm": 1.951818585395813, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7116507887840271, + "num_tokens": 396832690.0, + "step": 15564 + }, + { + "epoch": 1.7093125411816383, + "grad_norm": 2.171330451965332, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7411344647407532, + "num_tokens": 396857766.0, + "step": 15565 + }, + { + "epoch": 1.709422358884252, + "grad_norm": 2.3439602851867676, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7310219407081604, + "num_tokens": 396882003.0, + "step": 15566 + }, + { + "epoch": 1.7095321765868658, + "grad_norm": 1.9848690032958984, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7130901217460632, + "num_tokens": 396914370.0, + "step": 15567 + }, + { + "epoch": 1.7096419942894796, + "grad_norm": 2.323854923248291, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.6987041234970093, + "num_tokens": 396940357.0, + "step": 15568 + }, + { + "epoch": 1.7097518119920931, + "grad_norm": 2.1807518005371094, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.6874156594276428, + "num_tokens": 396966787.0, + "step": 15569 + }, + { + "epoch": 1.7098616296947067, + "grad_norm": 2.47512149810791, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7318123579025269, + "num_tokens": 396989324.0, + "step": 15570 + }, + { + "epoch": 1.7099714473973204, + "grad_norm": 2.590367078781128, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7275288701057434, + "num_tokens": 397009315.0, + "step": 15571 + }, + { + "epoch": 1.7100812650999342, + "grad_norm": 2.125237464904785, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7046762108802795, + "num_tokens": 397037549.0, + "step": 15572 + }, + { + "epoch": 1.7101910828025477, + "grad_norm": 2.3788163661956787, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.740151047706604, + "num_tokens": 397058501.0, + "step": 15573 + }, + { + "epoch": 1.7103009005051615, + "grad_norm": 2.2912018299102783, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7153599858283997, + "num_tokens": 397083958.0, + "step": 15574 + }, + { + "epoch": 1.710410718207775, + "grad_norm": 2.726046562194824, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7228270769119263, + "num_tokens": 397103239.0, + "step": 15575 + }, + { + "epoch": 1.7105205359103888, + "grad_norm": 2.5064682960510254, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7387718558311462, + "num_tokens": 397125368.0, + "step": 15576 + }, + { + "epoch": 1.7106303536130025, + "grad_norm": 2.15775203704834, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7265195846557617, + "num_tokens": 397152724.0, + "step": 15577 + }, + { + "epoch": 1.710740171315616, + "grad_norm": 2.0902135372161865, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7248722314834595, + "num_tokens": 397181451.0, + "step": 15578 + }, + { + "epoch": 1.7108499890182296, + "grad_norm": 2.080697774887085, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7364124059677124, + "num_tokens": 397208173.0, + "step": 15579 + }, + { + "epoch": 1.7109598067208434, + "grad_norm": 2.287731647491455, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7202021479606628, + "num_tokens": 397233723.0, + "step": 15580 + }, + { + "epoch": 1.7110696244234571, + "grad_norm": 2.464278221130371, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7313471436500549, + "num_tokens": 397256747.0, + "step": 15581 + }, + { + "epoch": 1.7111794421260709, + "grad_norm": 2.433851957321167, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7319512367248535, + "num_tokens": 397278703.0, + "step": 15582 + }, + { + "epoch": 1.7112892598286844, + "grad_norm": 2.1969056129455566, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7291557788848877, + "num_tokens": 397304735.0, + "step": 15583 + }, + { + "epoch": 1.711399077531298, + "grad_norm": 2.1951589584350586, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7194216847419739, + "num_tokens": 397331837.0, + "step": 15584 + }, + { + "epoch": 1.7115088952339117, + "grad_norm": 2.3330912590026855, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7109489440917969, + "num_tokens": 397358607.0, + "step": 15585 + }, + { + "epoch": 1.7116187129365255, + "grad_norm": 1.9979596138000488, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7159409523010254, + "num_tokens": 397388178.0, + "step": 15586 + }, + { + "epoch": 1.711728530639139, + "grad_norm": 2.1242198944091797, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7376253604888916, + "num_tokens": 397414456.0, + "step": 15587 + }, + { + "epoch": 1.7118383483417525, + "grad_norm": 2.0492115020751953, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7513754367828369, + "num_tokens": 397440548.0, + "step": 15588 + }, + { + "epoch": 1.7119481660443663, + "grad_norm": 2.3800265789031982, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.687711238861084, + "num_tokens": 397466485.0, + "step": 15589 + }, + { + "epoch": 1.71205798374698, + "grad_norm": 2.0733635425567627, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7205299139022827, + "num_tokens": 397497777.0, + "step": 15590 + }, + { + "epoch": 1.7121678014495938, + "grad_norm": 2.2137484550476074, + "learning_rate": 1e-06, + "loss": 0.7641, + "mean_token_accuracy": 0.7540196776390076, + "num_tokens": 397522022.0, + "step": 15591 + }, + { + "epoch": 1.7122776191522073, + "grad_norm": 2.101928949356079, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7178539037704468, + "num_tokens": 397548687.0, + "step": 15592 + }, + { + "epoch": 1.7123874368548209, + "grad_norm": 2.429849147796631, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7158797979354858, + "num_tokens": 397571972.0, + "step": 15593 + }, + { + "epoch": 1.7124972545574346, + "grad_norm": 2.1917431354522705, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.739517331123352, + "num_tokens": 397596569.0, + "step": 15594 + }, + { + "epoch": 1.7126070722600484, + "grad_norm": 2.3886876106262207, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7343736886978149, + "num_tokens": 397617527.0, + "step": 15595 + }, + { + "epoch": 1.7127168899626621, + "grad_norm": 2.0606789588928223, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7132548689842224, + "num_tokens": 397645838.0, + "step": 15596 + }, + { + "epoch": 1.7128267076652757, + "grad_norm": 2.035850763320923, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7298708558082581, + "num_tokens": 397676047.0, + "step": 15597 + }, + { + "epoch": 1.7129365253678892, + "grad_norm": 2.114800453186035, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.6984916925430298, + "num_tokens": 397706949.0, + "step": 15598 + }, + { + "epoch": 1.713046343070503, + "grad_norm": 2.4009039402008057, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.6961606740951538, + "num_tokens": 397730388.0, + "step": 15599 + }, + { + "epoch": 1.7131561607731167, + "grad_norm": 2.1639997959136963, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7381385564804077, + "num_tokens": 397758373.0, + "step": 15600 + }, + { + "epoch": 1.7132659784757303, + "grad_norm": 2.440087080001831, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7070164680480957, + "num_tokens": 397782263.0, + "step": 15601 + }, + { + "epoch": 1.7133757961783438, + "grad_norm": 2.5687594413757324, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7184840440750122, + "num_tokens": 397808507.0, + "step": 15602 + }, + { + "epoch": 1.7134856138809575, + "grad_norm": 2.2407944202423096, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7164795398712158, + "num_tokens": 397834636.0, + "step": 15603 + }, + { + "epoch": 1.7135954315835713, + "grad_norm": 2.118661403656006, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7257209420204163, + "num_tokens": 397866135.0, + "step": 15604 + }, + { + "epoch": 1.713705249286185, + "grad_norm": 1.7984811067581177, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.698857843875885, + "num_tokens": 397905375.0, + "step": 15605 + }, + { + "epoch": 1.7138150669887986, + "grad_norm": 2.307278871536255, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7123934030532837, + "num_tokens": 397930883.0, + "step": 15606 + }, + { + "epoch": 1.7139248846914121, + "grad_norm": 2.1463072299957275, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7267048358917236, + "num_tokens": 397957399.0, + "step": 15607 + }, + { + "epoch": 1.7140347023940259, + "grad_norm": 2.0905368328094482, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7279713749885559, + "num_tokens": 397983532.0, + "step": 15608 + }, + { + "epoch": 1.7141445200966396, + "grad_norm": 2.30183482170105, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6863446831703186, + "num_tokens": 398009024.0, + "step": 15609 + }, + { + "epoch": 1.7142543377992534, + "grad_norm": 2.254103660583496, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7107120752334595, + "num_tokens": 398035426.0, + "step": 15610 + }, + { + "epoch": 1.714364155501867, + "grad_norm": 2.3780791759490967, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7267606854438782, + "num_tokens": 398058371.0, + "step": 15611 + }, + { + "epoch": 1.7144739732044805, + "grad_norm": 2.510129451751709, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6937934160232544, + "num_tokens": 398080003.0, + "step": 15612 + }, + { + "epoch": 1.7145837909070942, + "grad_norm": 2.217683792114258, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7122739553451538, + "num_tokens": 398106724.0, + "step": 15613 + }, + { + "epoch": 1.714693608609708, + "grad_norm": 2.334998369216919, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7117667198181152, + "num_tokens": 398133068.0, + "step": 15614 + }, + { + "epoch": 1.7148034263123215, + "grad_norm": 2.2421233654022217, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7162798643112183, + "num_tokens": 398160773.0, + "step": 15615 + }, + { + "epoch": 1.714913244014935, + "grad_norm": 2.357133388519287, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7407123446464539, + "num_tokens": 398182884.0, + "step": 15616 + }, + { + "epoch": 1.7150230617175488, + "grad_norm": 2.1127474308013916, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7171547412872314, + "num_tokens": 398212452.0, + "step": 15617 + }, + { + "epoch": 1.7151328794201626, + "grad_norm": 2.0496585369110107, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7027873992919922, + "num_tokens": 398241904.0, + "step": 15618 + }, + { + "epoch": 1.7152426971227763, + "grad_norm": 2.334624767303467, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7346286177635193, + "num_tokens": 398265785.0, + "step": 15619 + }, + { + "epoch": 1.7153525148253899, + "grad_norm": 2.244615316390991, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7371639013290405, + "num_tokens": 398290226.0, + "step": 15620 + }, + { + "epoch": 1.7154623325280034, + "grad_norm": 2.4978110790252686, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.723296582698822, + "num_tokens": 398311235.0, + "step": 15621 + }, + { + "epoch": 1.7155721502306172, + "grad_norm": 1.944891095161438, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7026864290237427, + "num_tokens": 398344941.0, + "step": 15622 + }, + { + "epoch": 1.715681967933231, + "grad_norm": 2.1372179985046387, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7293288707733154, + "num_tokens": 398371379.0, + "step": 15623 + }, + { + "epoch": 1.7157917856358444, + "grad_norm": 2.113070487976074, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7214206457138062, + "num_tokens": 398399585.0, + "step": 15624 + }, + { + "epoch": 1.7159016033384582, + "grad_norm": 1.9633598327636719, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7320987582206726, + "num_tokens": 398428423.0, + "step": 15625 + }, + { + "epoch": 1.7160114210410717, + "grad_norm": 2.2718417644500732, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7122994661331177, + "num_tokens": 398452920.0, + "step": 15626 + }, + { + "epoch": 1.7161212387436855, + "grad_norm": 2.7622954845428467, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7251117825508118, + "num_tokens": 398472827.0, + "step": 15627 + }, + { + "epoch": 1.7162310564462993, + "grad_norm": 2.518066167831421, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7210735082626343, + "num_tokens": 398495446.0, + "step": 15628 + }, + { + "epoch": 1.7163408741489128, + "grad_norm": 2.505781412124634, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7299055457115173, + "num_tokens": 398517842.0, + "step": 15629 + }, + { + "epoch": 1.7164506918515263, + "grad_norm": 2.262709140777588, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7144046425819397, + "num_tokens": 398544776.0, + "step": 15630 + }, + { + "epoch": 1.71656050955414, + "grad_norm": 2.6312856674194336, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7334503531455994, + "num_tokens": 398565326.0, + "step": 15631 + }, + { + "epoch": 1.7166703272567538, + "grad_norm": 1.9736216068267822, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.717483401298523, + "num_tokens": 398593855.0, + "step": 15632 + }, + { + "epoch": 1.7167801449593676, + "grad_norm": 2.123692512512207, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7055873870849609, + "num_tokens": 398622826.0, + "step": 15633 + }, + { + "epoch": 1.7168899626619811, + "grad_norm": 2.3755581378936768, + "learning_rate": 1e-06, + "loss": 0.8176, + "mean_token_accuracy": 0.7344405651092529, + "num_tokens": 398645572.0, + "step": 15634 + }, + { + "epoch": 1.7169997803645947, + "grad_norm": 2.2511565685272217, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7145440578460693, + "num_tokens": 398670928.0, + "step": 15635 + }, + { + "epoch": 1.7171095980672084, + "grad_norm": 2.1037092208862305, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.6985384821891785, + "num_tokens": 398699333.0, + "step": 15636 + }, + { + "epoch": 1.7172194157698222, + "grad_norm": 2.1720786094665527, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7206819653511047, + "num_tokens": 398726434.0, + "step": 15637 + }, + { + "epoch": 1.7173292334724357, + "grad_norm": 2.599647045135498, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7373719215393066, + "num_tokens": 398745511.0, + "step": 15638 + }, + { + "epoch": 1.7174390511750495, + "grad_norm": 2.178231954574585, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7015342712402344, + "num_tokens": 398772291.0, + "step": 15639 + }, + { + "epoch": 1.717548868877663, + "grad_norm": 2.0325095653533936, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6898958086967468, + "num_tokens": 398802181.0, + "step": 15640 + }, + { + "epoch": 1.7176586865802768, + "grad_norm": 2.358806610107422, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7227600812911987, + "num_tokens": 398826589.0, + "step": 15641 + }, + { + "epoch": 1.7177685042828905, + "grad_norm": 2.1070494651794434, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7213566899299622, + "num_tokens": 398854861.0, + "step": 15642 + }, + { + "epoch": 1.717878321985504, + "grad_norm": 2.1586194038391113, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7289313673973083, + "num_tokens": 398882697.0, + "step": 15643 + }, + { + "epoch": 1.7179881396881176, + "grad_norm": 2.498624324798584, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7218586206436157, + "num_tokens": 398903965.0, + "step": 15644 + }, + { + "epoch": 1.7180979573907313, + "grad_norm": 2.02630352973938, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.6993504762649536, + "num_tokens": 398934655.0, + "step": 15645 + }, + { + "epoch": 1.718207775093345, + "grad_norm": 1.9092046022415161, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7183240652084351, + "num_tokens": 398968267.0, + "step": 15646 + }, + { + "epoch": 1.7183175927959589, + "grad_norm": 2.6245431900024414, + "learning_rate": 1e-06, + "loss": 0.8158, + "mean_token_accuracy": 0.7386967539787292, + "num_tokens": 398988247.0, + "step": 15647 + }, + { + "epoch": 1.7184274104985724, + "grad_norm": 2.3299012184143066, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7148784399032593, + "num_tokens": 399011460.0, + "step": 15648 + }, + { + "epoch": 1.718537228201186, + "grad_norm": 2.512443780899048, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.7518976926803589, + "num_tokens": 399031321.0, + "step": 15649 + }, + { + "epoch": 1.7186470459037997, + "grad_norm": 2.356333017349243, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7517842054367065, + "num_tokens": 399054081.0, + "step": 15650 + }, + { + "epoch": 1.7187568636064134, + "grad_norm": 2.3535516262054443, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7120245695114136, + "num_tokens": 399079252.0, + "step": 15651 + }, + { + "epoch": 1.718866681309027, + "grad_norm": 2.371049404144287, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7185304164886475, + "num_tokens": 399101516.0, + "step": 15652 + }, + { + "epoch": 1.7189764990116405, + "grad_norm": 2.2334072589874268, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7074306607246399, + "num_tokens": 399126470.0, + "step": 15653 + }, + { + "epoch": 1.7190863167142543, + "grad_norm": 2.4070887565612793, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7207620143890381, + "num_tokens": 399148310.0, + "step": 15654 + }, + { + "epoch": 1.719196134416868, + "grad_norm": 2.1164886951446533, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7211010456085205, + "num_tokens": 399174934.0, + "step": 15655 + }, + { + "epoch": 1.7193059521194818, + "grad_norm": 2.165703058242798, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7033496499061584, + "num_tokens": 399202900.0, + "step": 15656 + }, + { + "epoch": 1.7194157698220953, + "grad_norm": 2.3569815158843994, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7218965291976929, + "num_tokens": 399227174.0, + "step": 15657 + }, + { + "epoch": 1.7195255875247089, + "grad_norm": 2.248020887374878, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.719347357749939, + "num_tokens": 399253519.0, + "step": 15658 + }, + { + "epoch": 1.7196354052273226, + "grad_norm": 2.1306283473968506, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7112985253334045, + "num_tokens": 399282610.0, + "step": 15659 + }, + { + "epoch": 1.7197452229299364, + "grad_norm": 2.1110177040100098, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7061681151390076, + "num_tokens": 399311098.0, + "step": 15660 + }, + { + "epoch": 1.7198550406325501, + "grad_norm": 2.4252467155456543, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7212185859680176, + "num_tokens": 399334200.0, + "step": 15661 + }, + { + "epoch": 1.7199648583351637, + "grad_norm": 2.685835123062134, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7345743775367737, + "num_tokens": 399353423.0, + "step": 15662 + }, + { + "epoch": 1.7200746760377772, + "grad_norm": 2.3915131092071533, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7294551134109497, + "num_tokens": 399376638.0, + "step": 15663 + }, + { + "epoch": 1.720184493740391, + "grad_norm": 2.10082745552063, + "learning_rate": 1e-06, + "loss": 1.0687, + "mean_token_accuracy": 0.6911491751670837, + "num_tokens": 399408674.0, + "step": 15664 + }, + { + "epoch": 1.7202943114430047, + "grad_norm": 2.3953189849853516, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7156834006309509, + "num_tokens": 399431998.0, + "step": 15665 + }, + { + "epoch": 1.7204041291456182, + "grad_norm": 2.3056044578552246, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7034628391265869, + "num_tokens": 399456681.0, + "step": 15666 + }, + { + "epoch": 1.7205139468482318, + "grad_norm": 1.916700005531311, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.6939377188682556, + "num_tokens": 399490539.0, + "step": 15667 + }, + { + "epoch": 1.7206237645508455, + "grad_norm": 2.6069447994232178, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7268787622451782, + "num_tokens": 399511644.0, + "step": 15668 + }, + { + "epoch": 1.7207335822534593, + "grad_norm": 1.9597045183181763, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7117107510566711, + "num_tokens": 399545322.0, + "step": 15669 + }, + { + "epoch": 1.720843399956073, + "grad_norm": 2.272751569747925, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7247011065483093, + "num_tokens": 399569657.0, + "step": 15670 + }, + { + "epoch": 1.7209532176586866, + "grad_norm": 2.28609299659729, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7174381613731384, + "num_tokens": 399595116.0, + "step": 15671 + }, + { + "epoch": 1.7210630353613001, + "grad_norm": 1.9916430711746216, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7072975039482117, + "num_tokens": 399627147.0, + "step": 15672 + }, + { + "epoch": 1.7211728530639139, + "grad_norm": 2.2604596614837646, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7153198719024658, + "num_tokens": 399653533.0, + "step": 15673 + }, + { + "epoch": 1.7212826707665276, + "grad_norm": 2.2652711868286133, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7079845666885376, + "num_tokens": 399679913.0, + "step": 15674 + }, + { + "epoch": 1.7213924884691414, + "grad_norm": 2.1547935009002686, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.6988049745559692, + "num_tokens": 399709376.0, + "step": 15675 + }, + { + "epoch": 1.721502306171755, + "grad_norm": 2.1842918395996094, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7307215929031372, + "num_tokens": 399737350.0, + "step": 15676 + }, + { + "epoch": 1.7216121238743685, + "grad_norm": 2.2667882442474365, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7424255609512329, + "num_tokens": 399762559.0, + "step": 15677 + }, + { + "epoch": 1.7217219415769822, + "grad_norm": 2.454103946685791, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7216382026672363, + "num_tokens": 399783252.0, + "step": 15678 + }, + { + "epoch": 1.721831759279596, + "grad_norm": 2.4888110160827637, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7372456789016724, + "num_tokens": 399803508.0, + "step": 15679 + }, + { + "epoch": 1.7219415769822095, + "grad_norm": 2.194924831390381, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7234385013580322, + "num_tokens": 399829794.0, + "step": 15680 + }, + { + "epoch": 1.722051394684823, + "grad_norm": 2.0277466773986816, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.7026084661483765, + "num_tokens": 399861345.0, + "step": 15681 + }, + { + "epoch": 1.7221612123874368, + "grad_norm": 2.4697470664978027, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7520483732223511, + "num_tokens": 399881241.0, + "step": 15682 + }, + { + "epoch": 1.7222710300900506, + "grad_norm": 2.0732815265655518, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7133750915527344, + "num_tokens": 399909913.0, + "step": 15683 + }, + { + "epoch": 1.7223808477926643, + "grad_norm": 2.1790101528167725, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7140711545944214, + "num_tokens": 399936768.0, + "step": 15684 + }, + { + "epoch": 1.7224906654952779, + "grad_norm": 2.3739306926727295, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7135815024375916, + "num_tokens": 399959660.0, + "step": 15685 + }, + { + "epoch": 1.7226004831978914, + "grad_norm": 2.4571948051452637, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6941019892692566, + "num_tokens": 399982825.0, + "step": 15686 + }, + { + "epoch": 1.7227103009005051, + "grad_norm": 2.5769855976104736, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7326752543449402, + "num_tokens": 400003533.0, + "step": 15687 + }, + { + "epoch": 1.722820118603119, + "grad_norm": 2.1675004959106445, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7149515151977539, + "num_tokens": 400029307.0, + "step": 15688 + }, + { + "epoch": 1.7229299363057324, + "grad_norm": 2.1052279472351074, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7093694806098938, + "num_tokens": 400058218.0, + "step": 15689 + }, + { + "epoch": 1.7230397540083462, + "grad_norm": 2.196776866912842, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7187748551368713, + "num_tokens": 400085234.0, + "step": 15690 + }, + { + "epoch": 1.7231495717109597, + "grad_norm": 1.9657691717147827, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7183930277824402, + "num_tokens": 400115115.0, + "step": 15691 + }, + { + "epoch": 1.7232593894135735, + "grad_norm": 2.4250810146331787, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.7390269041061401, + "num_tokens": 400138937.0, + "step": 15692 + }, + { + "epoch": 1.7233692071161872, + "grad_norm": 2.3375444412231445, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7340515851974487, + "num_tokens": 400163109.0, + "step": 15693 + }, + { + "epoch": 1.7234790248188008, + "grad_norm": 2.3336679935455322, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7067985534667969, + "num_tokens": 400188399.0, + "step": 15694 + }, + { + "epoch": 1.7235888425214143, + "grad_norm": 1.9237842559814453, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.6938807964324951, + "num_tokens": 400221776.0, + "step": 15695 + }, + { + "epoch": 1.723698660224028, + "grad_norm": 2.2084622383117676, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7213677167892456, + "num_tokens": 400247749.0, + "step": 15696 + }, + { + "epoch": 1.7238084779266418, + "grad_norm": 2.2731802463531494, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7127004265785217, + "num_tokens": 400273083.0, + "step": 15697 + }, + { + "epoch": 1.7239182956292556, + "grad_norm": 2.3301913738250732, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7279192805290222, + "num_tokens": 400298706.0, + "step": 15698 + }, + { + "epoch": 1.7240281133318691, + "grad_norm": 2.1131396293640137, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7307575345039368, + "num_tokens": 400325480.0, + "step": 15699 + }, + { + "epoch": 1.7241379310344827, + "grad_norm": 2.1719422340393066, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7180364727973938, + "num_tokens": 400352201.0, + "step": 15700 + }, + { + "epoch": 1.7242477487370964, + "grad_norm": 2.073577642440796, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7109460830688477, + "num_tokens": 400381256.0, + "step": 15701 + }, + { + "epoch": 1.7243575664397102, + "grad_norm": 2.0259218215942383, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7025325298309326, + "num_tokens": 400411667.0, + "step": 15702 + }, + { + "epoch": 1.7244673841423237, + "grad_norm": 2.605050802230835, + "learning_rate": 1e-06, + "loss": 0.7867, + "mean_token_accuracy": 0.7425128817558289, + "num_tokens": 400430671.0, + "step": 15703 + }, + { + "epoch": 1.7245772018449375, + "grad_norm": 2.500530481338501, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.739262580871582, + "num_tokens": 400451679.0, + "step": 15704 + }, + { + "epoch": 1.724687019547551, + "grad_norm": 2.509783983230591, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7402681112289429, + "num_tokens": 400472171.0, + "step": 15705 + }, + { + "epoch": 1.7247968372501647, + "grad_norm": 2.349138021469116, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.7029159069061279, + "num_tokens": 400498075.0, + "step": 15706 + }, + { + "epoch": 1.7249066549527785, + "grad_norm": 2.680516481399536, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7234401106834412, + "num_tokens": 400517583.0, + "step": 15707 + }, + { + "epoch": 1.725016472655392, + "grad_norm": 2.2716941833496094, + "learning_rate": 1e-06, + "loss": 0.8043, + "mean_token_accuracy": 0.7446630001068115, + "num_tokens": 400542236.0, + "step": 15708 + }, + { + "epoch": 1.7251262903580056, + "grad_norm": 2.3963851928710938, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7363206744194031, + "num_tokens": 400565189.0, + "step": 15709 + }, + { + "epoch": 1.7252361080606193, + "grad_norm": 2.414991617202759, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7101888060569763, + "num_tokens": 400587390.0, + "step": 15710 + }, + { + "epoch": 1.725345925763233, + "grad_norm": 2.2146213054656982, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7283587455749512, + "num_tokens": 400613374.0, + "step": 15711 + }, + { + "epoch": 1.7254557434658468, + "grad_norm": 2.497375726699829, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.7439877390861511, + "num_tokens": 400634134.0, + "step": 15712 + }, + { + "epoch": 1.7255655611684604, + "grad_norm": 2.687570095062256, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7286142110824585, + "num_tokens": 400654775.0, + "step": 15713 + }, + { + "epoch": 1.725675378871074, + "grad_norm": 2.179227113723755, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7114908695220947, + "num_tokens": 400681219.0, + "step": 15714 + }, + { + "epoch": 1.7257851965736877, + "grad_norm": 2.4687705039978027, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7264044284820557, + "num_tokens": 400702232.0, + "step": 15715 + }, + { + "epoch": 1.7258950142763014, + "grad_norm": 2.3107099533081055, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7158985137939453, + "num_tokens": 400726492.0, + "step": 15716 + }, + { + "epoch": 1.726004831978915, + "grad_norm": 2.604858875274658, + "learning_rate": 1e-06, + "loss": 0.7752, + "mean_token_accuracy": 0.7538805603981018, + "num_tokens": 400745937.0, + "step": 15717 + }, + { + "epoch": 1.7261146496815285, + "grad_norm": 2.342700481414795, + "learning_rate": 1e-06, + "loss": 0.7877, + "mean_token_accuracy": 0.7482350468635559, + "num_tokens": 400767656.0, + "step": 15718 + }, + { + "epoch": 1.7262244673841423, + "grad_norm": 2.2398300170898438, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7166193723678589, + "num_tokens": 400791283.0, + "step": 15719 + }, + { + "epoch": 1.726334285086756, + "grad_norm": 2.1992735862731934, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7101165652275085, + "num_tokens": 400817931.0, + "step": 15720 + }, + { + "epoch": 1.7264441027893698, + "grad_norm": 2.0859789848327637, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7155148386955261, + "num_tokens": 400847190.0, + "step": 15721 + }, + { + "epoch": 1.7265539204919833, + "grad_norm": 2.3015761375427246, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7181237936019897, + "num_tokens": 400871933.0, + "step": 15722 + }, + { + "epoch": 1.7266637381945968, + "grad_norm": 2.1942930221557617, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.712590217590332, + "num_tokens": 400900756.0, + "step": 15723 + }, + { + "epoch": 1.7267735558972106, + "grad_norm": 2.3688976764678955, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7397744655609131, + "num_tokens": 400923061.0, + "step": 15724 + }, + { + "epoch": 1.7268833735998244, + "grad_norm": 2.3810267448425293, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7130990028381348, + "num_tokens": 400947337.0, + "step": 15725 + }, + { + "epoch": 1.7269931913024381, + "grad_norm": 2.0971105098724365, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.730351984500885, + "num_tokens": 400974500.0, + "step": 15726 + }, + { + "epoch": 1.7271030090050516, + "grad_norm": 2.4520645141601562, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7354627251625061, + "num_tokens": 400997080.0, + "step": 15727 + }, + { + "epoch": 1.7272128267076652, + "grad_norm": 2.146547794342041, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7187309861183167, + "num_tokens": 401025559.0, + "step": 15728 + }, + { + "epoch": 1.727322644410279, + "grad_norm": 2.300781488418579, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7202397584915161, + "num_tokens": 401049712.0, + "step": 15729 + }, + { + "epoch": 1.7274324621128927, + "grad_norm": 2.401085138320923, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7198032140731812, + "num_tokens": 401073514.0, + "step": 15730 + }, + { + "epoch": 1.7275422798155062, + "grad_norm": 2.5977261066436768, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7142576575279236, + "num_tokens": 401093919.0, + "step": 15731 + }, + { + "epoch": 1.7276520975181198, + "grad_norm": 2.2235281467437744, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.700632631778717, + "num_tokens": 401121969.0, + "step": 15732 + }, + { + "epoch": 1.7277619152207335, + "grad_norm": 2.0722529888153076, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6952801942825317, + "num_tokens": 401151664.0, + "step": 15733 + }, + { + "epoch": 1.7278717329233473, + "grad_norm": 1.9630157947540283, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7320218682289124, + "num_tokens": 401182365.0, + "step": 15734 + }, + { + "epoch": 1.727981550625961, + "grad_norm": 2.277653932571411, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7231137752532959, + "num_tokens": 401206226.0, + "step": 15735 + }, + { + "epoch": 1.7280913683285746, + "grad_norm": 2.105959177017212, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7204942107200623, + "num_tokens": 401234377.0, + "step": 15736 + }, + { + "epoch": 1.728201186031188, + "grad_norm": 2.1083407402038574, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.6873884201049805, + "num_tokens": 401265381.0, + "step": 15737 + }, + { + "epoch": 1.7283110037338019, + "grad_norm": 2.1390256881713867, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7012834548950195, + "num_tokens": 401295829.0, + "step": 15738 + }, + { + "epoch": 1.7284208214364156, + "grad_norm": 2.1866989135742188, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.6951318979263306, + "num_tokens": 401322511.0, + "step": 15739 + }, + { + "epoch": 1.7285306391390292, + "grad_norm": 2.2721478939056396, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7089004516601562, + "num_tokens": 401347636.0, + "step": 15740 + }, + { + "epoch": 1.728640456841643, + "grad_norm": 2.1239235401153564, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7147946357727051, + "num_tokens": 401375904.0, + "step": 15741 + }, + { + "epoch": 1.7287502745442564, + "grad_norm": 2.3777713775634766, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6960493922233582, + "num_tokens": 401399560.0, + "step": 15742 + }, + { + "epoch": 1.7288600922468702, + "grad_norm": 1.909967303276062, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.6984497308731079, + "num_tokens": 401432623.0, + "step": 15743 + }, + { + "epoch": 1.728969909949484, + "grad_norm": 2.00061297416687, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6945540904998779, + "num_tokens": 401466368.0, + "step": 15744 + }, + { + "epoch": 1.7290797276520975, + "grad_norm": 2.103212356567383, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7152204513549805, + "num_tokens": 401496695.0, + "step": 15745 + }, + { + "epoch": 1.729189545354711, + "grad_norm": 2.2275936603546143, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7213603258132935, + "num_tokens": 401522499.0, + "step": 15746 + }, + { + "epoch": 1.7292993630573248, + "grad_norm": 2.2055230140686035, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7327655553817749, + "num_tokens": 401547079.0, + "step": 15747 + }, + { + "epoch": 1.7294091807599385, + "grad_norm": 2.0730371475219727, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7032715082168579, + "num_tokens": 401574805.0, + "step": 15748 + }, + { + "epoch": 1.7295189984625523, + "grad_norm": 1.9419430494308472, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7056601047515869, + "num_tokens": 401606487.0, + "step": 15749 + }, + { + "epoch": 1.7296288161651658, + "grad_norm": 2.1359174251556396, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7176069617271423, + "num_tokens": 401634013.0, + "step": 15750 + }, + { + "epoch": 1.7297386338677794, + "grad_norm": 2.091218948364258, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7117956876754761, + "num_tokens": 401662992.0, + "step": 15751 + }, + { + "epoch": 1.7298484515703931, + "grad_norm": 2.0059573650360107, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6924686431884766, + "num_tokens": 401695843.0, + "step": 15752 + }, + { + "epoch": 1.7299582692730069, + "grad_norm": 2.0678396224975586, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.6995625495910645, + "num_tokens": 401725773.0, + "step": 15753 + }, + { + "epoch": 1.7300680869756204, + "grad_norm": 2.080146312713623, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7333195209503174, + "num_tokens": 401754358.0, + "step": 15754 + }, + { + "epoch": 1.7301779046782342, + "grad_norm": 1.8940938711166382, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6914339065551758, + "num_tokens": 401787605.0, + "step": 15755 + }, + { + "epoch": 1.7302877223808477, + "grad_norm": 2.1310107707977295, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7297238707542419, + "num_tokens": 401814049.0, + "step": 15756 + }, + { + "epoch": 1.7303975400834615, + "grad_norm": 2.2675962448120117, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7243428826332092, + "num_tokens": 401838874.0, + "step": 15757 + }, + { + "epoch": 1.7305073577860752, + "grad_norm": 2.2252142429351807, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7207337617874146, + "num_tokens": 401865813.0, + "step": 15758 + }, + { + "epoch": 1.7306171754886888, + "grad_norm": 2.032409906387329, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6878403425216675, + "num_tokens": 401898373.0, + "step": 15759 + }, + { + "epoch": 1.7307269931913023, + "grad_norm": 2.37565016746521, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7274249196052551, + "num_tokens": 401921470.0, + "step": 15760 + }, + { + "epoch": 1.730836810893916, + "grad_norm": 2.1995272636413574, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.708450198173523, + "num_tokens": 401951057.0, + "step": 15761 + }, + { + "epoch": 1.7309466285965298, + "grad_norm": 2.1947546005249023, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7241798043251038, + "num_tokens": 401979099.0, + "step": 15762 + }, + { + "epoch": 1.7310564462991436, + "grad_norm": 2.7799012660980225, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7445424795150757, + "num_tokens": 401996221.0, + "step": 15763 + }, + { + "epoch": 1.731166264001757, + "grad_norm": 2.0978574752807617, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7107970714569092, + "num_tokens": 402025889.0, + "step": 15764 + }, + { + "epoch": 1.7312760817043706, + "grad_norm": 2.059549570083618, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.6996946930885315, + "num_tokens": 402056391.0, + "step": 15765 + }, + { + "epoch": 1.7313858994069844, + "grad_norm": 2.3750052452087402, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7409999966621399, + "num_tokens": 402080206.0, + "step": 15766 + }, + { + "epoch": 1.7314957171095982, + "grad_norm": 2.095052719116211, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7387164235115051, + "num_tokens": 402106333.0, + "step": 15767 + }, + { + "epoch": 1.7316055348122117, + "grad_norm": 2.0618207454681396, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7167030572891235, + "num_tokens": 402139988.0, + "step": 15768 + }, + { + "epoch": 1.7317153525148252, + "grad_norm": 1.9683747291564941, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7033290266990662, + "num_tokens": 402174142.0, + "step": 15769 + }, + { + "epoch": 1.731825170217439, + "grad_norm": 2.4588236808776855, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7090299725532532, + "num_tokens": 402197174.0, + "step": 15770 + }, + { + "epoch": 1.7319349879200527, + "grad_norm": 2.3165717124938965, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.732876718044281, + "num_tokens": 402221067.0, + "step": 15771 + }, + { + "epoch": 1.7320448056226665, + "grad_norm": 2.2977585792541504, + "learning_rate": 1e-06, + "loss": 0.781, + "mean_token_accuracy": 0.7529045343399048, + "num_tokens": 402245534.0, + "step": 15772 + }, + { + "epoch": 1.73215462332528, + "grad_norm": 2.041802406311035, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7127116322517395, + "num_tokens": 402274732.0, + "step": 15773 + }, + { + "epoch": 1.7322644410278936, + "grad_norm": 2.026615619659424, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7214360237121582, + "num_tokens": 402306770.0, + "step": 15774 + }, + { + "epoch": 1.7323742587305073, + "grad_norm": 2.348209857940674, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7039275169372559, + "num_tokens": 402331237.0, + "step": 15775 + }, + { + "epoch": 1.732484076433121, + "grad_norm": 2.2451319694519043, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7441930770874023, + "num_tokens": 402356541.0, + "step": 15776 + }, + { + "epoch": 1.7325938941357348, + "grad_norm": 2.0108065605163574, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.730950653553009, + "num_tokens": 402385664.0, + "step": 15777 + }, + { + "epoch": 1.7327037118383484, + "grad_norm": 2.3914127349853516, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7042996287345886, + "num_tokens": 402409549.0, + "step": 15778 + }, + { + "epoch": 1.732813529540962, + "grad_norm": 2.584968090057373, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7243267297744751, + "num_tokens": 402429920.0, + "step": 15779 + }, + { + "epoch": 1.7329233472435757, + "grad_norm": 2.4575681686401367, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7125665545463562, + "num_tokens": 402452753.0, + "step": 15780 + }, + { + "epoch": 1.7330331649461894, + "grad_norm": 2.136070728302002, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7211669683456421, + "num_tokens": 402479398.0, + "step": 15781 + }, + { + "epoch": 1.733142982648803, + "grad_norm": 2.215538740158081, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7231330871582031, + "num_tokens": 402504881.0, + "step": 15782 + }, + { + "epoch": 1.7332528003514165, + "grad_norm": 2.32022762298584, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7407504916191101, + "num_tokens": 402527529.0, + "step": 15783 + }, + { + "epoch": 1.7333626180540302, + "grad_norm": 2.5709784030914307, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.746872067451477, + "num_tokens": 402546271.0, + "step": 15784 + }, + { + "epoch": 1.733472435756644, + "grad_norm": 2.1259348392486572, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7387640476226807, + "num_tokens": 402572937.0, + "step": 15785 + }, + { + "epoch": 1.7335822534592578, + "grad_norm": 2.268970489501953, + "learning_rate": 1e-06, + "loss": 0.7467, + "mean_token_accuracy": 0.7579994797706604, + "num_tokens": 402596037.0, + "step": 15786 + }, + { + "epoch": 1.7336920711618713, + "grad_norm": 2.3009777069091797, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7120044827461243, + "num_tokens": 402619864.0, + "step": 15787 + }, + { + "epoch": 1.7338018888644848, + "grad_norm": 2.0540173053741455, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.691636323928833, + "num_tokens": 402650908.0, + "step": 15788 + }, + { + "epoch": 1.7339117065670986, + "grad_norm": 2.4113526344299316, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7320702075958252, + "num_tokens": 402672735.0, + "step": 15789 + }, + { + "epoch": 1.7340215242697123, + "grad_norm": 2.025745391845703, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.6968204975128174, + "num_tokens": 402705560.0, + "step": 15790 + }, + { + "epoch": 1.734131341972326, + "grad_norm": 2.393671989440918, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.716839075088501, + "num_tokens": 402727874.0, + "step": 15791 + }, + { + "epoch": 1.7342411596749396, + "grad_norm": 2.189403533935547, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7099288105964661, + "num_tokens": 402754136.0, + "step": 15792 + }, + { + "epoch": 1.7343509773775532, + "grad_norm": 2.1341443061828613, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7250452041625977, + "num_tokens": 402781933.0, + "step": 15793 + }, + { + "epoch": 1.734460795080167, + "grad_norm": 2.483792304992676, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7201005816459656, + "num_tokens": 402804794.0, + "step": 15794 + }, + { + "epoch": 1.7345706127827807, + "grad_norm": 1.9042823314666748, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.7009752988815308, + "num_tokens": 402839819.0, + "step": 15795 + }, + { + "epoch": 1.7346804304853942, + "grad_norm": 2.3029215335845947, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7202514410018921, + "num_tokens": 402865728.0, + "step": 15796 + }, + { + "epoch": 1.7347902481880078, + "grad_norm": 2.530555248260498, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7180648446083069, + "num_tokens": 402886521.0, + "step": 15797 + }, + { + "epoch": 1.7349000658906215, + "grad_norm": 2.149803638458252, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7323033809661865, + "num_tokens": 402913489.0, + "step": 15798 + }, + { + "epoch": 1.7350098835932353, + "grad_norm": 2.741506576538086, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7148664593696594, + "num_tokens": 402932457.0, + "step": 15799 + }, + { + "epoch": 1.735119701295849, + "grad_norm": 2.4583346843719482, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7443361282348633, + "num_tokens": 402952757.0, + "step": 15800 + }, + { + "epoch": 1.7352295189984626, + "grad_norm": 1.990249514579773, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7304990291595459, + "num_tokens": 402982002.0, + "step": 15801 + }, + { + "epoch": 1.735339336701076, + "grad_norm": 2.0844924449920654, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7236918807029724, + "num_tokens": 403011087.0, + "step": 15802 + }, + { + "epoch": 1.7354491544036899, + "grad_norm": 2.260000705718994, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7350422143936157, + "num_tokens": 403037800.0, + "step": 15803 + }, + { + "epoch": 1.7355589721063036, + "grad_norm": 2.1498653888702393, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7257521748542786, + "num_tokens": 403064931.0, + "step": 15804 + }, + { + "epoch": 1.7356687898089171, + "grad_norm": 2.1678168773651123, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7034318447113037, + "num_tokens": 403092495.0, + "step": 15805 + }, + { + "epoch": 1.735778607511531, + "grad_norm": 2.386209011077881, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7147378325462341, + "num_tokens": 403115939.0, + "step": 15806 + }, + { + "epoch": 1.7358884252141444, + "grad_norm": 2.834228515625, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7416816353797913, + "num_tokens": 403132385.0, + "step": 15807 + }, + { + "epoch": 1.7359982429167582, + "grad_norm": 2.4435441493988037, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7315556406974792, + "num_tokens": 403154799.0, + "step": 15808 + }, + { + "epoch": 1.736108060619372, + "grad_norm": 2.2536542415618896, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7136955261230469, + "num_tokens": 403180232.0, + "step": 15809 + }, + { + "epoch": 1.7362178783219855, + "grad_norm": 2.0476508140563965, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7230615615844727, + "num_tokens": 403208257.0, + "step": 15810 + }, + { + "epoch": 1.736327696024599, + "grad_norm": 2.0322635173797607, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7227833271026611, + "num_tokens": 403239767.0, + "step": 15811 + }, + { + "epoch": 1.7364375137272128, + "grad_norm": 2.0916717052459717, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7185164093971252, + "num_tokens": 403268444.0, + "step": 15812 + }, + { + "epoch": 1.7365473314298265, + "grad_norm": 2.108351230621338, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.71211838722229, + "num_tokens": 403297499.0, + "step": 15813 + }, + { + "epoch": 1.7366571491324403, + "grad_norm": 2.068943738937378, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.696906328201294, + "num_tokens": 403327420.0, + "step": 15814 + }, + { + "epoch": 1.7367669668350538, + "grad_norm": 2.052915334701538, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7265730500221252, + "num_tokens": 403356009.0, + "step": 15815 + }, + { + "epoch": 1.7368767845376674, + "grad_norm": 2.401855230331421, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7167720198631287, + "num_tokens": 403380051.0, + "step": 15816 + }, + { + "epoch": 1.7369866022402811, + "grad_norm": 2.3229568004608154, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7243595123291016, + "num_tokens": 403404763.0, + "step": 15817 + }, + { + "epoch": 1.7370964199428949, + "grad_norm": 2.358454942703247, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7071840763092041, + "num_tokens": 403427688.0, + "step": 15818 + }, + { + "epoch": 1.7372062376455084, + "grad_norm": 2.5879287719726562, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.73061203956604, + "num_tokens": 403447374.0, + "step": 15819 + }, + { + "epoch": 1.7373160553481222, + "grad_norm": 2.0462963581085205, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.6954492330551147, + "num_tokens": 403480343.0, + "step": 15820 + }, + { + "epoch": 1.7374258730507357, + "grad_norm": 2.470857620239258, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7161500453948975, + "num_tokens": 403501450.0, + "step": 15821 + }, + { + "epoch": 1.7375356907533495, + "grad_norm": 2.192075729370117, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7236801385879517, + "num_tokens": 403531099.0, + "step": 15822 + }, + { + "epoch": 1.7376455084559632, + "grad_norm": 2.258164644241333, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7422972321510315, + "num_tokens": 403556375.0, + "step": 15823 + }, + { + "epoch": 1.7377553261585768, + "grad_norm": 2.325913429260254, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7110390663146973, + "num_tokens": 403581029.0, + "step": 15824 + }, + { + "epoch": 1.7378651438611903, + "grad_norm": 2.423905372619629, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7030373811721802, + "num_tokens": 403606479.0, + "step": 15825 + }, + { + "epoch": 1.737974961563804, + "grad_norm": 2.0752334594726562, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7092930674552917, + "num_tokens": 403637296.0, + "step": 15826 + }, + { + "epoch": 1.7380847792664178, + "grad_norm": 2.1626744270324707, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7232356667518616, + "num_tokens": 403663543.0, + "step": 15827 + }, + { + "epoch": 1.7381945969690316, + "grad_norm": 2.391338348388672, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7274502515792847, + "num_tokens": 403687880.0, + "step": 15828 + }, + { + "epoch": 1.738304414671645, + "grad_norm": 2.0631747245788574, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7299226522445679, + "num_tokens": 403715078.0, + "step": 15829 + }, + { + "epoch": 1.7384142323742586, + "grad_norm": 1.9831904172897339, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.719935953617096, + "num_tokens": 403745096.0, + "step": 15830 + }, + { + "epoch": 1.7385240500768724, + "grad_norm": 2.2565724849700928, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.725178062915802, + "num_tokens": 403771471.0, + "step": 15831 + }, + { + "epoch": 1.7386338677794861, + "grad_norm": 2.2955551147460938, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7162395715713501, + "num_tokens": 403797499.0, + "step": 15832 + }, + { + "epoch": 1.7387436854820997, + "grad_norm": 2.412834405899048, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7183389067649841, + "num_tokens": 403820550.0, + "step": 15833 + }, + { + "epoch": 1.7388535031847132, + "grad_norm": 2.525998830795288, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7357321977615356, + "num_tokens": 403841867.0, + "step": 15834 + }, + { + "epoch": 1.738963320887327, + "grad_norm": 2.3245255947113037, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7132010459899902, + "num_tokens": 403867430.0, + "step": 15835 + }, + { + "epoch": 1.7390731385899407, + "grad_norm": 2.161611318588257, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7105528116226196, + "num_tokens": 403896170.0, + "step": 15836 + }, + { + "epoch": 1.7391829562925545, + "grad_norm": 2.173164129257202, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7090778350830078, + "num_tokens": 403924204.0, + "step": 15837 + }, + { + "epoch": 1.739292773995168, + "grad_norm": 2.353203058242798, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7249503135681152, + "num_tokens": 403949329.0, + "step": 15838 + }, + { + "epoch": 1.7394025916977816, + "grad_norm": 2.4501190185546875, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7222034931182861, + "num_tokens": 403972013.0, + "step": 15839 + }, + { + "epoch": 1.7395124094003953, + "grad_norm": 2.162407636642456, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6928683519363403, + "num_tokens": 403999834.0, + "step": 15840 + }, + { + "epoch": 1.739622227103009, + "grad_norm": 2.468391180038452, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7303357124328613, + "num_tokens": 404022427.0, + "step": 15841 + }, + { + "epoch": 1.7397320448056228, + "grad_norm": 2.190279722213745, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.725906491279602, + "num_tokens": 404048190.0, + "step": 15842 + }, + { + "epoch": 1.7398418625082364, + "grad_norm": 2.138996124267578, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7135207653045654, + "num_tokens": 404076778.0, + "step": 15843 + }, + { + "epoch": 1.73995168021085, + "grad_norm": 2.3246169090270996, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7158180475234985, + "num_tokens": 404103927.0, + "step": 15844 + }, + { + "epoch": 1.7400614979134637, + "grad_norm": 1.9249584674835205, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7111214399337769, + "num_tokens": 404135526.0, + "step": 15845 + }, + { + "epoch": 1.7401713156160774, + "grad_norm": 2.3587405681610107, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7421631217002869, + "num_tokens": 404157976.0, + "step": 15846 + }, + { + "epoch": 1.740281133318691, + "grad_norm": 2.2702105045318604, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6921058893203735, + "num_tokens": 404182984.0, + "step": 15847 + }, + { + "epoch": 1.7403909510213045, + "grad_norm": 2.3060224056243896, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7199165225028992, + "num_tokens": 404208370.0, + "step": 15848 + }, + { + "epoch": 1.7405007687239182, + "grad_norm": 2.008657455444336, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.7612376809120178, + "num_tokens": 404236054.0, + "step": 15849 + }, + { + "epoch": 1.740610586426532, + "grad_norm": 2.1733336448669434, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7255383133888245, + "num_tokens": 404260710.0, + "step": 15850 + }, + { + "epoch": 1.7407204041291457, + "grad_norm": 2.0683186054229736, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7187375426292419, + "num_tokens": 404291091.0, + "step": 15851 + }, + { + "epoch": 1.7408302218317593, + "grad_norm": 2.119560480117798, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7125184535980225, + "num_tokens": 404317214.0, + "step": 15852 + }, + { + "epoch": 1.7409400395343728, + "grad_norm": 2.6282007694244385, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7231840491294861, + "num_tokens": 404338237.0, + "step": 15853 + }, + { + "epoch": 1.7410498572369866, + "grad_norm": 2.0369341373443604, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.7012231945991516, + "num_tokens": 404366727.0, + "step": 15854 + }, + { + "epoch": 1.7411596749396003, + "grad_norm": 2.360468626022339, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7478783130645752, + "num_tokens": 404389347.0, + "step": 15855 + }, + { + "epoch": 1.741269492642214, + "grad_norm": 2.3714091777801514, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.746407151222229, + "num_tokens": 404412012.0, + "step": 15856 + }, + { + "epoch": 1.7413793103448276, + "grad_norm": 2.1997077465057373, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7153387069702148, + "num_tokens": 404438370.0, + "step": 15857 + }, + { + "epoch": 1.7414891280474412, + "grad_norm": 2.116077423095703, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7047111988067627, + "num_tokens": 404467648.0, + "step": 15858 + }, + { + "epoch": 1.741598945750055, + "grad_norm": 2.4800665378570557, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7272466421127319, + "num_tokens": 404489477.0, + "step": 15859 + }, + { + "epoch": 1.7417087634526687, + "grad_norm": 2.154754638671875, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7344966530799866, + "num_tokens": 404515706.0, + "step": 15860 + }, + { + "epoch": 1.7418185811552822, + "grad_norm": 2.3625354766845703, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7200050354003906, + "num_tokens": 404538674.0, + "step": 15861 + }, + { + "epoch": 1.7419283988578957, + "grad_norm": 2.988445997238159, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7360585927963257, + "num_tokens": 404554936.0, + "step": 15862 + }, + { + "epoch": 1.7420382165605095, + "grad_norm": 2.1259005069732666, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.720211386680603, + "num_tokens": 404581991.0, + "step": 15863 + }, + { + "epoch": 1.7421480342631233, + "grad_norm": 2.002204418182373, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7160582542419434, + "num_tokens": 404613941.0, + "step": 15864 + }, + { + "epoch": 1.742257851965737, + "grad_norm": 2.320564031600952, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7186379432678223, + "num_tokens": 404637239.0, + "step": 15865 + }, + { + "epoch": 1.7423676696683505, + "grad_norm": 2.2456295490264893, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7182011008262634, + "num_tokens": 404662576.0, + "step": 15866 + }, + { + "epoch": 1.742477487370964, + "grad_norm": 1.9897043704986572, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7052798271179199, + "num_tokens": 404696676.0, + "step": 15867 + }, + { + "epoch": 1.7425873050735778, + "grad_norm": 2.212268114089966, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7270456552505493, + "num_tokens": 404720834.0, + "step": 15868 + }, + { + "epoch": 1.7426971227761916, + "grad_norm": 2.1399426460266113, + "learning_rate": 1e-06, + "loss": 1.0394, + "mean_token_accuracy": 0.6763449311256409, + "num_tokens": 404751512.0, + "step": 15869 + }, + { + "epoch": 1.7428069404788051, + "grad_norm": 2.5180442333221436, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7315853834152222, + "num_tokens": 404773254.0, + "step": 15870 + }, + { + "epoch": 1.742916758181419, + "grad_norm": 1.999886393547058, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7319732904434204, + "num_tokens": 404802765.0, + "step": 15871 + }, + { + "epoch": 1.7430265758840324, + "grad_norm": 2.2625129222869873, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.72263503074646, + "num_tokens": 404827840.0, + "step": 15872 + }, + { + "epoch": 1.7431363935866462, + "grad_norm": 2.5230798721313477, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7102569937705994, + "num_tokens": 404849972.0, + "step": 15873 + }, + { + "epoch": 1.74324621128926, + "grad_norm": 2.4252843856811523, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.73415607213974, + "num_tokens": 404872184.0, + "step": 15874 + }, + { + "epoch": 1.7433560289918735, + "grad_norm": 2.167706251144409, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.745644211769104, + "num_tokens": 404896848.0, + "step": 15875 + }, + { + "epoch": 1.743465846694487, + "grad_norm": 2.3656206130981445, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7299131155014038, + "num_tokens": 404918157.0, + "step": 15876 + }, + { + "epoch": 1.7435756643971008, + "grad_norm": 2.186028003692627, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7047587633132935, + "num_tokens": 404946704.0, + "step": 15877 + }, + { + "epoch": 1.7436854820997145, + "grad_norm": 2.363731861114502, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7196893095970154, + "num_tokens": 404969931.0, + "step": 15878 + }, + { + "epoch": 1.7437952998023283, + "grad_norm": 2.3808512687683105, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7173665761947632, + "num_tokens": 404993795.0, + "step": 15879 + }, + { + "epoch": 1.7439051175049418, + "grad_norm": 2.243849754333496, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7134333848953247, + "num_tokens": 405020289.0, + "step": 15880 + }, + { + "epoch": 1.7440149352075554, + "grad_norm": 2.0875349044799805, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7125476598739624, + "num_tokens": 405050372.0, + "step": 15881 + }, + { + "epoch": 1.744124752910169, + "grad_norm": 2.324124336242676, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7141807079315186, + "num_tokens": 405075028.0, + "step": 15882 + }, + { + "epoch": 1.7442345706127829, + "grad_norm": 2.3833210468292236, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7102776765823364, + "num_tokens": 405098702.0, + "step": 15883 + }, + { + "epoch": 1.7443443883153964, + "grad_norm": 2.134687900543213, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7081669569015503, + "num_tokens": 405126243.0, + "step": 15884 + }, + { + "epoch": 1.7444542060180102, + "grad_norm": 2.6572954654693604, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7437994480133057, + "num_tokens": 405144414.0, + "step": 15885 + }, + { + "epoch": 1.7445640237206237, + "grad_norm": 2.693384885787964, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7359206676483154, + "num_tokens": 405163324.0, + "step": 15886 + }, + { + "epoch": 1.7446738414232374, + "grad_norm": 2.3078255653381348, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7118504047393799, + "num_tokens": 405188505.0, + "step": 15887 + }, + { + "epoch": 1.7447836591258512, + "grad_norm": 2.1855907440185547, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7338614463806152, + "num_tokens": 405216443.0, + "step": 15888 + }, + { + "epoch": 1.7448934768284647, + "grad_norm": 2.5654332637786865, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7350497841835022, + "num_tokens": 405236735.0, + "step": 15889 + }, + { + "epoch": 1.7450032945310783, + "grad_norm": 2.253678560256958, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.745599091053009, + "num_tokens": 405260963.0, + "step": 15890 + }, + { + "epoch": 1.745113112233692, + "grad_norm": 2.267270088195801, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7197597622871399, + "num_tokens": 405288180.0, + "step": 15891 + }, + { + "epoch": 1.7452229299363058, + "grad_norm": 2.392350912094116, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7138718366622925, + "num_tokens": 405312754.0, + "step": 15892 + }, + { + "epoch": 1.7453327476389195, + "grad_norm": 2.092801809310913, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.6984889507293701, + "num_tokens": 405343219.0, + "step": 15893 + }, + { + "epoch": 1.745442565341533, + "grad_norm": 2.309739589691162, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7097460627555847, + "num_tokens": 405368820.0, + "step": 15894 + }, + { + "epoch": 1.7455523830441466, + "grad_norm": 2.178415536880493, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7227935791015625, + "num_tokens": 405395307.0, + "step": 15895 + }, + { + "epoch": 1.7456622007467604, + "grad_norm": 2.1807968616485596, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7061736583709717, + "num_tokens": 405423810.0, + "step": 15896 + }, + { + "epoch": 1.7457720184493741, + "grad_norm": 2.1516425609588623, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7448990941047668, + "num_tokens": 405449345.0, + "step": 15897 + }, + { + "epoch": 1.7458818361519877, + "grad_norm": 2.2749993801116943, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.700778603553772, + "num_tokens": 405472910.0, + "step": 15898 + }, + { + "epoch": 1.7459916538546012, + "grad_norm": 2.1274094581604004, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7132402658462524, + "num_tokens": 405502958.0, + "step": 15899 + }, + { + "epoch": 1.746101471557215, + "grad_norm": 2.074366807937622, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7030367851257324, + "num_tokens": 405535977.0, + "step": 15900 + }, + { + "epoch": 1.7462112892598287, + "grad_norm": 2.30617618560791, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7325608730316162, + "num_tokens": 405558720.0, + "step": 15901 + }, + { + "epoch": 1.7463211069624425, + "grad_norm": 2.1708037853240967, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7098684310913086, + "num_tokens": 405585467.0, + "step": 15902 + }, + { + "epoch": 1.746430924665056, + "grad_norm": 2.0699093341827393, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7085858583450317, + "num_tokens": 405615047.0, + "step": 15903 + }, + { + "epoch": 1.7465407423676695, + "grad_norm": 2.168168306350708, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7225431203842163, + "num_tokens": 405642395.0, + "step": 15904 + }, + { + "epoch": 1.7466505600702833, + "grad_norm": 2.6376750469207764, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7597885131835938, + "num_tokens": 405660104.0, + "step": 15905 + }, + { + "epoch": 1.746760377772897, + "grad_norm": 2.255686044692993, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7105973958969116, + "num_tokens": 405687305.0, + "step": 15906 + }, + { + "epoch": 1.7468701954755108, + "grad_norm": 2.405482053756714, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7131941318511963, + "num_tokens": 405712194.0, + "step": 15907 + }, + { + "epoch": 1.7469800131781243, + "grad_norm": 2.3136136531829834, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7273331880569458, + "num_tokens": 405735170.0, + "step": 15908 + }, + { + "epoch": 1.7470898308807379, + "grad_norm": 2.6216816902160645, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7381744384765625, + "num_tokens": 405754173.0, + "step": 15909 + }, + { + "epoch": 1.7471996485833516, + "grad_norm": 2.5710086822509766, + "learning_rate": 1e-06, + "loss": 0.8057, + "mean_token_accuracy": 0.743730902671814, + "num_tokens": 405774458.0, + "step": 15910 + }, + { + "epoch": 1.7473094662859654, + "grad_norm": 1.9770125150680542, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7095599174499512, + "num_tokens": 405806960.0, + "step": 15911 + }, + { + "epoch": 1.747419283988579, + "grad_norm": 2.103912353515625, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7275534868240356, + "num_tokens": 405832626.0, + "step": 15912 + }, + { + "epoch": 1.7475291016911925, + "grad_norm": 2.22334361076355, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.700090229511261, + "num_tokens": 405858723.0, + "step": 15913 + }, + { + "epoch": 1.7476389193938062, + "grad_norm": 2.24222469329834, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.6998229622840881, + "num_tokens": 405884419.0, + "step": 15914 + }, + { + "epoch": 1.74774873709642, + "grad_norm": 2.005552291870117, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7009884715080261, + "num_tokens": 405917740.0, + "step": 15915 + }, + { + "epoch": 1.7478585547990337, + "grad_norm": 2.0471620559692383, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7038296461105347, + "num_tokens": 405946649.0, + "step": 15916 + }, + { + "epoch": 1.7479683725016473, + "grad_norm": 2.1972618103027344, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7258046865463257, + "num_tokens": 405971251.0, + "step": 15917 + }, + { + "epoch": 1.7480781902042608, + "grad_norm": 2.2782208919525146, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7223122119903564, + "num_tokens": 405997274.0, + "step": 15918 + }, + { + "epoch": 1.7481880079068746, + "grad_norm": 2.034180164337158, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7259169220924377, + "num_tokens": 406027379.0, + "step": 15919 + }, + { + "epoch": 1.7482978256094883, + "grad_norm": 2.030930519104004, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7171449065208435, + "num_tokens": 406055434.0, + "step": 15920 + }, + { + "epoch": 1.7484076433121019, + "grad_norm": 2.348229169845581, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7374132871627808, + "num_tokens": 406079008.0, + "step": 15921 + }, + { + "epoch": 1.7485174610147156, + "grad_norm": 2.121000051498413, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7235002517700195, + "num_tokens": 406105183.0, + "step": 15922 + }, + { + "epoch": 1.7486272787173291, + "grad_norm": 2.587709665298462, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7118304371833801, + "num_tokens": 406126275.0, + "step": 15923 + }, + { + "epoch": 1.748737096419943, + "grad_norm": 2.603006601333618, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7046365737915039, + "num_tokens": 406146477.0, + "step": 15924 + }, + { + "epoch": 1.7488469141225567, + "grad_norm": 2.2559759616851807, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.696780800819397, + "num_tokens": 406174725.0, + "step": 15925 + }, + { + "epoch": 1.7489567318251702, + "grad_norm": 2.0726847648620605, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7153209447860718, + "num_tokens": 406203791.0, + "step": 15926 + }, + { + "epoch": 1.7490665495277837, + "grad_norm": 2.5368831157684326, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7241857051849365, + "num_tokens": 406225911.0, + "step": 15927 + }, + { + "epoch": 1.7491763672303975, + "grad_norm": 2.2824299335479736, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7118602991104126, + "num_tokens": 406250020.0, + "step": 15928 + }, + { + "epoch": 1.7492861849330112, + "grad_norm": 2.2797787189483643, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7275803685188293, + "num_tokens": 406275618.0, + "step": 15929 + }, + { + "epoch": 1.749396002635625, + "grad_norm": 2.3152267932891846, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.711338996887207, + "num_tokens": 406300586.0, + "step": 15930 + }, + { + "epoch": 1.7495058203382385, + "grad_norm": 2.284554958343506, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7165766954421997, + "num_tokens": 406325888.0, + "step": 15931 + }, + { + "epoch": 1.749615638040852, + "grad_norm": 2.177795648574829, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7146129608154297, + "num_tokens": 406352510.0, + "step": 15932 + }, + { + "epoch": 1.7497254557434658, + "grad_norm": 2.1606035232543945, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.7031933069229126, + "num_tokens": 406380989.0, + "step": 15933 + }, + { + "epoch": 1.7498352734460796, + "grad_norm": 2.029679536819458, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7027682065963745, + "num_tokens": 406412551.0, + "step": 15934 + }, + { + "epoch": 1.7499450911486931, + "grad_norm": 2.103297472000122, + "learning_rate": 1e-06, + "loss": 0.7836, + "mean_token_accuracy": 0.7540750503540039, + "num_tokens": 406439003.0, + "step": 15935 + }, + { + "epoch": 1.7500549088513069, + "grad_norm": 2.1081693172454834, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7143800258636475, + "num_tokens": 406467550.0, + "step": 15936 + }, + { + "epoch": 1.7501647265539204, + "grad_norm": 2.5027987957000732, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7254488468170166, + "num_tokens": 406488812.0, + "step": 15937 + }, + { + "epoch": 1.7502745442565342, + "grad_norm": 2.0242772102355957, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7030826210975647, + "num_tokens": 406519006.0, + "step": 15938 + }, + { + "epoch": 1.750384361959148, + "grad_norm": 2.0270907878875732, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7470982074737549, + "num_tokens": 406546376.0, + "step": 15939 + }, + { + "epoch": 1.7504941796617615, + "grad_norm": 2.298888683319092, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.728742778301239, + "num_tokens": 406571381.0, + "step": 15940 + }, + { + "epoch": 1.750603997364375, + "grad_norm": 2.2148067951202393, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7249064445495605, + "num_tokens": 406597577.0, + "step": 15941 + }, + { + "epoch": 1.7507138150669888, + "grad_norm": 2.493752956390381, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.758715808391571, + "num_tokens": 406616778.0, + "step": 15942 + }, + { + "epoch": 1.7508236327696025, + "grad_norm": 2.36041522026062, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7038458585739136, + "num_tokens": 406639772.0, + "step": 15943 + }, + { + "epoch": 1.7509334504722163, + "grad_norm": 2.114166259765625, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7127748727798462, + "num_tokens": 406667167.0, + "step": 15944 + }, + { + "epoch": 1.7510432681748298, + "grad_norm": 2.0003554821014404, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7064379453659058, + "num_tokens": 406697194.0, + "step": 15945 + }, + { + "epoch": 1.7511530858774433, + "grad_norm": 2.4363203048706055, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7317795753479004, + "num_tokens": 406718944.0, + "step": 15946 + }, + { + "epoch": 1.751262903580057, + "grad_norm": 2.514907121658325, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7434958815574646, + "num_tokens": 406738936.0, + "step": 15947 + }, + { + "epoch": 1.7513727212826709, + "grad_norm": 2.0258541107177734, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7000796794891357, + "num_tokens": 406767559.0, + "step": 15948 + }, + { + "epoch": 1.7514825389852844, + "grad_norm": 2.2394282817840576, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7164691686630249, + "num_tokens": 406793619.0, + "step": 15949 + }, + { + "epoch": 1.7515923566878981, + "grad_norm": 2.3020315170288086, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7036802768707275, + "num_tokens": 406820672.0, + "step": 15950 + }, + { + "epoch": 1.7517021743905117, + "grad_norm": 2.2641148567199707, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7101390361785889, + "num_tokens": 406846628.0, + "step": 15951 + }, + { + "epoch": 1.7518119920931254, + "grad_norm": 2.1944432258605957, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7340697050094604, + "num_tokens": 406871203.0, + "step": 15952 + }, + { + "epoch": 1.7519218097957392, + "grad_norm": 2.2284648418426514, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7068154811859131, + "num_tokens": 406898582.0, + "step": 15953 + }, + { + "epoch": 1.7520316274983527, + "grad_norm": 2.3921096324920654, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7248754501342773, + "num_tokens": 406921084.0, + "step": 15954 + }, + { + "epoch": 1.7521414452009663, + "grad_norm": 2.389279842376709, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7209572196006775, + "num_tokens": 406943505.0, + "step": 15955 + }, + { + "epoch": 1.75225126290358, + "grad_norm": 2.286895513534546, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7258095741271973, + "num_tokens": 406968123.0, + "step": 15956 + }, + { + "epoch": 1.7523610806061938, + "grad_norm": 2.1068625450134277, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7295716404914856, + "num_tokens": 406994829.0, + "step": 15957 + }, + { + "epoch": 1.7524708983088075, + "grad_norm": 2.151167869567871, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7288315296173096, + "num_tokens": 407019794.0, + "step": 15958 + }, + { + "epoch": 1.752580716011421, + "grad_norm": 2.032758951187134, + "learning_rate": 1e-06, + "loss": 1.0238, + "mean_token_accuracy": 0.687389612197876, + "num_tokens": 407052157.0, + "step": 15959 + }, + { + "epoch": 1.7526905337140346, + "grad_norm": 2.2347190380096436, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7261704802513123, + "num_tokens": 407078351.0, + "step": 15960 + }, + { + "epoch": 1.7528003514166484, + "grad_norm": 2.415977716445923, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7262757420539856, + "num_tokens": 407099393.0, + "step": 15961 + }, + { + "epoch": 1.7529101691192621, + "grad_norm": 2.4094126224517822, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.6983625888824463, + "num_tokens": 407123216.0, + "step": 15962 + }, + { + "epoch": 1.7530199868218757, + "grad_norm": 1.926924467086792, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6851466298103333, + "num_tokens": 407155486.0, + "step": 15963 + }, + { + "epoch": 1.7531298045244892, + "grad_norm": 1.9767467975616455, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7003748416900635, + "num_tokens": 407189086.0, + "step": 15964 + }, + { + "epoch": 1.753239622227103, + "grad_norm": 2.528351068496704, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7171573042869568, + "num_tokens": 407210090.0, + "step": 15965 + }, + { + "epoch": 1.7533494399297167, + "grad_norm": 2.2579715251922607, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7182600498199463, + "num_tokens": 407234820.0, + "step": 15966 + }, + { + "epoch": 1.7534592576323305, + "grad_norm": 2.0512216091156006, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7218952775001526, + "num_tokens": 407265356.0, + "step": 15967 + }, + { + "epoch": 1.753569075334944, + "grad_norm": 2.2713239192962646, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7214966416358948, + "num_tokens": 407291544.0, + "step": 15968 + }, + { + "epoch": 1.7536788930375575, + "grad_norm": 2.1170263290405273, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7041946649551392, + "num_tokens": 407320544.0, + "step": 15969 + }, + { + "epoch": 1.7537887107401713, + "grad_norm": 2.2520298957824707, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7254811525344849, + "num_tokens": 407345743.0, + "step": 15970 + }, + { + "epoch": 1.753898528442785, + "grad_norm": 2.334735155105591, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7258783578872681, + "num_tokens": 407369912.0, + "step": 15971 + }, + { + "epoch": 1.7540083461453988, + "grad_norm": 2.430499792098999, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7204739451408386, + "num_tokens": 407392476.0, + "step": 15972 + }, + { + "epoch": 1.7541181638480123, + "grad_norm": 2.382556438446045, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7180558443069458, + "num_tokens": 407417496.0, + "step": 15973 + }, + { + "epoch": 1.7542279815506259, + "grad_norm": 2.513206958770752, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7132529616355896, + "num_tokens": 407440321.0, + "step": 15974 + }, + { + "epoch": 1.7543377992532396, + "grad_norm": 2.0291383266448975, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7384860515594482, + "num_tokens": 407468863.0, + "step": 15975 + }, + { + "epoch": 1.7544476169558534, + "grad_norm": 2.389967203140259, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7062305212020874, + "num_tokens": 407493078.0, + "step": 15976 + }, + { + "epoch": 1.754557434658467, + "grad_norm": 2.2422168254852295, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7238605618476868, + "num_tokens": 407517552.0, + "step": 15977 + }, + { + "epoch": 1.7546672523610805, + "grad_norm": 1.9807966947555542, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7018018960952759, + "num_tokens": 407548890.0, + "step": 15978 + }, + { + "epoch": 1.7547770700636942, + "grad_norm": 2.160627841949463, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7138375639915466, + "num_tokens": 407575756.0, + "step": 15979 + }, + { + "epoch": 1.754886887766308, + "grad_norm": 2.30023455619812, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7328022718429565, + "num_tokens": 407598531.0, + "step": 15980 + }, + { + "epoch": 1.7549967054689217, + "grad_norm": 2.542559862136841, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7390472292900085, + "num_tokens": 407619430.0, + "step": 15981 + }, + { + "epoch": 1.7551065231715353, + "grad_norm": 2.3561062812805176, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7264302968978882, + "num_tokens": 407643380.0, + "step": 15982 + }, + { + "epoch": 1.7552163408741488, + "grad_norm": 2.3428797721862793, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7279841899871826, + "num_tokens": 407667745.0, + "step": 15983 + }, + { + "epoch": 1.7553261585767626, + "grad_norm": 2.0149199962615967, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7334673404693604, + "num_tokens": 407696809.0, + "step": 15984 + }, + { + "epoch": 1.7554359762793763, + "grad_norm": 2.067765951156616, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6963990926742554, + "num_tokens": 407728913.0, + "step": 15985 + }, + { + "epoch": 1.7555457939819898, + "grad_norm": 2.6967406272888184, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7376518845558167, + "num_tokens": 407748584.0, + "step": 15986 + }, + { + "epoch": 1.7556556116846036, + "grad_norm": 1.8931095600128174, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7037510275840759, + "num_tokens": 407780898.0, + "step": 15987 + }, + { + "epoch": 1.7557654293872171, + "grad_norm": 1.9778177738189697, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7064285278320312, + "num_tokens": 407813231.0, + "step": 15988 + }, + { + "epoch": 1.755875247089831, + "grad_norm": 2.2278635501861572, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.727698028087616, + "num_tokens": 407839505.0, + "step": 15989 + }, + { + "epoch": 1.7559850647924446, + "grad_norm": 2.2286081314086914, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.6906365156173706, + "num_tokens": 407867567.0, + "step": 15990 + }, + { + "epoch": 1.7560948824950582, + "grad_norm": 2.608767032623291, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7271072864532471, + "num_tokens": 407887969.0, + "step": 15991 + }, + { + "epoch": 1.7562047001976717, + "grad_norm": 2.1783413887023926, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7111754417419434, + "num_tokens": 407914798.0, + "step": 15992 + }, + { + "epoch": 1.7563145179002855, + "grad_norm": 2.329404354095459, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7283569574356079, + "num_tokens": 407937611.0, + "step": 15993 + }, + { + "epoch": 1.7564243356028992, + "grad_norm": 2.310816526412964, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7326496839523315, + "num_tokens": 407962568.0, + "step": 15994 + }, + { + "epoch": 1.756534153305513, + "grad_norm": 2.3212172985076904, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.739920973777771, + "num_tokens": 407986531.0, + "step": 15995 + }, + { + "epoch": 1.7566439710081265, + "grad_norm": 2.1882970333099365, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7108883857727051, + "num_tokens": 408010982.0, + "step": 15996 + }, + { + "epoch": 1.75675378871074, + "grad_norm": 2.473700761795044, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7281770706176758, + "num_tokens": 408032318.0, + "step": 15997 + }, + { + "epoch": 1.7568636064133538, + "grad_norm": 2.3537046909332275, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7367056608200073, + "num_tokens": 408055082.0, + "step": 15998 + }, + { + "epoch": 1.7569734241159676, + "grad_norm": 2.5475687980651855, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7501906156539917, + "num_tokens": 408074138.0, + "step": 15999 + }, + { + "epoch": 1.757083241818581, + "grad_norm": 2.3691272735595703, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7380954027175903, + "num_tokens": 408096955.0, + "step": 16000 + }, + { + "epoch": 1.7571930595211949, + "grad_norm": 2.050865650177002, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6881564855575562, + "num_tokens": 408127889.0, + "step": 16001 + }, + { + "epoch": 1.7573028772238084, + "grad_norm": 2.474097728729248, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7216935157775879, + "num_tokens": 408149864.0, + "step": 16002 + }, + { + "epoch": 1.7574126949264222, + "grad_norm": 2.23343825340271, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.720589280128479, + "num_tokens": 408176439.0, + "step": 16003 + }, + { + "epoch": 1.757522512629036, + "grad_norm": 2.127711772918701, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7099137306213379, + "num_tokens": 408204558.0, + "step": 16004 + }, + { + "epoch": 1.7576323303316495, + "grad_norm": 2.1184639930725098, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7209707498550415, + "num_tokens": 408232826.0, + "step": 16005 + }, + { + "epoch": 1.757742148034263, + "grad_norm": 2.202263116836548, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7231436371803284, + "num_tokens": 408259126.0, + "step": 16006 + }, + { + "epoch": 1.7578519657368767, + "grad_norm": 2.5252480506896973, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7247090339660645, + "num_tokens": 408282029.0, + "step": 16007 + }, + { + "epoch": 1.7579617834394905, + "grad_norm": 2.3418338298797607, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7148350477218628, + "num_tokens": 408306083.0, + "step": 16008 + }, + { + "epoch": 1.7580716011421043, + "grad_norm": 2.3280341625213623, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7028241157531738, + "num_tokens": 408331042.0, + "step": 16009 + }, + { + "epoch": 1.7581814188447178, + "grad_norm": 2.5089786052703857, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7102375030517578, + "num_tokens": 408352433.0, + "step": 16010 + }, + { + "epoch": 1.7582912365473313, + "grad_norm": 2.443743944168091, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7316474914550781, + "num_tokens": 408374387.0, + "step": 16011 + }, + { + "epoch": 1.758401054249945, + "grad_norm": 2.1125147342681885, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6963102221488953, + "num_tokens": 408404540.0, + "step": 16012 + }, + { + "epoch": 1.7585108719525588, + "grad_norm": 2.4244868755340576, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7287642955780029, + "num_tokens": 408426670.0, + "step": 16013 + }, + { + "epoch": 1.7586206896551724, + "grad_norm": 2.0607407093048096, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.7466038465499878, + "num_tokens": 408454790.0, + "step": 16014 + }, + { + "epoch": 1.758730507357786, + "grad_norm": 2.0696654319763184, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7098250389099121, + "num_tokens": 408483820.0, + "step": 16015 + }, + { + "epoch": 1.7588403250603997, + "grad_norm": 2.161024808883667, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7076430320739746, + "num_tokens": 408511092.0, + "step": 16016 + }, + { + "epoch": 1.7589501427630134, + "grad_norm": 2.0533034801483154, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.714054524898529, + "num_tokens": 408542040.0, + "step": 16017 + }, + { + "epoch": 1.7590599604656272, + "grad_norm": 2.1616580486297607, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.6999019384384155, + "num_tokens": 408570368.0, + "step": 16018 + }, + { + "epoch": 1.7591697781682407, + "grad_norm": 2.6113600730895996, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.74016934633255, + "num_tokens": 408590281.0, + "step": 16019 + }, + { + "epoch": 1.7592795958708543, + "grad_norm": 2.514739990234375, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7275800108909607, + "num_tokens": 408611862.0, + "step": 16020 + }, + { + "epoch": 1.759389413573468, + "grad_norm": 2.505521774291992, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.73976069688797, + "num_tokens": 408630586.0, + "step": 16021 + }, + { + "epoch": 1.7594992312760818, + "grad_norm": 2.234612226486206, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.712348222732544, + "num_tokens": 408654358.0, + "step": 16022 + }, + { + "epoch": 1.7596090489786955, + "grad_norm": 2.067784070968628, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7131996154785156, + "num_tokens": 408684509.0, + "step": 16023 + }, + { + "epoch": 1.759718866681309, + "grad_norm": 2.120732307434082, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7424290776252747, + "num_tokens": 408710586.0, + "step": 16024 + }, + { + "epoch": 1.7598286843839226, + "grad_norm": 2.2133371829986572, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7031601667404175, + "num_tokens": 408738398.0, + "step": 16025 + }, + { + "epoch": 1.7599385020865363, + "grad_norm": 2.19750714302063, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7103819251060486, + "num_tokens": 408764659.0, + "step": 16026 + }, + { + "epoch": 1.76004831978915, + "grad_norm": 2.6738767623901367, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7195698022842407, + "num_tokens": 408783023.0, + "step": 16027 + }, + { + "epoch": 1.7601581374917636, + "grad_norm": 1.8791431188583374, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.6975754499435425, + "num_tokens": 408817837.0, + "step": 16028 + }, + { + "epoch": 1.7602679551943772, + "grad_norm": 2.095592975616455, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7163563370704651, + "num_tokens": 408848278.0, + "step": 16029 + }, + { + "epoch": 1.760377772896991, + "grad_norm": 2.2115731239318848, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7280874252319336, + "num_tokens": 408874474.0, + "step": 16030 + }, + { + "epoch": 1.7604875905996047, + "grad_norm": 2.0159175395965576, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.727139949798584, + "num_tokens": 408903894.0, + "step": 16031 + }, + { + "epoch": 1.7605974083022184, + "grad_norm": 2.2409515380859375, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7356739044189453, + "num_tokens": 408929256.0, + "step": 16032 + }, + { + "epoch": 1.760707226004832, + "grad_norm": 2.2423439025878906, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7230523824691772, + "num_tokens": 408954923.0, + "step": 16033 + }, + { + "epoch": 1.7608170437074455, + "grad_norm": 2.1827762126922607, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7138659954071045, + "num_tokens": 408983667.0, + "step": 16034 + }, + { + "epoch": 1.7609268614100593, + "grad_norm": 2.0940709114074707, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7387088537216187, + "num_tokens": 409011697.0, + "step": 16035 + }, + { + "epoch": 1.761036679112673, + "grad_norm": 2.2503979206085205, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7204397916793823, + "num_tokens": 409037663.0, + "step": 16036 + }, + { + "epoch": 1.7611464968152868, + "grad_norm": 2.133284330368042, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.7003471255302429, + "num_tokens": 409067782.0, + "step": 16037 + }, + { + "epoch": 1.7612563145179003, + "grad_norm": 2.1928892135620117, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7169139385223389, + "num_tokens": 409095596.0, + "step": 16038 + }, + { + "epoch": 1.7613661322205139, + "grad_norm": 2.1348156929016113, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7085888385772705, + "num_tokens": 409122653.0, + "step": 16039 + }, + { + "epoch": 1.7614759499231276, + "grad_norm": 2.3627777099609375, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7003159523010254, + "num_tokens": 409145988.0, + "step": 16040 + }, + { + "epoch": 1.7615857676257414, + "grad_norm": 2.1464121341705322, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.722266674041748, + "num_tokens": 409172889.0, + "step": 16041 + }, + { + "epoch": 1.761695585328355, + "grad_norm": 2.228788137435913, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7336753010749817, + "num_tokens": 409196579.0, + "step": 16042 + }, + { + "epoch": 1.7618054030309684, + "grad_norm": 2.2875754833221436, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7193713784217834, + "num_tokens": 409223428.0, + "step": 16043 + }, + { + "epoch": 1.7619152207335822, + "grad_norm": 2.131474733352661, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7332719564437866, + "num_tokens": 409251634.0, + "step": 16044 + }, + { + "epoch": 1.762025038436196, + "grad_norm": 2.2144839763641357, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7053537368774414, + "num_tokens": 409278090.0, + "step": 16045 + }, + { + "epoch": 1.7621348561388097, + "grad_norm": 2.4666008949279785, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7400169372558594, + "num_tokens": 409299577.0, + "step": 16046 + }, + { + "epoch": 1.7622446738414232, + "grad_norm": 2.278764247894287, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.718201756477356, + "num_tokens": 409324371.0, + "step": 16047 + }, + { + "epoch": 1.7623544915440368, + "grad_norm": 2.3630266189575195, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7019039392471313, + "num_tokens": 409345687.0, + "step": 16048 + }, + { + "epoch": 1.7624643092466505, + "grad_norm": 2.530224084854126, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7266018986701965, + "num_tokens": 409367251.0, + "step": 16049 + }, + { + "epoch": 1.7625741269492643, + "grad_norm": 2.808948040008545, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7339071035385132, + "num_tokens": 409385017.0, + "step": 16050 + }, + { + "epoch": 1.7626839446518778, + "grad_norm": 2.4127309322357178, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7211721539497375, + "num_tokens": 409407177.0, + "step": 16051 + }, + { + "epoch": 1.7627937623544916, + "grad_norm": 2.2662339210510254, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7242920398712158, + "num_tokens": 409430730.0, + "step": 16052 + }, + { + "epoch": 1.7629035800571051, + "grad_norm": 1.9167195558547974, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7472978234291077, + "num_tokens": 409459718.0, + "step": 16053 + }, + { + "epoch": 1.7630133977597189, + "grad_norm": 2.490243673324585, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7022498250007629, + "num_tokens": 409481904.0, + "step": 16054 + }, + { + "epoch": 1.7631232154623326, + "grad_norm": 2.3429434299468994, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7194521427154541, + "num_tokens": 409505068.0, + "step": 16055 + }, + { + "epoch": 1.7632330331649462, + "grad_norm": 2.145932197570801, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7064370512962341, + "num_tokens": 409532547.0, + "step": 16056 + }, + { + "epoch": 1.7633428508675597, + "grad_norm": 2.2054336071014404, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7312318086624146, + "num_tokens": 409558287.0, + "step": 16057 + }, + { + "epoch": 1.7634526685701735, + "grad_norm": 2.0490365028381348, + "learning_rate": 1e-06, + "loss": 0.8266, + "mean_token_accuracy": 0.7398085594177246, + "num_tokens": 409587535.0, + "step": 16058 + }, + { + "epoch": 1.7635624862727872, + "grad_norm": 2.11118745803833, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7166954278945923, + "num_tokens": 409614402.0, + "step": 16059 + }, + { + "epoch": 1.763672303975401, + "grad_norm": 2.224620819091797, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7283730506896973, + "num_tokens": 409639923.0, + "step": 16060 + }, + { + "epoch": 1.7637821216780145, + "grad_norm": 2.3233933448791504, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7241508960723877, + "num_tokens": 409663584.0, + "step": 16061 + }, + { + "epoch": 1.763891939380628, + "grad_norm": 2.317582845687866, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7089270949363708, + "num_tokens": 409688802.0, + "step": 16062 + }, + { + "epoch": 1.7640017570832418, + "grad_norm": 2.292160987854004, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7296320199966431, + "num_tokens": 409713338.0, + "step": 16063 + }, + { + "epoch": 1.7641115747858556, + "grad_norm": 2.446260452270508, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7162603139877319, + "num_tokens": 409734421.0, + "step": 16064 + }, + { + "epoch": 1.764221392488469, + "grad_norm": 2.2119345664978027, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7311740517616272, + "num_tokens": 409760441.0, + "step": 16065 + }, + { + "epoch": 1.7643312101910829, + "grad_norm": 2.3397483825683594, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7204879522323608, + "num_tokens": 409783472.0, + "step": 16066 + }, + { + "epoch": 1.7644410278936964, + "grad_norm": 2.439267635345459, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7339292168617249, + "num_tokens": 409805968.0, + "step": 16067 + }, + { + "epoch": 1.7645508455963101, + "grad_norm": 2.379206895828247, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7215365171432495, + "num_tokens": 409829494.0, + "step": 16068 + }, + { + "epoch": 1.764660663298924, + "grad_norm": 1.9591896533966064, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.6972966194152832, + "num_tokens": 409860554.0, + "step": 16069 + }, + { + "epoch": 1.7647704810015374, + "grad_norm": 2.5440468788146973, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.7370540499687195, + "num_tokens": 409880778.0, + "step": 16070 + }, + { + "epoch": 1.764880298704151, + "grad_norm": 2.264875650405884, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7212488651275635, + "num_tokens": 409904959.0, + "step": 16071 + }, + { + "epoch": 1.7649901164067647, + "grad_norm": 2.609323501586914, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7215045690536499, + "num_tokens": 409933084.0, + "step": 16072 + }, + { + "epoch": 1.7650999341093785, + "grad_norm": 2.783799886703491, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7177276611328125, + "num_tokens": 409949745.0, + "step": 16073 + }, + { + "epoch": 1.7652097518119922, + "grad_norm": 2.385875940322876, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7214993238449097, + "num_tokens": 409973465.0, + "step": 16074 + }, + { + "epoch": 1.7653195695146058, + "grad_norm": 2.1450016498565674, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7056964635848999, + "num_tokens": 410001555.0, + "step": 16075 + }, + { + "epoch": 1.7654293872172193, + "grad_norm": 2.5832302570343018, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7301737666130066, + "num_tokens": 410022075.0, + "step": 16076 + }, + { + "epoch": 1.765539204919833, + "grad_norm": 2.152952194213867, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7345608472824097, + "num_tokens": 410049036.0, + "step": 16077 + }, + { + "epoch": 1.7656490226224468, + "grad_norm": 2.0929417610168457, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7382241487503052, + "num_tokens": 410078473.0, + "step": 16078 + }, + { + "epoch": 1.7657588403250604, + "grad_norm": 2.286572217941284, + "learning_rate": 1e-06, + "loss": 0.8198, + "mean_token_accuracy": 0.7404050230979919, + "num_tokens": 410101422.0, + "step": 16079 + }, + { + "epoch": 1.765868658027674, + "grad_norm": 2.383270740509033, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7261921167373657, + "num_tokens": 410125125.0, + "step": 16080 + }, + { + "epoch": 1.7659784757302877, + "grad_norm": 2.3117756843566895, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7061673998832703, + "num_tokens": 410150369.0, + "step": 16081 + }, + { + "epoch": 1.7660882934329014, + "grad_norm": 2.8149077892303467, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7119284868240356, + "num_tokens": 410167931.0, + "step": 16082 + }, + { + "epoch": 1.7661981111355152, + "grad_norm": 2.098750114440918, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7098000645637512, + "num_tokens": 410195268.0, + "step": 16083 + }, + { + "epoch": 1.7663079288381287, + "grad_norm": 2.3711276054382324, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7328956127166748, + "num_tokens": 410217318.0, + "step": 16084 + }, + { + "epoch": 1.7664177465407422, + "grad_norm": 2.233847141265869, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.734165370464325, + "num_tokens": 410241579.0, + "step": 16085 + }, + { + "epoch": 1.766527564243356, + "grad_norm": 2.035707712173462, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7000159025192261, + "num_tokens": 410273024.0, + "step": 16086 + }, + { + "epoch": 1.7666373819459698, + "grad_norm": 2.2881925106048584, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7040364146232605, + "num_tokens": 410298740.0, + "step": 16087 + }, + { + "epoch": 1.7667471996485835, + "grad_norm": 2.0184080600738525, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.6887615323066711, + "num_tokens": 410330111.0, + "step": 16088 + }, + { + "epoch": 1.766857017351197, + "grad_norm": 2.1692967414855957, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7181277275085449, + "num_tokens": 410355356.0, + "step": 16089 + }, + { + "epoch": 1.7669668350538106, + "grad_norm": 2.0957963466644287, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7276783585548401, + "num_tokens": 410383642.0, + "step": 16090 + }, + { + "epoch": 1.7670766527564243, + "grad_norm": 2.2863149642944336, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7139590978622437, + "num_tokens": 410408221.0, + "step": 16091 + }, + { + "epoch": 1.767186470459038, + "grad_norm": 2.1185693740844727, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.713301420211792, + "num_tokens": 410435235.0, + "step": 16092 + }, + { + "epoch": 1.7672962881616516, + "grad_norm": 2.110511302947998, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7121243476867676, + "num_tokens": 410463609.0, + "step": 16093 + }, + { + "epoch": 1.7674061058642652, + "grad_norm": 2.1577956676483154, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7280882000923157, + "num_tokens": 410489598.0, + "step": 16094 + }, + { + "epoch": 1.767515923566879, + "grad_norm": 2.305384635925293, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7261162400245667, + "num_tokens": 410514412.0, + "step": 16095 + }, + { + "epoch": 1.7676257412694927, + "grad_norm": 2.2527925968170166, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.714243471622467, + "num_tokens": 410538782.0, + "step": 16096 + }, + { + "epoch": 1.7677355589721064, + "grad_norm": 2.050236225128174, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7162120342254639, + "num_tokens": 410567399.0, + "step": 16097 + }, + { + "epoch": 1.76784537667472, + "grad_norm": 2.07389497756958, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7296535968780518, + "num_tokens": 410594695.0, + "step": 16098 + }, + { + "epoch": 1.7679551943773335, + "grad_norm": 2.257040500640869, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7100481986999512, + "num_tokens": 410619366.0, + "step": 16099 + }, + { + "epoch": 1.7680650120799473, + "grad_norm": 2.2587780952453613, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7318377494812012, + "num_tokens": 410645384.0, + "step": 16100 + }, + { + "epoch": 1.768174829782561, + "grad_norm": 2.4776782989501953, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7197228670120239, + "num_tokens": 410666384.0, + "step": 16101 + }, + { + "epoch": 1.7682846474851748, + "grad_norm": 2.0705032348632812, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7237288951873779, + "num_tokens": 410694819.0, + "step": 16102 + }, + { + "epoch": 1.7683944651877883, + "grad_norm": 2.085935115814209, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7057026624679565, + "num_tokens": 410724564.0, + "step": 16103 + }, + { + "epoch": 1.7685042828904018, + "grad_norm": 2.322906255722046, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7252193689346313, + "num_tokens": 410748269.0, + "step": 16104 + }, + { + "epoch": 1.7686141005930156, + "grad_norm": 2.3532586097717285, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7291660308837891, + "num_tokens": 410772768.0, + "step": 16105 + }, + { + "epoch": 1.7687239182956294, + "grad_norm": 2.1647679805755615, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7293568253517151, + "num_tokens": 410799298.0, + "step": 16106 + }, + { + "epoch": 1.768833735998243, + "grad_norm": 2.3708975315093994, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6955134868621826, + "num_tokens": 410824111.0, + "step": 16107 + }, + { + "epoch": 1.7689435537008564, + "grad_norm": 2.228647232055664, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.727792501449585, + "num_tokens": 410849590.0, + "step": 16108 + }, + { + "epoch": 1.7690533714034702, + "grad_norm": 2.678218364715576, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7310856580734253, + "num_tokens": 410868580.0, + "step": 16109 + }, + { + "epoch": 1.769163189106084, + "grad_norm": 2.1827011108398438, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.699984610080719, + "num_tokens": 410895330.0, + "step": 16110 + }, + { + "epoch": 1.7692730068086977, + "grad_norm": 2.2403178215026855, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.714368462562561, + "num_tokens": 410921647.0, + "step": 16111 + }, + { + "epoch": 1.7693828245113112, + "grad_norm": 2.0953173637390137, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7229294776916504, + "num_tokens": 410948377.0, + "step": 16112 + }, + { + "epoch": 1.7694926422139248, + "grad_norm": 2.5005900859832764, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7087477445602417, + "num_tokens": 410969890.0, + "step": 16113 + }, + { + "epoch": 1.7696024599165385, + "grad_norm": 2.50297212600708, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7058476209640503, + "num_tokens": 410991839.0, + "step": 16114 + }, + { + "epoch": 1.7697122776191523, + "grad_norm": 2.276641845703125, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7083745002746582, + "num_tokens": 411018149.0, + "step": 16115 + }, + { + "epoch": 1.7698220953217658, + "grad_norm": 2.535836935043335, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7303305864334106, + "num_tokens": 411039334.0, + "step": 16116 + }, + { + "epoch": 1.7699319130243796, + "grad_norm": 2.4072697162628174, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7305338382720947, + "num_tokens": 411061278.0, + "step": 16117 + }, + { + "epoch": 1.7700417307269931, + "grad_norm": 2.2066550254821777, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7170277833938599, + "num_tokens": 411086867.0, + "step": 16118 + }, + { + "epoch": 1.7701515484296069, + "grad_norm": 2.1563992500305176, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7176151871681213, + "num_tokens": 411113949.0, + "step": 16119 + }, + { + "epoch": 1.7702613661322206, + "grad_norm": 2.1512017250061035, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7165127992630005, + "num_tokens": 411142161.0, + "step": 16120 + }, + { + "epoch": 1.7703711838348342, + "grad_norm": 2.4011242389678955, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7226688861846924, + "num_tokens": 411164798.0, + "step": 16121 + }, + { + "epoch": 1.7704810015374477, + "grad_norm": 2.2530770301818848, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.741127610206604, + "num_tokens": 411189786.0, + "step": 16122 + }, + { + "epoch": 1.7705908192400615, + "grad_norm": 2.5900049209594727, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7426159381866455, + "num_tokens": 411209692.0, + "step": 16123 + }, + { + "epoch": 1.7707006369426752, + "grad_norm": 2.5049550533294678, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7327935695648193, + "num_tokens": 411230123.0, + "step": 16124 + }, + { + "epoch": 1.770810454645289, + "grad_norm": 2.177459239959717, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7203202247619629, + "num_tokens": 411258796.0, + "step": 16125 + }, + { + "epoch": 1.7709202723479025, + "grad_norm": 1.9158856868743896, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7172858119010925, + "num_tokens": 411293279.0, + "step": 16126 + }, + { + "epoch": 1.771030090050516, + "grad_norm": 2.0594940185546875, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7163151502609253, + "num_tokens": 411322420.0, + "step": 16127 + }, + { + "epoch": 1.7711399077531298, + "grad_norm": 2.0224077701568604, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.6944048404693604, + "num_tokens": 411352624.0, + "step": 16128 + }, + { + "epoch": 1.7712497254557436, + "grad_norm": 2.311612367630005, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7385042905807495, + "num_tokens": 411376287.0, + "step": 16129 + }, + { + "epoch": 1.771359543158357, + "grad_norm": 2.0706915855407715, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7175426483154297, + "num_tokens": 411406126.0, + "step": 16130 + }, + { + "epoch": 1.7714693608609708, + "grad_norm": 2.2893528938293457, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.729779839515686, + "num_tokens": 411430492.0, + "step": 16131 + }, + { + "epoch": 1.7715791785635844, + "grad_norm": 2.2558815479278564, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7121464014053345, + "num_tokens": 411456765.0, + "step": 16132 + }, + { + "epoch": 1.7716889962661981, + "grad_norm": 2.4278688430786133, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7304904460906982, + "num_tokens": 411477330.0, + "step": 16133 + }, + { + "epoch": 1.771798813968812, + "grad_norm": 2.1985628604888916, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7247530221939087, + "num_tokens": 411504469.0, + "step": 16134 + }, + { + "epoch": 1.7719086316714254, + "grad_norm": 2.308763265609741, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.7395011186599731, + "num_tokens": 411529506.0, + "step": 16135 + }, + { + "epoch": 1.772018449374039, + "grad_norm": 2.370868682861328, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7174303531646729, + "num_tokens": 411552419.0, + "step": 16136 + }, + { + "epoch": 1.7721282670766527, + "grad_norm": 2.576995849609375, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7293630242347717, + "num_tokens": 411572980.0, + "step": 16137 + }, + { + "epoch": 1.7722380847792665, + "grad_norm": 2.238806962966919, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7141042947769165, + "num_tokens": 411599657.0, + "step": 16138 + }, + { + "epoch": 1.7723479024818802, + "grad_norm": 2.3675222396850586, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7428813576698303, + "num_tokens": 411622417.0, + "step": 16139 + }, + { + "epoch": 1.7724577201844938, + "grad_norm": 2.385464906692505, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7323463559150696, + "num_tokens": 411644715.0, + "step": 16140 + }, + { + "epoch": 1.7725675378871073, + "grad_norm": 2.2511110305786133, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7371294498443604, + "num_tokens": 411668941.0, + "step": 16141 + }, + { + "epoch": 1.772677355589721, + "grad_norm": 2.7204558849334717, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.7375008463859558, + "num_tokens": 411686758.0, + "step": 16142 + }, + { + "epoch": 1.7727871732923348, + "grad_norm": 2.062126398086548, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7076843976974487, + "num_tokens": 411714434.0, + "step": 16143 + }, + { + "epoch": 1.7728969909949484, + "grad_norm": 2.364036798477173, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7138713002204895, + "num_tokens": 411737358.0, + "step": 16144 + }, + { + "epoch": 1.7730068086975619, + "grad_norm": 2.114912748336792, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7305968403816223, + "num_tokens": 411765064.0, + "step": 16145 + }, + { + "epoch": 1.7731166264001756, + "grad_norm": 2.509826898574829, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7338978052139282, + "num_tokens": 411786516.0, + "step": 16146 + }, + { + "epoch": 1.7732264441027894, + "grad_norm": 2.288804292678833, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7208057641983032, + "num_tokens": 411811922.0, + "step": 16147 + }, + { + "epoch": 1.7733362618054032, + "grad_norm": 1.994133710861206, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6992566585540771, + "num_tokens": 411843819.0, + "step": 16148 + }, + { + "epoch": 1.7734460795080167, + "grad_norm": 1.9591859579086304, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7079551815986633, + "num_tokens": 411877423.0, + "step": 16149 + }, + { + "epoch": 1.7735558972106302, + "grad_norm": 2.620695114135742, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7355202436447144, + "num_tokens": 411897804.0, + "step": 16150 + }, + { + "epoch": 1.773665714913244, + "grad_norm": 2.0554723739624023, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7348253726959229, + "num_tokens": 411926719.0, + "step": 16151 + }, + { + "epoch": 1.7737755326158577, + "grad_norm": 2.0418434143066406, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7349109649658203, + "num_tokens": 411955004.0, + "step": 16152 + }, + { + "epoch": 1.7738853503184715, + "grad_norm": 2.4323909282684326, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7113596200942993, + "num_tokens": 411979025.0, + "step": 16153 + }, + { + "epoch": 1.773995168021085, + "grad_norm": 2.2602627277374268, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7253661155700684, + "num_tokens": 412002372.0, + "step": 16154 + }, + { + "epoch": 1.7741049857236986, + "grad_norm": 2.034558057785034, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6870622634887695, + "num_tokens": 412035426.0, + "step": 16155 + }, + { + "epoch": 1.7742148034263123, + "grad_norm": 2.293801784515381, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7040139436721802, + "num_tokens": 412060767.0, + "step": 16156 + }, + { + "epoch": 1.774324621128926, + "grad_norm": 2.2443721294403076, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7097734808921814, + "num_tokens": 412084926.0, + "step": 16157 + }, + { + "epoch": 1.7744344388315396, + "grad_norm": 1.9932968616485596, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7247551083564758, + "num_tokens": 412116133.0, + "step": 16158 + }, + { + "epoch": 1.7745442565341532, + "grad_norm": 2.0286803245544434, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7244491577148438, + "num_tokens": 412147412.0, + "step": 16159 + }, + { + "epoch": 1.774654074236767, + "grad_norm": 2.3181726932525635, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7253318428993225, + "num_tokens": 412171459.0, + "step": 16160 + }, + { + "epoch": 1.7747638919393807, + "grad_norm": 1.9606400728225708, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7174519300460815, + "num_tokens": 412204937.0, + "step": 16161 + }, + { + "epoch": 1.7748737096419944, + "grad_norm": 2.586665630340576, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7454122304916382, + "num_tokens": 412223889.0, + "step": 16162 + }, + { + "epoch": 1.774983527344608, + "grad_norm": 2.1213178634643555, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7187562584877014, + "num_tokens": 412252616.0, + "step": 16163 + }, + { + "epoch": 1.7750933450472215, + "grad_norm": 2.2828917503356934, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.726731538772583, + "num_tokens": 412277966.0, + "step": 16164 + }, + { + "epoch": 1.7752031627498353, + "grad_norm": 1.7852360010147095, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7134866714477539, + "num_tokens": 412313818.0, + "step": 16165 + }, + { + "epoch": 1.775312980452449, + "grad_norm": 2.397996187210083, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7086864709854126, + "num_tokens": 412338755.0, + "step": 16166 + }, + { + "epoch": 1.7754227981550625, + "grad_norm": 2.570274591445923, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7349791526794434, + "num_tokens": 412358743.0, + "step": 16167 + }, + { + "epoch": 1.7755326158576763, + "grad_norm": 2.374403238296509, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7098348140716553, + "num_tokens": 412383627.0, + "step": 16168 + }, + { + "epoch": 1.7756424335602898, + "grad_norm": 2.26478910446167, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7468547821044922, + "num_tokens": 412407321.0, + "step": 16169 + }, + { + "epoch": 1.7757522512629036, + "grad_norm": 2.298171281814575, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7214301824569702, + "num_tokens": 412431744.0, + "step": 16170 + }, + { + "epoch": 1.7758620689655173, + "grad_norm": 2.134157657623291, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7011496424674988, + "num_tokens": 412461456.0, + "step": 16171 + }, + { + "epoch": 1.7759718866681309, + "grad_norm": 2.0506293773651123, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7257195711135864, + "num_tokens": 412490178.0, + "step": 16172 + }, + { + "epoch": 1.7760817043707444, + "grad_norm": 2.3771326541900635, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7293877601623535, + "num_tokens": 412512081.0, + "step": 16173 + }, + { + "epoch": 1.7761915220733582, + "grad_norm": 2.1407601833343506, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7018225193023682, + "num_tokens": 412539227.0, + "step": 16174 + }, + { + "epoch": 1.776301339775972, + "grad_norm": 2.618450164794922, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7259543538093567, + "num_tokens": 412560199.0, + "step": 16175 + }, + { + "epoch": 1.7764111574785857, + "grad_norm": 2.2914083003997803, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7293205857276917, + "num_tokens": 412584406.0, + "step": 16176 + }, + { + "epoch": 1.7765209751811992, + "grad_norm": 2.2196714878082275, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7029421329498291, + "num_tokens": 412612044.0, + "step": 16177 + }, + { + "epoch": 1.7766307928838128, + "grad_norm": 2.338007688522339, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7229962944984436, + "num_tokens": 412635356.0, + "step": 16178 + }, + { + "epoch": 1.7767406105864265, + "grad_norm": 2.202479839324951, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.696262001991272, + "num_tokens": 412662758.0, + "step": 16179 + }, + { + "epoch": 1.7768504282890403, + "grad_norm": 2.1336135864257812, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7318054437637329, + "num_tokens": 412689664.0, + "step": 16180 + }, + { + "epoch": 1.7769602459916538, + "grad_norm": 2.2661046981811523, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7161325216293335, + "num_tokens": 412714839.0, + "step": 16181 + }, + { + "epoch": 1.7770700636942676, + "grad_norm": 2.197817802429199, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7238078117370605, + "num_tokens": 412741746.0, + "step": 16182 + }, + { + "epoch": 1.777179881396881, + "grad_norm": 2.070653200149536, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7111786603927612, + "num_tokens": 412769727.0, + "step": 16183 + }, + { + "epoch": 1.7772896990994949, + "grad_norm": 2.2316038608551025, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7168378829956055, + "num_tokens": 412794465.0, + "step": 16184 + }, + { + "epoch": 1.7773995168021086, + "grad_norm": 1.8999141454696655, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7013938426971436, + "num_tokens": 412830221.0, + "step": 16185 + }, + { + "epoch": 1.7775093345047221, + "grad_norm": 2.0609078407287598, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7095523476600647, + "num_tokens": 412859665.0, + "step": 16186 + }, + { + "epoch": 1.7776191522073357, + "grad_norm": 2.064504384994507, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.7326074838638306, + "num_tokens": 412886897.0, + "step": 16187 + }, + { + "epoch": 1.7777289699099494, + "grad_norm": 2.7318224906921387, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7448748350143433, + "num_tokens": 412904064.0, + "step": 16188 + }, + { + "epoch": 1.7778387876125632, + "grad_norm": 2.491398334503174, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7238099575042725, + "num_tokens": 412924485.0, + "step": 16189 + }, + { + "epoch": 1.777948605315177, + "grad_norm": 2.3687870502471924, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.721379280090332, + "num_tokens": 412947241.0, + "step": 16190 + }, + { + "epoch": 1.7780584230177905, + "grad_norm": 2.207408905029297, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7357994914054871, + "num_tokens": 412976011.0, + "step": 16191 + }, + { + "epoch": 1.778168240720404, + "grad_norm": 2.4504575729370117, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7257530689239502, + "num_tokens": 412998478.0, + "step": 16192 + }, + { + "epoch": 1.7782780584230178, + "grad_norm": 2.180558681488037, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.6971071362495422, + "num_tokens": 413027129.0, + "step": 16193 + }, + { + "epoch": 1.7783878761256315, + "grad_norm": 2.2200448513031006, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7088524103164673, + "num_tokens": 413052431.0, + "step": 16194 + }, + { + "epoch": 1.778497693828245, + "grad_norm": 2.1597073078155518, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7067943811416626, + "num_tokens": 413078877.0, + "step": 16195 + }, + { + "epoch": 1.7786075115308586, + "grad_norm": 2.0460145473480225, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7359538674354553, + "num_tokens": 413105751.0, + "step": 16196 + }, + { + "epoch": 1.7787173292334724, + "grad_norm": 2.1013662815093994, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.6949678659439087, + "num_tokens": 413136880.0, + "step": 16197 + }, + { + "epoch": 1.7788271469360861, + "grad_norm": 2.30380916595459, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7266165018081665, + "num_tokens": 413160724.0, + "step": 16198 + }, + { + "epoch": 1.7789369646386999, + "grad_norm": 2.1865224838256836, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.6997243165969849, + "num_tokens": 413189259.0, + "step": 16199 + }, + { + "epoch": 1.7790467823413134, + "grad_norm": 2.2489137649536133, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7090458869934082, + "num_tokens": 413213481.0, + "step": 16200 + }, + { + "epoch": 1.779156600043927, + "grad_norm": 2.6155636310577393, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7381004095077515, + "num_tokens": 413232079.0, + "step": 16201 + }, + { + "epoch": 1.7792664177465407, + "grad_norm": 2.6620564460754395, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7329235076904297, + "num_tokens": 413253224.0, + "step": 16202 + }, + { + "epoch": 1.7793762354491545, + "grad_norm": 2.266422986984253, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7140452861785889, + "num_tokens": 413278983.0, + "step": 16203 + }, + { + "epoch": 1.7794860531517682, + "grad_norm": 2.1861000061035156, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7210355997085571, + "num_tokens": 413303605.0, + "step": 16204 + }, + { + "epoch": 1.7795958708543818, + "grad_norm": 1.9868243932724, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.726140022277832, + "num_tokens": 413334670.0, + "step": 16205 + }, + { + "epoch": 1.7797056885569953, + "grad_norm": 2.2332513332366943, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7285857200622559, + "num_tokens": 413360000.0, + "step": 16206 + }, + { + "epoch": 1.779815506259609, + "grad_norm": 2.0847220420837402, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7086912989616394, + "num_tokens": 413389751.0, + "step": 16207 + }, + { + "epoch": 1.7799253239622228, + "grad_norm": 2.486555337905884, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7195967435836792, + "num_tokens": 413412201.0, + "step": 16208 + }, + { + "epoch": 1.7800351416648363, + "grad_norm": 2.38962459564209, + "learning_rate": 1e-06, + "loss": 0.7817, + "mean_token_accuracy": 0.7570748329162598, + "num_tokens": 413434868.0, + "step": 16209 + }, + { + "epoch": 1.7801449593674499, + "grad_norm": 2.323829412460327, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7167561054229736, + "num_tokens": 413459132.0, + "step": 16210 + }, + { + "epoch": 1.7802547770700636, + "grad_norm": 2.226926326751709, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.723160445690155, + "num_tokens": 413485788.0, + "step": 16211 + }, + { + "epoch": 1.7803645947726774, + "grad_norm": 2.2355940341949463, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7396094799041748, + "num_tokens": 413509936.0, + "step": 16212 + }, + { + "epoch": 1.7804744124752911, + "grad_norm": 2.277491807937622, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7146053314208984, + "num_tokens": 413533902.0, + "step": 16213 + }, + { + "epoch": 1.7805842301779047, + "grad_norm": 2.0400500297546387, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7020144462585449, + "num_tokens": 413565811.0, + "step": 16214 + }, + { + "epoch": 1.7806940478805182, + "grad_norm": 2.4276726245880127, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7140458822250366, + "num_tokens": 413590610.0, + "step": 16215 + }, + { + "epoch": 1.780803865583132, + "grad_norm": 2.8001763820648193, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7364756464958191, + "num_tokens": 413608611.0, + "step": 16216 + }, + { + "epoch": 1.7809136832857457, + "grad_norm": 2.425597906112671, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7010502219200134, + "num_tokens": 413631244.0, + "step": 16217 + }, + { + "epoch": 1.7810235009883595, + "grad_norm": 2.2845661640167236, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7311223745346069, + "num_tokens": 413655447.0, + "step": 16218 + }, + { + "epoch": 1.781133318690973, + "grad_norm": 2.1347618103027344, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7148178815841675, + "num_tokens": 413684960.0, + "step": 16219 + }, + { + "epoch": 1.7812431363935866, + "grad_norm": 2.09238600730896, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.7007899284362793, + "num_tokens": 413715237.0, + "step": 16220 + }, + { + "epoch": 1.7813529540962003, + "grad_norm": 2.4887712001800537, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7142543792724609, + "num_tokens": 413737410.0, + "step": 16221 + }, + { + "epoch": 1.781462771798814, + "grad_norm": 2.078413248062134, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6908746957778931, + "num_tokens": 413766663.0, + "step": 16222 + }, + { + "epoch": 1.7815725895014276, + "grad_norm": 2.328857898712158, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7166810035705566, + "num_tokens": 413789416.0, + "step": 16223 + }, + { + "epoch": 1.7816824072040411, + "grad_norm": 2.1915841102600098, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7321521043777466, + "num_tokens": 413814415.0, + "step": 16224 + }, + { + "epoch": 1.781792224906655, + "grad_norm": 2.235870838165283, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7048976421356201, + "num_tokens": 413840370.0, + "step": 16225 + }, + { + "epoch": 1.7819020426092687, + "grad_norm": 2.766153335571289, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7316486239433289, + "num_tokens": 413858169.0, + "step": 16226 + }, + { + "epoch": 1.7820118603118824, + "grad_norm": 2.0951988697052, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7247700095176697, + "num_tokens": 413886545.0, + "step": 16227 + }, + { + "epoch": 1.782121678014496, + "grad_norm": 2.2190089225769043, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7129373550415039, + "num_tokens": 413913238.0, + "step": 16228 + }, + { + "epoch": 1.7822314957171095, + "grad_norm": 2.122204303741455, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7355855703353882, + "num_tokens": 413939805.0, + "step": 16229 + }, + { + "epoch": 1.7823413134197232, + "grad_norm": 2.4150447845458984, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7339763641357422, + "num_tokens": 413963045.0, + "step": 16230 + }, + { + "epoch": 1.782451131122337, + "grad_norm": 2.296884536743164, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7215370535850525, + "num_tokens": 413986164.0, + "step": 16231 + }, + { + "epoch": 1.7825609488249505, + "grad_norm": 2.4859378337860107, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7090851068496704, + "num_tokens": 414008253.0, + "step": 16232 + }, + { + "epoch": 1.7826707665275643, + "grad_norm": 2.190992832183838, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7186865210533142, + "num_tokens": 414034016.0, + "step": 16233 + }, + { + "epoch": 1.7827805842301778, + "grad_norm": 2.4568753242492676, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7316655516624451, + "num_tokens": 414057027.0, + "step": 16234 + }, + { + "epoch": 1.7828904019327916, + "grad_norm": 2.1951730251312256, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7204725742340088, + "num_tokens": 414083358.0, + "step": 16235 + }, + { + "epoch": 1.7830002196354053, + "grad_norm": 2.139549493789673, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7011115550994873, + "num_tokens": 414111466.0, + "step": 16236 + }, + { + "epoch": 1.7831100373380189, + "grad_norm": 2.1211047172546387, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7132232189178467, + "num_tokens": 414138832.0, + "step": 16237 + }, + { + "epoch": 1.7832198550406324, + "grad_norm": 2.1338109970092773, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7235751152038574, + "num_tokens": 414165251.0, + "step": 16238 + }, + { + "epoch": 1.7833296727432462, + "grad_norm": 2.282820224761963, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7048400640487671, + "num_tokens": 414190084.0, + "step": 16239 + }, + { + "epoch": 1.78343949044586, + "grad_norm": 2.227325439453125, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.7022119760513306, + "num_tokens": 414216738.0, + "step": 16240 + }, + { + "epoch": 1.7835493081484737, + "grad_norm": 2.532036542892456, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7162035703659058, + "num_tokens": 414239524.0, + "step": 16241 + }, + { + "epoch": 1.7836591258510872, + "grad_norm": 2.2129883766174316, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7297552824020386, + "num_tokens": 414266135.0, + "step": 16242 + }, + { + "epoch": 1.7837689435537007, + "grad_norm": 2.290807008743286, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7355968952178955, + "num_tokens": 414291560.0, + "step": 16243 + }, + { + "epoch": 1.7838787612563145, + "grad_norm": 2.44667387008667, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7244240045547485, + "num_tokens": 414313970.0, + "step": 16244 + }, + { + "epoch": 1.7839885789589283, + "grad_norm": 2.2942473888397217, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7393447160720825, + "num_tokens": 414337286.0, + "step": 16245 + }, + { + "epoch": 1.7840983966615418, + "grad_norm": 1.9442328214645386, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6913658380508423, + "num_tokens": 414369649.0, + "step": 16246 + }, + { + "epoch": 1.7842082143641556, + "grad_norm": 2.072075366973877, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7173032760620117, + "num_tokens": 414398372.0, + "step": 16247 + }, + { + "epoch": 1.784318032066769, + "grad_norm": 2.0638058185577393, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7106770277023315, + "num_tokens": 414429640.0, + "step": 16248 + }, + { + "epoch": 1.7844278497693828, + "grad_norm": 2.1907060146331787, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7316364645957947, + "num_tokens": 414458550.0, + "step": 16249 + }, + { + "epoch": 1.7845376674719966, + "grad_norm": 2.488825559616089, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7274951338768005, + "num_tokens": 414479498.0, + "step": 16250 + }, + { + "epoch": 1.7846474851746101, + "grad_norm": 2.177098274230957, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.735442578792572, + "num_tokens": 414505930.0, + "step": 16251 + }, + { + "epoch": 1.7847573028772237, + "grad_norm": 2.235405206680298, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7234067320823669, + "num_tokens": 414531354.0, + "step": 16252 + }, + { + "epoch": 1.7848671205798374, + "grad_norm": 2.5427629947662354, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7400606274604797, + "num_tokens": 414552445.0, + "step": 16253 + }, + { + "epoch": 1.7849769382824512, + "grad_norm": 2.3892533779144287, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7365207076072693, + "num_tokens": 414573878.0, + "step": 16254 + }, + { + "epoch": 1.785086755985065, + "grad_norm": 2.307774782180786, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.7520148754119873, + "num_tokens": 414595543.0, + "step": 16255 + }, + { + "epoch": 1.7851965736876785, + "grad_norm": 2.333867311477661, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7066847681999207, + "num_tokens": 414620866.0, + "step": 16256 + }, + { + "epoch": 1.785306391390292, + "grad_norm": 2.1860673427581787, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7302860617637634, + "num_tokens": 414646015.0, + "step": 16257 + }, + { + "epoch": 1.7854162090929058, + "grad_norm": 2.066957473754883, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.732585072517395, + "num_tokens": 414674690.0, + "step": 16258 + }, + { + "epoch": 1.7855260267955195, + "grad_norm": 2.2233777046203613, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7221865057945251, + "num_tokens": 414702775.0, + "step": 16259 + }, + { + "epoch": 1.785635844498133, + "grad_norm": 2.220585584640503, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7237169146537781, + "num_tokens": 414728075.0, + "step": 16260 + }, + { + "epoch": 1.7857456622007466, + "grad_norm": 2.3033618927001953, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.725392758846283, + "num_tokens": 414753663.0, + "step": 16261 + }, + { + "epoch": 1.7858554799033604, + "grad_norm": 2.2861459255218506, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7150890231132507, + "num_tokens": 414779929.0, + "step": 16262 + }, + { + "epoch": 1.7859652976059741, + "grad_norm": 2.427105665206909, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7208353281021118, + "num_tokens": 414802112.0, + "step": 16263 + }, + { + "epoch": 1.7860751153085879, + "grad_norm": 2.347160577774048, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7525526285171509, + "num_tokens": 414824958.0, + "step": 16264 + }, + { + "epoch": 1.7861849330112014, + "grad_norm": 2.0672171115875244, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7343233823776245, + "num_tokens": 414853663.0, + "step": 16265 + }, + { + "epoch": 1.786294750713815, + "grad_norm": 2.527392864227295, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7348814010620117, + "num_tokens": 414875691.0, + "step": 16266 + }, + { + "epoch": 1.7864045684164287, + "grad_norm": 2.3975343704223633, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7262383699417114, + "num_tokens": 414899075.0, + "step": 16267 + }, + { + "epoch": 1.7865143861190425, + "grad_norm": 2.2172539234161377, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7128994464874268, + "num_tokens": 414924677.0, + "step": 16268 + }, + { + "epoch": 1.7866242038216562, + "grad_norm": 2.168079137802124, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.711761474609375, + "num_tokens": 414952427.0, + "step": 16269 + }, + { + "epoch": 1.7867340215242697, + "grad_norm": 2.652763605117798, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7417364120483398, + "num_tokens": 414970076.0, + "step": 16270 + }, + { + "epoch": 1.7868438392268833, + "grad_norm": 2.744678020477295, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7284265756607056, + "num_tokens": 414989295.0, + "step": 16271 + }, + { + "epoch": 1.786953656929497, + "grad_norm": 2.200359582901001, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.6970912218093872, + "num_tokens": 415014183.0, + "step": 16272 + }, + { + "epoch": 1.7870634746321108, + "grad_norm": 2.154628038406372, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7170369625091553, + "num_tokens": 415040663.0, + "step": 16273 + }, + { + "epoch": 1.7871732923347243, + "grad_norm": 2.3200325965881348, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7307732105255127, + "num_tokens": 415063205.0, + "step": 16274 + }, + { + "epoch": 1.7872831100373379, + "grad_norm": 2.2575786113739014, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7053529024124146, + "num_tokens": 415089361.0, + "step": 16275 + }, + { + "epoch": 1.7873929277399516, + "grad_norm": 2.170485496520996, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7202969789505005, + "num_tokens": 415116908.0, + "step": 16276 + }, + { + "epoch": 1.7875027454425654, + "grad_norm": 2.318610906600952, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.71091228723526, + "num_tokens": 415140175.0, + "step": 16277 + }, + { + "epoch": 1.7876125631451791, + "grad_norm": 2.296586513519287, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7286590337753296, + "num_tokens": 415161414.0, + "step": 16278 + }, + { + "epoch": 1.7877223808477927, + "grad_norm": 1.9920449256896973, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7286491990089417, + "num_tokens": 415190512.0, + "step": 16279 + }, + { + "epoch": 1.7878321985504062, + "grad_norm": 2.1075518131256104, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.6939610242843628, + "num_tokens": 415221850.0, + "step": 16280 + }, + { + "epoch": 1.78794201625302, + "grad_norm": 2.4505536556243896, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.69841068983078, + "num_tokens": 415245945.0, + "step": 16281 + }, + { + "epoch": 1.7880518339556337, + "grad_norm": 2.3721063137054443, + "learning_rate": 1e-06, + "loss": 0.7626, + "mean_token_accuracy": 0.7557816505432129, + "num_tokens": 415267585.0, + "step": 16282 + }, + { + "epoch": 1.7881616516582475, + "grad_norm": 2.1012678146362305, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7171425819396973, + "num_tokens": 415295761.0, + "step": 16283 + }, + { + "epoch": 1.788271469360861, + "grad_norm": 2.397181987762451, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7351160645484924, + "num_tokens": 415316470.0, + "step": 16284 + }, + { + "epoch": 1.7883812870634745, + "grad_norm": 2.3120338916778564, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7016764879226685, + "num_tokens": 415343584.0, + "step": 16285 + }, + { + "epoch": 1.7884911047660883, + "grad_norm": 2.305769205093384, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7237083911895752, + "num_tokens": 415369442.0, + "step": 16286 + }, + { + "epoch": 1.788600922468702, + "grad_norm": 1.9045006036758423, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.716158926486969, + "num_tokens": 415403467.0, + "step": 16287 + }, + { + "epoch": 1.7887107401713156, + "grad_norm": 2.435882091522217, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7313297390937805, + "num_tokens": 415425815.0, + "step": 16288 + }, + { + "epoch": 1.7888205578739291, + "grad_norm": 2.111420154571533, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7129709720611572, + "num_tokens": 415452607.0, + "step": 16289 + }, + { + "epoch": 1.7889303755765429, + "grad_norm": 2.266425848007202, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7124526500701904, + "num_tokens": 415477537.0, + "step": 16290 + }, + { + "epoch": 1.7890401932791566, + "grad_norm": 2.0281145572662354, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.6943868398666382, + "num_tokens": 415510690.0, + "step": 16291 + }, + { + "epoch": 1.7891500109817704, + "grad_norm": 2.3788435459136963, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7154342532157898, + "num_tokens": 415532980.0, + "step": 16292 + }, + { + "epoch": 1.789259828684384, + "grad_norm": 2.378044843673706, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7077251672744751, + "num_tokens": 415557080.0, + "step": 16293 + }, + { + "epoch": 1.7893696463869975, + "grad_norm": 2.471613645553589, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7280668616294861, + "num_tokens": 415577699.0, + "step": 16294 + }, + { + "epoch": 1.7894794640896112, + "grad_norm": 2.6322262287139893, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.706843376159668, + "num_tokens": 415598949.0, + "step": 16295 + }, + { + "epoch": 1.789589281792225, + "grad_norm": 2.0560238361358643, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.703606128692627, + "num_tokens": 415627692.0, + "step": 16296 + }, + { + "epoch": 1.7896990994948385, + "grad_norm": 2.3132598400115967, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7300751805305481, + "num_tokens": 415652168.0, + "step": 16297 + }, + { + "epoch": 1.7898089171974523, + "grad_norm": 2.2120630741119385, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.691980242729187, + "num_tokens": 415681375.0, + "step": 16298 + }, + { + "epoch": 1.7899187349000658, + "grad_norm": 2.1256821155548096, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.708989679813385, + "num_tokens": 415709328.0, + "step": 16299 + }, + { + "epoch": 1.7900285526026796, + "grad_norm": 2.3059167861938477, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7088476419448853, + "num_tokens": 415733935.0, + "step": 16300 + }, + { + "epoch": 1.7901383703052933, + "grad_norm": 2.1372084617614746, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7264163494110107, + "num_tokens": 415760548.0, + "step": 16301 + }, + { + "epoch": 1.7902481880079069, + "grad_norm": 2.110243797302246, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7279907464981079, + "num_tokens": 415789262.0, + "step": 16302 + }, + { + "epoch": 1.7903580057105204, + "grad_norm": 2.7266383171081543, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7366948127746582, + "num_tokens": 415806935.0, + "step": 16303 + }, + { + "epoch": 1.7904678234131342, + "grad_norm": 2.123185634613037, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.706452488899231, + "num_tokens": 415837033.0, + "step": 16304 + }, + { + "epoch": 1.790577641115748, + "grad_norm": 2.227552652359009, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7257295250892639, + "num_tokens": 415864216.0, + "step": 16305 + }, + { + "epoch": 1.7906874588183617, + "grad_norm": 2.3842666149139404, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7377088069915771, + "num_tokens": 415886658.0, + "step": 16306 + }, + { + "epoch": 1.7907972765209752, + "grad_norm": 2.04935622215271, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7140042185783386, + "num_tokens": 415915994.0, + "step": 16307 + }, + { + "epoch": 1.7909070942235887, + "grad_norm": 2.177478075027466, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7084163427352905, + "num_tokens": 415946213.0, + "step": 16308 + }, + { + "epoch": 1.7910169119262025, + "grad_norm": 2.343305826187134, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7267114520072937, + "num_tokens": 415970063.0, + "step": 16309 + }, + { + "epoch": 1.7911267296288162, + "grad_norm": 2.3220365047454834, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7096621990203857, + "num_tokens": 415994686.0, + "step": 16310 + }, + { + "epoch": 1.7912365473314298, + "grad_norm": 2.612920045852661, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7346097826957703, + "num_tokens": 416014086.0, + "step": 16311 + }, + { + "epoch": 1.7913463650340435, + "grad_norm": 2.260432243347168, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7320238351821899, + "num_tokens": 416038836.0, + "step": 16312 + }, + { + "epoch": 1.791456182736657, + "grad_norm": 2.279531717300415, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7049790620803833, + "num_tokens": 416062360.0, + "step": 16313 + }, + { + "epoch": 1.7915660004392708, + "grad_norm": 2.414414167404175, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.727908730506897, + "num_tokens": 416084882.0, + "step": 16314 + }, + { + "epoch": 1.7916758181418846, + "grad_norm": 2.3398525714874268, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7271537780761719, + "num_tokens": 416107984.0, + "step": 16315 + }, + { + "epoch": 1.7917856358444981, + "grad_norm": 2.852993965148926, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7375016212463379, + "num_tokens": 416124812.0, + "step": 16316 + }, + { + "epoch": 1.7918954535471117, + "grad_norm": 2.7076382637023926, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7224727869033813, + "num_tokens": 416146248.0, + "step": 16317 + }, + { + "epoch": 1.7920052712497254, + "grad_norm": 2.1658737659454346, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.6952196359634399, + "num_tokens": 416173215.0, + "step": 16318 + }, + { + "epoch": 1.7921150889523392, + "grad_norm": 1.968201994895935, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7166140079498291, + "num_tokens": 416204876.0, + "step": 16319 + }, + { + "epoch": 1.792224906654953, + "grad_norm": 2.2109103202819824, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7243086099624634, + "num_tokens": 416232710.0, + "step": 16320 + }, + { + "epoch": 1.7923347243575665, + "grad_norm": 2.0483415126800537, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7119325399398804, + "num_tokens": 416260815.0, + "step": 16321 + }, + { + "epoch": 1.79244454206018, + "grad_norm": 2.2588961124420166, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6987246870994568, + "num_tokens": 416288370.0, + "step": 16322 + }, + { + "epoch": 1.7925543597627938, + "grad_norm": 2.397006034851074, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7345408201217651, + "num_tokens": 416311752.0, + "step": 16323 + }, + { + "epoch": 1.7926641774654075, + "grad_norm": 2.102890968322754, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7388485670089722, + "num_tokens": 416337808.0, + "step": 16324 + }, + { + "epoch": 1.792773995168021, + "grad_norm": 2.230595350265503, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7252969741821289, + "num_tokens": 416364488.0, + "step": 16325 + }, + { + "epoch": 1.7928838128706346, + "grad_norm": 2.401334285736084, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.694652795791626, + "num_tokens": 416391392.0, + "step": 16326 + }, + { + "epoch": 1.7929936305732483, + "grad_norm": 2.2340190410614014, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7165713310241699, + "num_tokens": 416417633.0, + "step": 16327 + }, + { + "epoch": 1.793103448275862, + "grad_norm": 2.0591235160827637, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7327541708946228, + "num_tokens": 416445662.0, + "step": 16328 + }, + { + "epoch": 1.7932132659784759, + "grad_norm": 2.645936965942383, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7416019439697266, + "num_tokens": 416464405.0, + "step": 16329 + }, + { + "epoch": 1.7933230836810894, + "grad_norm": 2.2793309688568115, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7136319875717163, + "num_tokens": 416488938.0, + "step": 16330 + }, + { + "epoch": 1.793432901383703, + "grad_norm": 1.8778411149978638, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7009758949279785, + "num_tokens": 416524826.0, + "step": 16331 + }, + { + "epoch": 1.7935427190863167, + "grad_norm": 2.623598337173462, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7246009111404419, + "num_tokens": 416547376.0, + "step": 16332 + }, + { + "epoch": 1.7936525367889304, + "grad_norm": 2.396449565887451, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7317421436309814, + "num_tokens": 416570573.0, + "step": 16333 + }, + { + "epoch": 1.7937623544915442, + "grad_norm": 2.2904789447784424, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7246106863021851, + "num_tokens": 416596622.0, + "step": 16334 + }, + { + "epoch": 1.7938721721941577, + "grad_norm": 2.3686978816986084, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7277726531028748, + "num_tokens": 416619791.0, + "step": 16335 + }, + { + "epoch": 1.7939819898967713, + "grad_norm": 2.2805802822113037, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7498928904533386, + "num_tokens": 416643036.0, + "step": 16336 + }, + { + "epoch": 1.794091807599385, + "grad_norm": 2.385615587234497, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7324185967445374, + "num_tokens": 416665154.0, + "step": 16337 + }, + { + "epoch": 1.7942016253019988, + "grad_norm": 2.0668137073516846, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.722683310508728, + "num_tokens": 416692767.0, + "step": 16338 + }, + { + "epoch": 1.7943114430046123, + "grad_norm": 2.009059190750122, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6980258822441101, + "num_tokens": 416725203.0, + "step": 16339 + }, + { + "epoch": 1.7944212607072259, + "grad_norm": 2.3525891304016113, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7328039407730103, + "num_tokens": 416748483.0, + "step": 16340 + }, + { + "epoch": 1.7945310784098396, + "grad_norm": 2.5952768325805664, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7184336185455322, + "num_tokens": 416768783.0, + "step": 16341 + }, + { + "epoch": 1.7946408961124534, + "grad_norm": 2.2364344596862793, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7183030247688293, + "num_tokens": 416795739.0, + "step": 16342 + }, + { + "epoch": 1.7947507138150671, + "grad_norm": 2.09194016456604, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.725077211856842, + "num_tokens": 416825137.0, + "step": 16343 + }, + { + "epoch": 1.7948605315176807, + "grad_norm": 2.0983378887176514, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7258059978485107, + "num_tokens": 416854517.0, + "step": 16344 + }, + { + "epoch": 1.7949703492202942, + "grad_norm": 2.081202983856201, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7154878377914429, + "num_tokens": 416886887.0, + "step": 16345 + }, + { + "epoch": 1.795080166922908, + "grad_norm": 3.00644588470459, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7341204285621643, + "num_tokens": 416902300.0, + "step": 16346 + }, + { + "epoch": 1.7951899846255217, + "grad_norm": 2.253685712814331, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7349344491958618, + "num_tokens": 416926956.0, + "step": 16347 + }, + { + "epoch": 1.7952998023281352, + "grad_norm": 2.1600325107574463, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6935781836509705, + "num_tokens": 416953316.0, + "step": 16348 + }, + { + "epoch": 1.795409620030749, + "grad_norm": 2.3041765689849854, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7245281934738159, + "num_tokens": 416978108.0, + "step": 16349 + }, + { + "epoch": 1.7955194377333625, + "grad_norm": 2.464425802230835, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7172498106956482, + "num_tokens": 417001748.0, + "step": 16350 + }, + { + "epoch": 1.7956292554359763, + "grad_norm": 2.3399710655212402, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.696295976638794, + "num_tokens": 417027766.0, + "step": 16351 + }, + { + "epoch": 1.79573907313859, + "grad_norm": 2.2061235904693604, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.7557015419006348, + "num_tokens": 417052422.0, + "step": 16352 + }, + { + "epoch": 1.7958488908412036, + "grad_norm": 2.452956199645996, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7369561791419983, + "num_tokens": 417077506.0, + "step": 16353 + }, + { + "epoch": 1.7959587085438171, + "grad_norm": 2.494950294494629, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7168313264846802, + "num_tokens": 417100840.0, + "step": 16354 + }, + { + "epoch": 1.7960685262464309, + "grad_norm": 2.0717287063598633, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7373350262641907, + "num_tokens": 417130163.0, + "step": 16355 + }, + { + "epoch": 1.7961783439490446, + "grad_norm": 1.9978541135787964, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7144032120704651, + "num_tokens": 417162186.0, + "step": 16356 + }, + { + "epoch": 1.7962881616516584, + "grad_norm": 2.6142337322235107, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7181745767593384, + "num_tokens": 417183399.0, + "step": 16357 + }, + { + "epoch": 1.796397979354272, + "grad_norm": 1.8681225776672363, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7157034277915955, + "num_tokens": 417218099.0, + "step": 16358 + }, + { + "epoch": 1.7965077970568855, + "grad_norm": 2.10720157623291, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6976964473724365, + "num_tokens": 417247533.0, + "step": 16359 + }, + { + "epoch": 1.7966176147594992, + "grad_norm": 2.193699836730957, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7103896141052246, + "num_tokens": 417272865.0, + "step": 16360 + }, + { + "epoch": 1.796727432462113, + "grad_norm": 2.283738851547241, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.7480093836784363, + "num_tokens": 417296536.0, + "step": 16361 + }, + { + "epoch": 1.7968372501647265, + "grad_norm": 2.385590076446533, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7122012376785278, + "num_tokens": 417319053.0, + "step": 16362 + }, + { + "epoch": 1.7969470678673403, + "grad_norm": 2.4680206775665283, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7201632261276245, + "num_tokens": 417342366.0, + "step": 16363 + }, + { + "epoch": 1.7970568855699538, + "grad_norm": 2.2402846813201904, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7342759370803833, + "num_tokens": 417367588.0, + "step": 16364 + }, + { + "epoch": 1.7971667032725676, + "grad_norm": 1.9457437992095947, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7079108357429504, + "num_tokens": 417399811.0, + "step": 16365 + }, + { + "epoch": 1.7972765209751813, + "grad_norm": 2.3098104000091553, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7336782813072205, + "num_tokens": 417424438.0, + "step": 16366 + }, + { + "epoch": 1.7973863386777948, + "grad_norm": 2.203172445297241, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7205048203468323, + "num_tokens": 417449440.0, + "step": 16367 + }, + { + "epoch": 1.7974961563804084, + "grad_norm": 2.227161169052124, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7080157995223999, + "num_tokens": 417480310.0, + "step": 16368 + }, + { + "epoch": 1.7976059740830221, + "grad_norm": 2.2019033432006836, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7305033802986145, + "num_tokens": 417507041.0, + "step": 16369 + }, + { + "epoch": 1.797715791785636, + "grad_norm": 2.335935354232788, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7315244078636169, + "num_tokens": 417530377.0, + "step": 16370 + }, + { + "epoch": 1.7978256094882497, + "grad_norm": 2.0658681392669678, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7270077466964722, + "num_tokens": 417558218.0, + "step": 16371 + }, + { + "epoch": 1.7979354271908632, + "grad_norm": 2.5148348808288574, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7161665558815002, + "num_tokens": 417578997.0, + "step": 16372 + }, + { + "epoch": 1.7980452448934767, + "grad_norm": 1.8870680332183838, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7364834547042847, + "num_tokens": 417611671.0, + "step": 16373 + }, + { + "epoch": 1.7981550625960905, + "grad_norm": 2.6540677547454834, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7178473472595215, + "num_tokens": 417632067.0, + "step": 16374 + }, + { + "epoch": 1.7982648802987042, + "grad_norm": 2.202284097671509, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7527707815170288, + "num_tokens": 417656283.0, + "step": 16375 + }, + { + "epoch": 1.7983746980013178, + "grad_norm": 2.1718502044677734, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.739589512348175, + "num_tokens": 417682480.0, + "step": 16376 + }, + { + "epoch": 1.7984845157039313, + "grad_norm": 2.314175605773926, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7246197462081909, + "num_tokens": 417705983.0, + "step": 16377 + }, + { + "epoch": 1.798594333406545, + "grad_norm": 2.403243064880371, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7261520624160767, + "num_tokens": 417728971.0, + "step": 16378 + }, + { + "epoch": 1.7987041511091588, + "grad_norm": 2.029214859008789, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.6972811222076416, + "num_tokens": 417759515.0, + "step": 16379 + }, + { + "epoch": 1.7988139688117726, + "grad_norm": 2.119198799133301, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7102324962615967, + "num_tokens": 417787330.0, + "step": 16380 + }, + { + "epoch": 1.7989237865143861, + "grad_norm": 2.002434492111206, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.724892258644104, + "num_tokens": 417820166.0, + "step": 16381 + }, + { + "epoch": 1.7990336042169996, + "grad_norm": 2.1106793880462646, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.731694221496582, + "num_tokens": 417847511.0, + "step": 16382 + }, + { + "epoch": 1.7991434219196134, + "grad_norm": 2.088925361633301, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7107099294662476, + "num_tokens": 417878573.0, + "step": 16383 + }, + { + "epoch": 1.7992532396222272, + "grad_norm": 2.318957567214966, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7475997805595398, + "num_tokens": 417901852.0, + "step": 16384 + }, + { + "epoch": 1.799363057324841, + "grad_norm": 2.332833766937256, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7030908465385437, + "num_tokens": 417925675.0, + "step": 16385 + }, + { + "epoch": 1.7994728750274545, + "grad_norm": 1.971445918083191, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7073162794113159, + "num_tokens": 417957832.0, + "step": 16386 + }, + { + "epoch": 1.799582692730068, + "grad_norm": 2.5019588470458984, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.6980481743812561, + "num_tokens": 417981698.0, + "step": 16387 + }, + { + "epoch": 1.7996925104326817, + "grad_norm": 2.544409990310669, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7451592683792114, + "num_tokens": 418001252.0, + "step": 16388 + }, + { + "epoch": 1.7998023281352955, + "grad_norm": 2.4570131301879883, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7303038835525513, + "num_tokens": 418021705.0, + "step": 16389 + }, + { + "epoch": 1.799912145837909, + "grad_norm": 2.175588607788086, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7074785232543945, + "num_tokens": 418047908.0, + "step": 16390 + }, + { + "epoch": 1.8000219635405226, + "grad_norm": 2.114267587661743, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7443568110466003, + "num_tokens": 418073440.0, + "step": 16391 + }, + { + "epoch": 1.8001317812431363, + "grad_norm": 2.2423954010009766, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7138683199882507, + "num_tokens": 418099480.0, + "step": 16392 + }, + { + "epoch": 1.80024159894575, + "grad_norm": 2.30045223236084, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.6948893070220947, + "num_tokens": 418122905.0, + "step": 16393 + }, + { + "epoch": 1.8003514166483638, + "grad_norm": 1.9377390146255493, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.6990825533866882, + "num_tokens": 418154637.0, + "step": 16394 + }, + { + "epoch": 1.8004612343509774, + "grad_norm": 2.0388236045837402, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7194892168045044, + "num_tokens": 418184238.0, + "step": 16395 + }, + { + "epoch": 1.800571052053591, + "grad_norm": 2.4019241333007812, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.718887984752655, + "num_tokens": 418209325.0, + "step": 16396 + }, + { + "epoch": 1.8006808697562047, + "grad_norm": 2.2232861518859863, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7202398777008057, + "num_tokens": 418236132.0, + "step": 16397 + }, + { + "epoch": 1.8007906874588184, + "grad_norm": 2.388671636581421, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.730840802192688, + "num_tokens": 418259546.0, + "step": 16398 + }, + { + "epoch": 1.8009005051614322, + "grad_norm": 2.2572147846221924, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7400486469268799, + "num_tokens": 418283332.0, + "step": 16399 + }, + { + "epoch": 1.8010103228640457, + "grad_norm": 2.125920295715332, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7093012928962708, + "num_tokens": 418310021.0, + "step": 16400 + }, + { + "epoch": 1.8011201405666593, + "grad_norm": 2.584761619567871, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.746540904045105, + "num_tokens": 418329636.0, + "step": 16401 + }, + { + "epoch": 1.801229958269273, + "grad_norm": 2.249871253967285, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7239623069763184, + "num_tokens": 418356820.0, + "step": 16402 + }, + { + "epoch": 1.8013397759718868, + "grad_norm": 2.249379873275757, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7357924580574036, + "num_tokens": 418381167.0, + "step": 16403 + }, + { + "epoch": 1.8014495936745003, + "grad_norm": 2.5413899421691895, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7250808477401733, + "num_tokens": 418401288.0, + "step": 16404 + }, + { + "epoch": 1.8015594113771138, + "grad_norm": 2.39627742767334, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7202724814414978, + "num_tokens": 418424373.0, + "step": 16405 + }, + { + "epoch": 1.8016692290797276, + "grad_norm": 2.0498499870300293, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.72423255443573, + "num_tokens": 418453374.0, + "step": 16406 + }, + { + "epoch": 1.8017790467823414, + "grad_norm": 2.2668726444244385, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7080365419387817, + "num_tokens": 418481639.0, + "step": 16407 + }, + { + "epoch": 1.801888864484955, + "grad_norm": 2.378580093383789, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7164402008056641, + "num_tokens": 418506259.0, + "step": 16408 + }, + { + "epoch": 1.8019986821875686, + "grad_norm": 2.227484941482544, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7027339339256287, + "num_tokens": 418535493.0, + "step": 16409 + }, + { + "epoch": 1.8021084998901822, + "grad_norm": 2.087782859802246, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7287704348564148, + "num_tokens": 418564027.0, + "step": 16410 + }, + { + "epoch": 1.802218317592796, + "grad_norm": 2.422891616821289, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7343335151672363, + "num_tokens": 418589081.0, + "step": 16411 + }, + { + "epoch": 1.8023281352954097, + "grad_norm": 2.0968234539031982, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7308964729309082, + "num_tokens": 418617698.0, + "step": 16412 + }, + { + "epoch": 1.8024379529980232, + "grad_norm": 2.1168227195739746, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7263718843460083, + "num_tokens": 418646240.0, + "step": 16413 + }, + { + "epoch": 1.802547770700637, + "grad_norm": 2.330273151397705, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.6974360942840576, + "num_tokens": 418671194.0, + "step": 16414 + }, + { + "epoch": 1.8026575884032505, + "grad_norm": 1.9412734508514404, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.6861072182655334, + "num_tokens": 418708599.0, + "step": 16415 + }, + { + "epoch": 1.8027674061058643, + "grad_norm": 2.041347026824951, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7208309769630432, + "num_tokens": 418737622.0, + "step": 16416 + }, + { + "epoch": 1.802877223808478, + "grad_norm": 1.9169813394546509, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7138904929161072, + "num_tokens": 418771747.0, + "step": 16417 + }, + { + "epoch": 1.8029870415110916, + "grad_norm": 2.2569706439971924, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7069337368011475, + "num_tokens": 418796638.0, + "step": 16418 + }, + { + "epoch": 1.803096859213705, + "grad_norm": 2.098520278930664, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7342262864112854, + "num_tokens": 418822722.0, + "step": 16419 + }, + { + "epoch": 1.8032066769163189, + "grad_norm": 2.292672872543335, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.6983098387718201, + "num_tokens": 418847153.0, + "step": 16420 + }, + { + "epoch": 1.8033164946189326, + "grad_norm": 2.525265693664551, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7038062810897827, + "num_tokens": 418869123.0, + "step": 16421 + }, + { + "epoch": 1.8034263123215464, + "grad_norm": 2.2640109062194824, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7142983675003052, + "num_tokens": 418893302.0, + "step": 16422 + }, + { + "epoch": 1.80353613002416, + "grad_norm": 2.317500591278076, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7176945805549622, + "num_tokens": 418916535.0, + "step": 16423 + }, + { + "epoch": 1.8036459477267734, + "grad_norm": 2.0634875297546387, + "learning_rate": 1e-06, + "loss": 1.0497, + "mean_token_accuracy": 0.682518482208252, + "num_tokens": 418946623.0, + "step": 16424 + }, + { + "epoch": 1.8037557654293872, + "grad_norm": 2.244947671890259, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7000539302825928, + "num_tokens": 418974723.0, + "step": 16425 + }, + { + "epoch": 1.803865583132001, + "grad_norm": 2.4075632095336914, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7123188972473145, + "num_tokens": 418997767.0, + "step": 16426 + }, + { + "epoch": 1.8039754008346145, + "grad_norm": 2.1576087474823, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7306252121925354, + "num_tokens": 419023757.0, + "step": 16427 + }, + { + "epoch": 1.8040852185372283, + "grad_norm": 1.9967163801193237, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7425774335861206, + "num_tokens": 419051045.0, + "step": 16428 + }, + { + "epoch": 1.8041950362398418, + "grad_norm": 2.5875086784362793, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7667446136474609, + "num_tokens": 419069868.0, + "step": 16429 + }, + { + "epoch": 1.8043048539424555, + "grad_norm": 2.2131454944610596, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7197766304016113, + "num_tokens": 419096224.0, + "step": 16430 + }, + { + "epoch": 1.8044146716450693, + "grad_norm": 1.9239234924316406, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7086765170097351, + "num_tokens": 419130579.0, + "step": 16431 + }, + { + "epoch": 1.8045244893476828, + "grad_norm": 2.3943300247192383, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7294955253601074, + "num_tokens": 419152892.0, + "step": 16432 + }, + { + "epoch": 1.8046343070502964, + "grad_norm": 2.012995481491089, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7338693737983704, + "num_tokens": 419182250.0, + "step": 16433 + }, + { + "epoch": 1.8047441247529101, + "grad_norm": 2.176560163497925, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7216625809669495, + "num_tokens": 419209042.0, + "step": 16434 + }, + { + "epoch": 1.8048539424555239, + "grad_norm": 2.066002130508423, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7132032513618469, + "num_tokens": 419239424.0, + "step": 16435 + }, + { + "epoch": 1.8049637601581376, + "grad_norm": 2.692453622817993, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7486284971237183, + "num_tokens": 419257979.0, + "step": 16436 + }, + { + "epoch": 1.8050735778607512, + "grad_norm": 2.2465436458587646, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7205067873001099, + "num_tokens": 419282103.0, + "step": 16437 + }, + { + "epoch": 1.8051833955633647, + "grad_norm": 2.0210113525390625, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.6992051601409912, + "num_tokens": 419313970.0, + "step": 16438 + }, + { + "epoch": 1.8052932132659785, + "grad_norm": 2.31876802444458, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7278978824615479, + "num_tokens": 419337751.0, + "step": 16439 + }, + { + "epoch": 1.8054030309685922, + "grad_norm": 2.229127883911133, + "learning_rate": 1e-06, + "loss": 1.0484, + "mean_token_accuracy": 0.6815271973609924, + "num_tokens": 419366231.0, + "step": 16440 + }, + { + "epoch": 1.8055128486712058, + "grad_norm": 2.336331367492676, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.6968141794204712, + "num_tokens": 419391067.0, + "step": 16441 + }, + { + "epoch": 1.8056226663738193, + "grad_norm": 2.331139087677002, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7258652448654175, + "num_tokens": 419416174.0, + "step": 16442 + }, + { + "epoch": 1.805732484076433, + "grad_norm": 2.0594186782836914, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7042137384414673, + "num_tokens": 419448632.0, + "step": 16443 + }, + { + "epoch": 1.8058423017790468, + "grad_norm": 2.24029541015625, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7148049473762512, + "num_tokens": 419476021.0, + "step": 16444 + }, + { + "epoch": 1.8059521194816606, + "grad_norm": 2.615316390991211, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7308285236358643, + "num_tokens": 419496871.0, + "step": 16445 + }, + { + "epoch": 1.806061937184274, + "grad_norm": 2.231649875640869, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7275124192237854, + "num_tokens": 419522724.0, + "step": 16446 + }, + { + "epoch": 1.8061717548868876, + "grad_norm": 2.1991608142852783, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7272525429725647, + "num_tokens": 419549419.0, + "step": 16447 + }, + { + "epoch": 1.8062815725895014, + "grad_norm": 2.379530429840088, + "learning_rate": 1e-06, + "loss": 0.8044, + "mean_token_accuracy": 0.7460193634033203, + "num_tokens": 419572515.0, + "step": 16448 + }, + { + "epoch": 1.8063913902921152, + "grad_norm": 2.0146725177764893, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7073363065719604, + "num_tokens": 419602476.0, + "step": 16449 + }, + { + "epoch": 1.806501207994729, + "grad_norm": 2.3793036937713623, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7405163049697876, + "num_tokens": 419626088.0, + "step": 16450 + }, + { + "epoch": 1.8066110256973424, + "grad_norm": 2.4318809509277344, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.71502286195755, + "num_tokens": 419649617.0, + "step": 16451 + }, + { + "epoch": 1.806720843399956, + "grad_norm": 2.305774450302124, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7210166454315186, + "num_tokens": 419674319.0, + "step": 16452 + }, + { + "epoch": 1.8068306611025697, + "grad_norm": 2.016874074935913, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7390145659446716, + "num_tokens": 419702635.0, + "step": 16453 + }, + { + "epoch": 1.8069404788051835, + "grad_norm": 2.205810308456421, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7275837063789368, + "num_tokens": 419727706.0, + "step": 16454 + }, + { + "epoch": 1.807050296507797, + "grad_norm": 2.5316708087921143, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7282418012619019, + "num_tokens": 419748908.0, + "step": 16455 + }, + { + "epoch": 1.8071601142104106, + "grad_norm": 2.073063611984253, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7137759327888489, + "num_tokens": 419778446.0, + "step": 16456 + }, + { + "epoch": 1.8072699319130243, + "grad_norm": 2.2290146350860596, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7140766382217407, + "num_tokens": 419803587.0, + "step": 16457 + }, + { + "epoch": 1.807379749615638, + "grad_norm": 2.1054296493530273, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7038476467132568, + "num_tokens": 419832334.0, + "step": 16458 + }, + { + "epoch": 1.8074895673182518, + "grad_norm": 1.9557671546936035, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7413369417190552, + "num_tokens": 419862237.0, + "step": 16459 + }, + { + "epoch": 1.8075993850208654, + "grad_norm": 2.5718393325805664, + "learning_rate": 1e-06, + "loss": 0.7715, + "mean_token_accuracy": 0.752329409122467, + "num_tokens": 419881509.0, + "step": 16460 + }, + { + "epoch": 1.807709202723479, + "grad_norm": 2.254445791244507, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7205367088317871, + "num_tokens": 419908639.0, + "step": 16461 + }, + { + "epoch": 1.8078190204260927, + "grad_norm": 2.381462574005127, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7424346208572388, + "num_tokens": 419930774.0, + "step": 16462 + }, + { + "epoch": 1.8079288381287064, + "grad_norm": 1.906084418296814, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7265533208847046, + "num_tokens": 419964360.0, + "step": 16463 + }, + { + "epoch": 1.8080386558313202, + "grad_norm": 2.3056516647338867, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7154008150100708, + "num_tokens": 419987172.0, + "step": 16464 + }, + { + "epoch": 1.8081484735339337, + "grad_norm": 2.1308388710021973, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.723604679107666, + "num_tokens": 420013411.0, + "step": 16465 + }, + { + "epoch": 1.8082582912365472, + "grad_norm": 2.168893814086914, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7065647840499878, + "num_tokens": 420040674.0, + "step": 16466 + }, + { + "epoch": 1.808368108939161, + "grad_norm": 1.8798986673355103, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.716759979724884, + "num_tokens": 420075282.0, + "step": 16467 + }, + { + "epoch": 1.8084779266417748, + "grad_norm": 2.060487747192383, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7027418613433838, + "num_tokens": 420106860.0, + "step": 16468 + }, + { + "epoch": 1.8085877443443883, + "grad_norm": 2.1051530838012695, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7223395109176636, + "num_tokens": 420136496.0, + "step": 16469 + }, + { + "epoch": 1.8086975620470018, + "grad_norm": 2.372309446334839, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7235826253890991, + "num_tokens": 420161488.0, + "step": 16470 + }, + { + "epoch": 1.8088073797496156, + "grad_norm": 1.9172747135162354, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7100725173950195, + "num_tokens": 420194479.0, + "step": 16471 + }, + { + "epoch": 1.8089171974522293, + "grad_norm": 2.3210883140563965, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7012870907783508, + "num_tokens": 420218921.0, + "step": 16472 + }, + { + "epoch": 1.809027015154843, + "grad_norm": 2.374863386154175, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7225358486175537, + "num_tokens": 420243676.0, + "step": 16473 + }, + { + "epoch": 1.8091368328574566, + "grad_norm": 2.400547742843628, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.697292685508728, + "num_tokens": 420268189.0, + "step": 16474 + }, + { + "epoch": 1.8092466505600702, + "grad_norm": 2.2402026653289795, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7196445465087891, + "num_tokens": 420292533.0, + "step": 16475 + }, + { + "epoch": 1.809356468262684, + "grad_norm": 2.253523349761963, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7050365209579468, + "num_tokens": 420316880.0, + "step": 16476 + }, + { + "epoch": 1.8094662859652977, + "grad_norm": 1.908613681793213, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.6893361806869507, + "num_tokens": 420350838.0, + "step": 16477 + }, + { + "epoch": 1.8095761036679112, + "grad_norm": 2.4971044063568115, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7210680246353149, + "num_tokens": 420372730.0, + "step": 16478 + }, + { + "epoch": 1.809685921370525, + "grad_norm": 2.2248101234436035, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7231897115707397, + "num_tokens": 420396738.0, + "step": 16479 + }, + { + "epoch": 1.8097957390731385, + "grad_norm": 2.472294807434082, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7392044067382812, + "num_tokens": 420419301.0, + "step": 16480 + }, + { + "epoch": 1.8099055567757523, + "grad_norm": 2.464632511138916, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7204588651657104, + "num_tokens": 420442141.0, + "step": 16481 + }, + { + "epoch": 1.810015374478366, + "grad_norm": 2.597818613052368, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7233564257621765, + "num_tokens": 420462615.0, + "step": 16482 + }, + { + "epoch": 1.8101251921809796, + "grad_norm": 2.0785434246063232, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7301982641220093, + "num_tokens": 420492145.0, + "step": 16483 + }, + { + "epoch": 1.810235009883593, + "grad_norm": 2.4319939613342285, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7310062646865845, + "num_tokens": 420515098.0, + "step": 16484 + }, + { + "epoch": 1.8103448275862069, + "grad_norm": 2.259829521179199, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7188153266906738, + "num_tokens": 420540507.0, + "step": 16485 + }, + { + "epoch": 1.8104546452888206, + "grad_norm": 2.4214258193969727, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7327051162719727, + "num_tokens": 420563523.0, + "step": 16486 + }, + { + "epoch": 1.8105644629914344, + "grad_norm": 2.3041906356811523, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7103237509727478, + "num_tokens": 420588147.0, + "step": 16487 + }, + { + "epoch": 1.810674280694048, + "grad_norm": 2.5626649856567383, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7225911617279053, + "num_tokens": 420608523.0, + "step": 16488 + }, + { + "epoch": 1.8107840983966614, + "grad_norm": 1.8942502737045288, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7173671126365662, + "num_tokens": 420643389.0, + "step": 16489 + }, + { + "epoch": 1.8108939160992752, + "grad_norm": 2.113095760345459, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7091223001480103, + "num_tokens": 420671631.0, + "step": 16490 + }, + { + "epoch": 1.811003733801889, + "grad_norm": 2.028099775314331, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7157700657844543, + "num_tokens": 420700037.0, + "step": 16491 + }, + { + "epoch": 1.8111135515045025, + "grad_norm": 2.2803478240966797, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7145373225212097, + "num_tokens": 420724493.0, + "step": 16492 + }, + { + "epoch": 1.8112233692071162, + "grad_norm": 2.204008102416992, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7247350811958313, + "num_tokens": 420750885.0, + "step": 16493 + }, + { + "epoch": 1.8113331869097298, + "grad_norm": 2.6975629329681396, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7366563081741333, + "num_tokens": 420769635.0, + "step": 16494 + }, + { + "epoch": 1.8114430046123435, + "grad_norm": 2.420178174972534, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7107703685760498, + "num_tokens": 420792887.0, + "step": 16495 + }, + { + "epoch": 1.8115528223149573, + "grad_norm": 2.032747983932495, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7337585687637329, + "num_tokens": 420823735.0, + "step": 16496 + }, + { + "epoch": 1.8116626400175708, + "grad_norm": 2.5345349311828613, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7215268611907959, + "num_tokens": 420843911.0, + "step": 16497 + }, + { + "epoch": 1.8117724577201844, + "grad_norm": 2.413917064666748, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.735430121421814, + "num_tokens": 420865086.0, + "step": 16498 + }, + { + "epoch": 1.8118822754227981, + "grad_norm": 2.014474630355835, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7220805883407593, + "num_tokens": 420895910.0, + "step": 16499 + }, + { + "epoch": 1.8119920931254119, + "grad_norm": 2.243185520172119, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7290515899658203, + "num_tokens": 420920732.0, + "step": 16500 + }, + { + "epoch": 1.8121019108280256, + "grad_norm": 2.0970335006713867, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7175356149673462, + "num_tokens": 420948251.0, + "step": 16501 + }, + { + "epoch": 1.8122117285306392, + "grad_norm": 2.4066202640533447, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7179074287414551, + "num_tokens": 420972643.0, + "step": 16502 + }, + { + "epoch": 1.8123215462332527, + "grad_norm": 1.8875337839126587, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.707602858543396, + "num_tokens": 421007450.0, + "step": 16503 + }, + { + "epoch": 1.8124313639358665, + "grad_norm": 2.4285855293273926, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7240875959396362, + "num_tokens": 421031026.0, + "step": 16504 + }, + { + "epoch": 1.8125411816384802, + "grad_norm": 2.1606690883636475, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.6998988389968872, + "num_tokens": 421061227.0, + "step": 16505 + }, + { + "epoch": 1.8126509993410937, + "grad_norm": 2.4484047889709473, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7078087329864502, + "num_tokens": 421086240.0, + "step": 16506 + }, + { + "epoch": 1.8127608170437073, + "grad_norm": 2.320436716079712, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7047336101531982, + "num_tokens": 421111290.0, + "step": 16507 + }, + { + "epoch": 1.812870634746321, + "grad_norm": 2.667833089828491, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7206571102142334, + "num_tokens": 421131456.0, + "step": 16508 + }, + { + "epoch": 1.8129804524489348, + "grad_norm": 2.174098491668701, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7144870758056641, + "num_tokens": 421158500.0, + "step": 16509 + }, + { + "epoch": 1.8130902701515486, + "grad_norm": 1.9478262662887573, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7206266522407532, + "num_tokens": 421193071.0, + "step": 16510 + }, + { + "epoch": 1.813200087854162, + "grad_norm": 2.4767229557037354, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7185018658638, + "num_tokens": 421215484.0, + "step": 16511 + }, + { + "epoch": 1.8133099055567756, + "grad_norm": 2.345444679260254, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7770962715148926, + "num_tokens": 421236084.0, + "step": 16512 + }, + { + "epoch": 1.8134197232593894, + "grad_norm": 2.359715700149536, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7151623368263245, + "num_tokens": 421260394.0, + "step": 16513 + }, + { + "epoch": 1.8135295409620031, + "grad_norm": 2.2627041339874268, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7142341136932373, + "num_tokens": 421286308.0, + "step": 16514 + }, + { + "epoch": 1.813639358664617, + "grad_norm": 2.0337235927581787, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7330586910247803, + "num_tokens": 421317507.0, + "step": 16515 + }, + { + "epoch": 1.8137491763672304, + "grad_norm": 2.28456449508667, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7219700217247009, + "num_tokens": 421341502.0, + "step": 16516 + }, + { + "epoch": 1.813858994069844, + "grad_norm": 2.67498517036438, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7236742377281189, + "num_tokens": 421361612.0, + "step": 16517 + }, + { + "epoch": 1.8139688117724577, + "grad_norm": 2.150329351425171, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7292556762695312, + "num_tokens": 421387584.0, + "step": 16518 + }, + { + "epoch": 1.8140786294750715, + "grad_norm": 2.1953837871551514, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7103185057640076, + "num_tokens": 421414140.0, + "step": 16519 + }, + { + "epoch": 1.814188447177685, + "grad_norm": 2.3588812351226807, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7289966344833374, + "num_tokens": 421437327.0, + "step": 16520 + }, + { + "epoch": 1.8142982648802986, + "grad_norm": 2.209709405899048, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7426235675811768, + "num_tokens": 421460888.0, + "step": 16521 + }, + { + "epoch": 1.8144080825829123, + "grad_norm": 2.128042697906494, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7147840261459351, + "num_tokens": 421489143.0, + "step": 16522 + }, + { + "epoch": 1.814517900285526, + "grad_norm": 2.4052963256835938, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7413526773452759, + "num_tokens": 421510587.0, + "step": 16523 + }, + { + "epoch": 1.8146277179881398, + "grad_norm": 2.6024062633514404, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7224818468093872, + "num_tokens": 421531810.0, + "step": 16524 + }, + { + "epoch": 1.8147375356907534, + "grad_norm": 2.5098977088928223, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7298036217689514, + "num_tokens": 421551685.0, + "step": 16525 + }, + { + "epoch": 1.814847353393367, + "grad_norm": 2.5924861431121826, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7277079820632935, + "num_tokens": 421572511.0, + "step": 16526 + }, + { + "epoch": 1.8149571710959806, + "grad_norm": 2.2340166568756104, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7016828060150146, + "num_tokens": 421600557.0, + "step": 16527 + }, + { + "epoch": 1.8150669887985944, + "grad_norm": 2.5779988765716553, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7268640995025635, + "num_tokens": 421621415.0, + "step": 16528 + }, + { + "epoch": 1.815176806501208, + "grad_norm": 2.1108474731445312, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7130851149559021, + "num_tokens": 421649739.0, + "step": 16529 + }, + { + "epoch": 1.8152866242038217, + "grad_norm": 2.2427687644958496, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.7333369255065918, + "num_tokens": 421675811.0, + "step": 16530 + }, + { + "epoch": 1.8153964419064352, + "grad_norm": 2.5586366653442383, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7102417945861816, + "num_tokens": 421697433.0, + "step": 16531 + }, + { + "epoch": 1.815506259609049, + "grad_norm": 1.9790059328079224, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7110910415649414, + "num_tokens": 421728801.0, + "step": 16532 + }, + { + "epoch": 1.8156160773116627, + "grad_norm": 2.233975648880005, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7239179611206055, + "num_tokens": 421753935.0, + "step": 16533 + }, + { + "epoch": 1.8157258950142763, + "grad_norm": 2.071331262588501, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7058334946632385, + "num_tokens": 421782506.0, + "step": 16534 + }, + { + "epoch": 1.8158357127168898, + "grad_norm": 2.167619228363037, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7205278277397156, + "num_tokens": 421810376.0, + "step": 16535 + }, + { + "epoch": 1.8159455304195036, + "grad_norm": 2.3626387119293213, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.6924842596054077, + "num_tokens": 421835702.0, + "step": 16536 + }, + { + "epoch": 1.8160553481221173, + "grad_norm": 2.1552271842956543, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7112553715705872, + "num_tokens": 421865299.0, + "step": 16537 + }, + { + "epoch": 1.816165165824731, + "grad_norm": 2.3256685733795166, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7203327417373657, + "num_tokens": 421889468.0, + "step": 16538 + }, + { + "epoch": 1.8162749835273446, + "grad_norm": 2.086963415145874, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7131813168525696, + "num_tokens": 421919582.0, + "step": 16539 + }, + { + "epoch": 1.8163848012299582, + "grad_norm": 2.416001319885254, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7295002937316895, + "num_tokens": 421941757.0, + "step": 16540 + }, + { + "epoch": 1.816494618932572, + "grad_norm": 2.005842685699463, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7162468433380127, + "num_tokens": 421975485.0, + "step": 16541 + }, + { + "epoch": 1.8166044366351857, + "grad_norm": 2.455474615097046, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7328657507896423, + "num_tokens": 421996875.0, + "step": 16542 + }, + { + "epoch": 1.8167142543377992, + "grad_norm": 2.309067726135254, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7241202592849731, + "num_tokens": 422022688.0, + "step": 16543 + }, + { + "epoch": 1.816824072040413, + "grad_norm": 1.9189738035202026, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6942227482795715, + "num_tokens": 422056976.0, + "step": 16544 + }, + { + "epoch": 1.8169338897430265, + "grad_norm": 2.0317606925964355, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.7040137648582458, + "num_tokens": 422087845.0, + "step": 16545 + }, + { + "epoch": 1.8170437074456403, + "grad_norm": 2.4322562217712402, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7455669641494751, + "num_tokens": 422110181.0, + "step": 16546 + }, + { + "epoch": 1.817153525148254, + "grad_norm": 2.0831799507141113, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7157453298568726, + "num_tokens": 422138092.0, + "step": 16547 + }, + { + "epoch": 1.8172633428508675, + "grad_norm": 2.524637222290039, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7066143751144409, + "num_tokens": 422161236.0, + "step": 16548 + }, + { + "epoch": 1.817373160553481, + "grad_norm": 2.274235963821411, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7155733108520508, + "num_tokens": 422185112.0, + "step": 16549 + }, + { + "epoch": 1.8174829782560948, + "grad_norm": 2.481677293777466, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7241648435592651, + "num_tokens": 422207452.0, + "step": 16550 + }, + { + "epoch": 1.8175927959587086, + "grad_norm": 2.062269926071167, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7091633081436157, + "num_tokens": 422237782.0, + "step": 16551 + }, + { + "epoch": 1.8177026136613224, + "grad_norm": 2.393198013305664, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7298607230186462, + "num_tokens": 422261065.0, + "step": 16552 + }, + { + "epoch": 1.8178124313639359, + "grad_norm": 2.0066895484924316, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7310230731964111, + "num_tokens": 422291522.0, + "step": 16553 + }, + { + "epoch": 1.8179222490665494, + "grad_norm": 2.1975114345550537, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7545655369758606, + "num_tokens": 422314620.0, + "step": 16554 + }, + { + "epoch": 1.8180320667691632, + "grad_norm": 2.54158616065979, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7012045383453369, + "num_tokens": 422339877.0, + "step": 16555 + }, + { + "epoch": 1.818141884471777, + "grad_norm": 2.3316726684570312, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7074331045150757, + "num_tokens": 422364893.0, + "step": 16556 + }, + { + "epoch": 1.8182517021743905, + "grad_norm": 2.4867725372314453, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7279900312423706, + "num_tokens": 422385569.0, + "step": 16557 + }, + { + "epoch": 1.818361519877004, + "grad_norm": 2.1291701793670654, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.6969195604324341, + "num_tokens": 422414700.0, + "step": 16558 + }, + { + "epoch": 1.8184713375796178, + "grad_norm": 2.2393832206726074, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7278999090194702, + "num_tokens": 422440259.0, + "step": 16559 + }, + { + "epoch": 1.8185811552822315, + "grad_norm": 2.2098662853240967, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7062686681747437, + "num_tokens": 422466531.0, + "step": 16560 + }, + { + "epoch": 1.8186909729848453, + "grad_norm": 2.2709832191467285, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7226152420043945, + "num_tokens": 422489914.0, + "step": 16561 + }, + { + "epoch": 1.8188007906874588, + "grad_norm": 2.350985288619995, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7133499383926392, + "num_tokens": 422513288.0, + "step": 16562 + }, + { + "epoch": 1.8189106083900723, + "grad_norm": 2.5592262744903564, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7093641757965088, + "num_tokens": 422536257.0, + "step": 16563 + }, + { + "epoch": 1.819020426092686, + "grad_norm": 2.084873676300049, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7152227163314819, + "num_tokens": 422565668.0, + "step": 16564 + }, + { + "epoch": 1.8191302437952999, + "grad_norm": 2.4320573806762695, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7259261608123779, + "num_tokens": 422589144.0, + "step": 16565 + }, + { + "epoch": 1.8192400614979136, + "grad_norm": 2.46712064743042, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.7521672248840332, + "num_tokens": 422610831.0, + "step": 16566 + }, + { + "epoch": 1.8193498792005272, + "grad_norm": 2.0520336627960205, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7278781533241272, + "num_tokens": 422640092.0, + "step": 16567 + }, + { + "epoch": 1.8194596969031407, + "grad_norm": 2.264775037765503, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.723764181137085, + "num_tokens": 422662707.0, + "step": 16568 + }, + { + "epoch": 1.8195695146057544, + "grad_norm": 2.2789509296417236, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7229541540145874, + "num_tokens": 422685946.0, + "step": 16569 + }, + { + "epoch": 1.8196793323083682, + "grad_norm": 2.516817569732666, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7062221765518188, + "num_tokens": 422707749.0, + "step": 16570 + }, + { + "epoch": 1.8197891500109817, + "grad_norm": 2.8689281940460205, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7235531806945801, + "num_tokens": 422725212.0, + "step": 16571 + }, + { + "epoch": 1.8198989677135953, + "grad_norm": 2.2867612838745117, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7007735967636108, + "num_tokens": 422751288.0, + "step": 16572 + }, + { + "epoch": 1.820008785416209, + "grad_norm": 2.438739538192749, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7321432828903198, + "num_tokens": 422772941.0, + "step": 16573 + }, + { + "epoch": 1.8201186031188228, + "grad_norm": 2.2367002964019775, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7084548473358154, + "num_tokens": 422799177.0, + "step": 16574 + }, + { + "epoch": 1.8202284208214365, + "grad_norm": 2.171194314956665, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7268961668014526, + "num_tokens": 422824629.0, + "step": 16575 + }, + { + "epoch": 1.82033823852405, + "grad_norm": 2.529876708984375, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7162134647369385, + "num_tokens": 422844740.0, + "step": 16576 + }, + { + "epoch": 1.8204480562266636, + "grad_norm": 2.3941566944122314, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7197392582893372, + "num_tokens": 422869790.0, + "step": 16577 + }, + { + "epoch": 1.8205578739292774, + "grad_norm": 2.4223735332489014, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7041690349578857, + "num_tokens": 422892789.0, + "step": 16578 + }, + { + "epoch": 1.8206676916318911, + "grad_norm": 2.0769786834716797, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7018081545829773, + "num_tokens": 422921592.0, + "step": 16579 + }, + { + "epoch": 1.8207775093345049, + "grad_norm": 2.2986347675323486, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7327970862388611, + "num_tokens": 422944880.0, + "step": 16580 + }, + { + "epoch": 1.8208873270371184, + "grad_norm": 2.5161259174346924, + "learning_rate": 1e-06, + "loss": 0.7773, + "mean_token_accuracy": 0.74901282787323, + "num_tokens": 422964690.0, + "step": 16581 + }, + { + "epoch": 1.820997144739732, + "grad_norm": 2.7755491733551025, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7287453413009644, + "num_tokens": 422985193.0, + "step": 16582 + }, + { + "epoch": 1.8211069624423457, + "grad_norm": 2.145627498626709, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7336252331733704, + "num_tokens": 423012963.0, + "step": 16583 + }, + { + "epoch": 1.8212167801449595, + "grad_norm": 2.0646467208862305, + "learning_rate": 1e-06, + "loss": 1.0599, + "mean_token_accuracy": 0.6846669316291809, + "num_tokens": 423042735.0, + "step": 16584 + }, + { + "epoch": 1.821326597847573, + "grad_norm": 2.4552078247070312, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.7521162033081055, + "num_tokens": 423063170.0, + "step": 16585 + }, + { + "epoch": 1.8214364155501865, + "grad_norm": 2.126862049102783, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7186620235443115, + "num_tokens": 423091821.0, + "step": 16586 + }, + { + "epoch": 1.8215462332528003, + "grad_norm": 2.2667722702026367, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7393157482147217, + "num_tokens": 423116448.0, + "step": 16587 + }, + { + "epoch": 1.821656050955414, + "grad_norm": 1.9343369007110596, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7071706652641296, + "num_tokens": 423150901.0, + "step": 16588 + }, + { + "epoch": 1.8217658686580278, + "grad_norm": 2.081084728240967, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7193341255187988, + "num_tokens": 423179702.0, + "step": 16589 + }, + { + "epoch": 1.8218756863606413, + "grad_norm": 2.235391855239868, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7261682748794556, + "num_tokens": 423204131.0, + "step": 16590 + }, + { + "epoch": 1.8219855040632549, + "grad_norm": 1.9900918006896973, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7084472179412842, + "num_tokens": 423233200.0, + "step": 16591 + }, + { + "epoch": 1.8220953217658686, + "grad_norm": 2.1902780532836914, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.722055196762085, + "num_tokens": 423259941.0, + "step": 16592 + }, + { + "epoch": 1.8222051394684824, + "grad_norm": 1.9285980463027954, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7192139029502869, + "num_tokens": 423292883.0, + "step": 16593 + }, + { + "epoch": 1.822314957171096, + "grad_norm": 2.0186142921447754, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7146495580673218, + "num_tokens": 423324302.0, + "step": 16594 + }, + { + "epoch": 1.8224247748737097, + "grad_norm": 2.3539528846740723, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.730092465877533, + "num_tokens": 423347569.0, + "step": 16595 + }, + { + "epoch": 1.8225345925763232, + "grad_norm": 2.2258780002593994, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7169016599655151, + "num_tokens": 423374019.0, + "step": 16596 + }, + { + "epoch": 1.822644410278937, + "grad_norm": 2.2015748023986816, + "learning_rate": 1e-06, + "loss": 1.0951, + "mean_token_accuracy": 0.6838704943656921, + "num_tokens": 423402815.0, + "step": 16597 + }, + { + "epoch": 1.8227542279815507, + "grad_norm": 2.202310562133789, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7133194208145142, + "num_tokens": 423428725.0, + "step": 16598 + }, + { + "epoch": 1.8228640456841643, + "grad_norm": 2.069003105163574, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7233772873878479, + "num_tokens": 423458252.0, + "step": 16599 + }, + { + "epoch": 1.8229738633867778, + "grad_norm": 2.1936604976654053, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7163114547729492, + "num_tokens": 423483383.0, + "step": 16600 + }, + { + "epoch": 1.8230836810893916, + "grad_norm": 2.1617655754089355, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.7527406215667725, + "num_tokens": 423511342.0, + "step": 16601 + }, + { + "epoch": 1.8231934987920053, + "grad_norm": 2.613007068634033, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7140833139419556, + "num_tokens": 423532102.0, + "step": 16602 + }, + { + "epoch": 1.823303316494619, + "grad_norm": 2.109306573867798, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7262653112411499, + "num_tokens": 423559670.0, + "step": 16603 + }, + { + "epoch": 1.8234131341972326, + "grad_norm": 2.1837613582611084, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7257976531982422, + "num_tokens": 423585166.0, + "step": 16604 + }, + { + "epoch": 1.8235229518998461, + "grad_norm": 2.129506826400757, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7061437368392944, + "num_tokens": 423612578.0, + "step": 16605 + }, + { + "epoch": 1.82363276960246, + "grad_norm": 2.2483839988708496, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7276921272277832, + "num_tokens": 423637137.0, + "step": 16606 + }, + { + "epoch": 1.8237425873050737, + "grad_norm": 2.3966305255889893, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7102238535881042, + "num_tokens": 423660186.0, + "step": 16607 + }, + { + "epoch": 1.8238524050076872, + "grad_norm": 2.113288402557373, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7063583135604858, + "num_tokens": 423686789.0, + "step": 16608 + }, + { + "epoch": 1.823962222710301, + "grad_norm": 2.4974300861358643, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7160339951515198, + "num_tokens": 423708114.0, + "step": 16609 + }, + { + "epoch": 1.8240720404129145, + "grad_norm": 2.222468614578247, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.719833493232727, + "num_tokens": 423733525.0, + "step": 16610 + }, + { + "epoch": 1.8241818581155282, + "grad_norm": 2.5841121673583984, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.6978172659873962, + "num_tokens": 423754965.0, + "step": 16611 + }, + { + "epoch": 1.824291675818142, + "grad_norm": 2.2820651531219482, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7272763252258301, + "num_tokens": 423778713.0, + "step": 16612 + }, + { + "epoch": 1.8244014935207555, + "grad_norm": 2.1761927604675293, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7265546917915344, + "num_tokens": 423806775.0, + "step": 16613 + }, + { + "epoch": 1.824511311223369, + "grad_norm": 2.0814433097839355, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7207266092300415, + "num_tokens": 423834585.0, + "step": 16614 + }, + { + "epoch": 1.8246211289259828, + "grad_norm": 2.12540864944458, + "learning_rate": 1e-06, + "loss": 0.7772, + "mean_token_accuracy": 0.7445089817047119, + "num_tokens": 423861442.0, + "step": 16615 + }, + { + "epoch": 1.8247309466285966, + "grad_norm": 2.144885540008545, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7339915633201599, + "num_tokens": 423886939.0, + "step": 16616 + }, + { + "epoch": 1.8248407643312103, + "grad_norm": 2.1950416564941406, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.71724534034729, + "num_tokens": 423914594.0, + "step": 16617 + }, + { + "epoch": 1.8249505820338239, + "grad_norm": 2.039900302886963, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.7020083665847778, + "num_tokens": 423946665.0, + "step": 16618 + }, + { + "epoch": 1.8250603997364374, + "grad_norm": 2.1136555671691895, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7177817821502686, + "num_tokens": 423974096.0, + "step": 16619 + }, + { + "epoch": 1.8251702174390512, + "grad_norm": 2.203430652618408, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7281302213668823, + "num_tokens": 424001764.0, + "step": 16620 + }, + { + "epoch": 1.825280035141665, + "grad_norm": 2.3577511310577393, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.710343062877655, + "num_tokens": 424025220.0, + "step": 16621 + }, + { + "epoch": 1.8253898528442785, + "grad_norm": 2.202695369720459, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7321749925613403, + "num_tokens": 424051025.0, + "step": 16622 + }, + { + "epoch": 1.825499670546892, + "grad_norm": 2.5109801292419434, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.718017578125, + "num_tokens": 424074373.0, + "step": 16623 + }, + { + "epoch": 1.8256094882495058, + "grad_norm": 2.480152130126953, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7599482536315918, + "num_tokens": 424093638.0, + "step": 16624 + }, + { + "epoch": 1.8257193059521195, + "grad_norm": 2.223816156387329, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7235375642776489, + "num_tokens": 424120355.0, + "step": 16625 + }, + { + "epoch": 1.8258291236547333, + "grad_norm": 2.3348705768585205, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7066152691841125, + "num_tokens": 424146015.0, + "step": 16626 + }, + { + "epoch": 1.8259389413573468, + "grad_norm": 2.141878604888916, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7107424139976501, + "num_tokens": 424173778.0, + "step": 16627 + }, + { + "epoch": 1.8260487590599603, + "grad_norm": 2.1689207553863525, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7316457629203796, + "num_tokens": 424200592.0, + "step": 16628 + }, + { + "epoch": 1.826158576762574, + "grad_norm": 2.136640787124634, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7119791507720947, + "num_tokens": 424227779.0, + "step": 16629 + }, + { + "epoch": 1.8262683944651878, + "grad_norm": 2.1983859539031982, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.716530978679657, + "num_tokens": 424252745.0, + "step": 16630 + }, + { + "epoch": 1.8263782121678016, + "grad_norm": 2.207131862640381, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.704011082649231, + "num_tokens": 424281984.0, + "step": 16631 + }, + { + "epoch": 1.8264880298704151, + "grad_norm": 2.338883876800537, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7052372097969055, + "num_tokens": 424306222.0, + "step": 16632 + }, + { + "epoch": 1.8265978475730287, + "grad_norm": 2.061659812927246, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7165082693099976, + "num_tokens": 424334133.0, + "step": 16633 + }, + { + "epoch": 1.8267076652756424, + "grad_norm": 2.044442892074585, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7240103483200073, + "num_tokens": 424364158.0, + "step": 16634 + }, + { + "epoch": 1.8268174829782562, + "grad_norm": 2.4412155151367188, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7182408571243286, + "num_tokens": 424387896.0, + "step": 16635 + }, + { + "epoch": 1.8269273006808697, + "grad_norm": 2.520012378692627, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7256377935409546, + "num_tokens": 424408780.0, + "step": 16636 + }, + { + "epoch": 1.8270371183834833, + "grad_norm": 2.298140525817871, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7038236856460571, + "num_tokens": 424433047.0, + "step": 16637 + }, + { + "epoch": 1.827146936086097, + "grad_norm": 1.9795011281967163, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7108680009841919, + "num_tokens": 424464162.0, + "step": 16638 + }, + { + "epoch": 1.8272567537887108, + "grad_norm": 2.092177152633667, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7305634617805481, + "num_tokens": 424491685.0, + "step": 16639 + }, + { + "epoch": 1.8273665714913245, + "grad_norm": 2.0160374641418457, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.705768883228302, + "num_tokens": 424523016.0, + "step": 16640 + }, + { + "epoch": 1.827476389193938, + "grad_norm": 2.603801727294922, + "learning_rate": 1e-06, + "loss": 0.7746, + "mean_token_accuracy": 0.7500604391098022, + "num_tokens": 424540466.0, + "step": 16641 + }, + { + "epoch": 1.8275862068965516, + "grad_norm": 2.2638473510742188, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7346680164337158, + "num_tokens": 424564580.0, + "step": 16642 + }, + { + "epoch": 1.8276960245991654, + "grad_norm": 2.3729798793792725, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7054015398025513, + "num_tokens": 424590197.0, + "step": 16643 + }, + { + "epoch": 1.8278058423017791, + "grad_norm": 1.9535257816314697, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6908715963363647, + "num_tokens": 424623639.0, + "step": 16644 + }, + { + "epoch": 1.8279156600043929, + "grad_norm": 2.464576482772827, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7583047151565552, + "num_tokens": 424644545.0, + "step": 16645 + }, + { + "epoch": 1.8280254777070064, + "grad_norm": 2.102457284927368, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.738940954208374, + "num_tokens": 424671683.0, + "step": 16646 + }, + { + "epoch": 1.82813529540962, + "grad_norm": 2.4313814640045166, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7107645273208618, + "num_tokens": 424693577.0, + "step": 16647 + }, + { + "epoch": 1.8282451131122337, + "grad_norm": 2.333486557006836, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7100603580474854, + "num_tokens": 424714597.0, + "step": 16648 + }, + { + "epoch": 1.8283549308148475, + "grad_norm": 2.1824774742126465, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7275454998016357, + "num_tokens": 424739846.0, + "step": 16649 + }, + { + "epoch": 1.828464748517461, + "grad_norm": 2.4391796588897705, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.6989593505859375, + "num_tokens": 424761794.0, + "step": 16650 + }, + { + "epoch": 1.8285745662200745, + "grad_norm": 2.162804126739502, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7166851758956909, + "num_tokens": 424789874.0, + "step": 16651 + }, + { + "epoch": 1.8286843839226883, + "grad_norm": 2.0531158447265625, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.718507707118988, + "num_tokens": 424818446.0, + "step": 16652 + }, + { + "epoch": 1.828794201625302, + "grad_norm": 2.2017505168914795, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7338382005691528, + "num_tokens": 424843719.0, + "step": 16653 + }, + { + "epoch": 1.8289040193279158, + "grad_norm": 2.373666524887085, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.714128315448761, + "num_tokens": 424868581.0, + "step": 16654 + }, + { + "epoch": 1.8290138370305293, + "grad_norm": 2.14024019241333, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7201519012451172, + "num_tokens": 424897132.0, + "step": 16655 + }, + { + "epoch": 1.8291236547331429, + "grad_norm": 1.9772030115127563, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7124037742614746, + "num_tokens": 424927774.0, + "step": 16656 + }, + { + "epoch": 1.8292334724357566, + "grad_norm": 2.447366237640381, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7287137508392334, + "num_tokens": 424948878.0, + "step": 16657 + }, + { + "epoch": 1.8293432901383704, + "grad_norm": 2.437260627746582, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7047233581542969, + "num_tokens": 424974637.0, + "step": 16658 + }, + { + "epoch": 1.829453107840984, + "grad_norm": 2.8574893474578857, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7211499214172363, + "num_tokens": 424992579.0, + "step": 16659 + }, + { + "epoch": 1.8295629255435977, + "grad_norm": 2.3388257026672363, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7220063209533691, + "num_tokens": 425020202.0, + "step": 16660 + }, + { + "epoch": 1.8296727432462112, + "grad_norm": 2.5181212425231934, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.730363667011261, + "num_tokens": 425042897.0, + "step": 16661 + }, + { + "epoch": 1.829782560948825, + "grad_norm": 2.055565118789673, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7172266840934753, + "num_tokens": 425071214.0, + "step": 16662 + }, + { + "epoch": 1.8298923786514387, + "grad_norm": 2.2532713413238525, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.728697657585144, + "num_tokens": 425094293.0, + "step": 16663 + }, + { + "epoch": 1.8300021963540523, + "grad_norm": 2.3196980953216553, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7091014385223389, + "num_tokens": 425118509.0, + "step": 16664 + }, + { + "epoch": 1.8301120140566658, + "grad_norm": 2.3763062953948975, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.717581033706665, + "num_tokens": 425142415.0, + "step": 16665 + }, + { + "epoch": 1.8302218317592795, + "grad_norm": 2.267164468765259, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7133245468139648, + "num_tokens": 425167544.0, + "step": 16666 + }, + { + "epoch": 1.8303316494618933, + "grad_norm": 2.0371575355529785, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.742110013961792, + "num_tokens": 425198057.0, + "step": 16667 + }, + { + "epoch": 1.830441467164507, + "grad_norm": 2.2893524169921875, + "learning_rate": 1e-06, + "loss": 0.77, + "mean_token_accuracy": 0.755983293056488, + "num_tokens": 425222024.0, + "step": 16668 + }, + { + "epoch": 1.8305512848671206, + "grad_norm": 2.2137744426727295, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7301178574562073, + "num_tokens": 425246123.0, + "step": 16669 + }, + { + "epoch": 1.8306611025697341, + "grad_norm": 2.511773109436035, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7269253730773926, + "num_tokens": 425266492.0, + "step": 16670 + }, + { + "epoch": 1.830770920272348, + "grad_norm": 1.9930505752563477, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7248963117599487, + "num_tokens": 425295250.0, + "step": 16671 + }, + { + "epoch": 1.8308807379749616, + "grad_norm": 2.1375572681427, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.719280481338501, + "num_tokens": 425325815.0, + "step": 16672 + }, + { + "epoch": 1.8309905556775752, + "grad_norm": 1.9735517501831055, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.6954197883605957, + "num_tokens": 425357286.0, + "step": 16673 + }, + { + "epoch": 1.831100373380189, + "grad_norm": 2.598264455795288, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7306568026542664, + "num_tokens": 425378735.0, + "step": 16674 + }, + { + "epoch": 1.8312101910828025, + "grad_norm": 2.407182455062866, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7483978271484375, + "num_tokens": 425401363.0, + "step": 16675 + }, + { + "epoch": 1.8313200087854162, + "grad_norm": 2.5163254737854004, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7114133238792419, + "num_tokens": 425425479.0, + "step": 16676 + }, + { + "epoch": 1.83142982648803, + "grad_norm": 2.2719907760620117, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7242003679275513, + "num_tokens": 425450986.0, + "step": 16677 + }, + { + "epoch": 1.8315396441906435, + "grad_norm": 2.5552308559417725, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7107409238815308, + "num_tokens": 425472828.0, + "step": 16678 + }, + { + "epoch": 1.831649461893257, + "grad_norm": 2.3423893451690674, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7185681462287903, + "num_tokens": 425496715.0, + "step": 16679 + }, + { + "epoch": 1.8317592795958708, + "grad_norm": 2.0434298515319824, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7331511378288269, + "num_tokens": 425526084.0, + "step": 16680 + }, + { + "epoch": 1.8318690972984846, + "grad_norm": 2.2198143005371094, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7124752402305603, + "num_tokens": 425551403.0, + "step": 16681 + }, + { + "epoch": 1.8319789150010983, + "grad_norm": 2.3653101921081543, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7407071590423584, + "num_tokens": 425575129.0, + "step": 16682 + }, + { + "epoch": 1.8320887327037119, + "grad_norm": 2.3033483028411865, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7297183871269226, + "num_tokens": 425597231.0, + "step": 16683 + }, + { + "epoch": 1.8321985504063254, + "grad_norm": 2.1800105571746826, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6950080394744873, + "num_tokens": 425625434.0, + "step": 16684 + }, + { + "epoch": 1.8323083681089392, + "grad_norm": 2.4741458892822266, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7178252339363098, + "num_tokens": 425647333.0, + "step": 16685 + }, + { + "epoch": 1.832418185811553, + "grad_norm": 2.273784637451172, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7287718653678894, + "num_tokens": 425670968.0, + "step": 16686 + }, + { + "epoch": 1.8325280035141664, + "grad_norm": 2.120325803756714, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7098526358604431, + "num_tokens": 425699408.0, + "step": 16687 + }, + { + "epoch": 1.83263782121678, + "grad_norm": 2.3995296955108643, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.738633394241333, + "num_tokens": 425721440.0, + "step": 16688 + }, + { + "epoch": 1.8327476389193937, + "grad_norm": 2.348701000213623, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7277647256851196, + "num_tokens": 425743258.0, + "step": 16689 + }, + { + "epoch": 1.8328574566220075, + "grad_norm": 2.3426358699798584, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7179446220397949, + "num_tokens": 425769695.0, + "step": 16690 + }, + { + "epoch": 1.8329672743246213, + "grad_norm": 2.263678789138794, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7045997381210327, + "num_tokens": 425794841.0, + "step": 16691 + }, + { + "epoch": 1.8330770920272348, + "grad_norm": 2.0503575801849365, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7129704356193542, + "num_tokens": 425823349.0, + "step": 16692 + }, + { + "epoch": 1.8331869097298483, + "grad_norm": 2.129946708679199, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7310816049575806, + "num_tokens": 425851580.0, + "step": 16693 + }, + { + "epoch": 1.833296727432462, + "grad_norm": 1.9478631019592285, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7098050117492676, + "num_tokens": 425884549.0, + "step": 16694 + }, + { + "epoch": 1.8334065451350758, + "grad_norm": 2.1024582386016846, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7323032021522522, + "num_tokens": 425910423.0, + "step": 16695 + }, + { + "epoch": 1.8335163628376896, + "grad_norm": 2.609602212905884, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7287323474884033, + "num_tokens": 425930001.0, + "step": 16696 + }, + { + "epoch": 1.8336261805403031, + "grad_norm": 2.0274581909179688, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7013813257217407, + "num_tokens": 425960891.0, + "step": 16697 + }, + { + "epoch": 1.8337359982429167, + "grad_norm": 2.3843204975128174, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7421354055404663, + "num_tokens": 425983674.0, + "step": 16698 + }, + { + "epoch": 1.8338458159455304, + "grad_norm": 2.171196699142456, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7218920588493347, + "num_tokens": 426012841.0, + "step": 16699 + }, + { + "epoch": 1.8339556336481442, + "grad_norm": 2.2685554027557373, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.72587651014328, + "num_tokens": 426038988.0, + "step": 16700 + }, + { + "epoch": 1.8340654513507577, + "grad_norm": 2.164219856262207, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7162965536117554, + "num_tokens": 426064315.0, + "step": 16701 + }, + { + "epoch": 1.8341752690533712, + "grad_norm": 2.6660029888153076, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7227567434310913, + "num_tokens": 426084952.0, + "step": 16702 + }, + { + "epoch": 1.834285086755985, + "grad_norm": 2.2605621814727783, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7152535319328308, + "num_tokens": 426110731.0, + "step": 16703 + }, + { + "epoch": 1.8343949044585988, + "grad_norm": 2.068730354309082, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.727576494216919, + "num_tokens": 426138554.0, + "step": 16704 + }, + { + "epoch": 1.8345047221612125, + "grad_norm": 2.0198974609375, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7122397422790527, + "num_tokens": 426170359.0, + "step": 16705 + }, + { + "epoch": 1.834614539863826, + "grad_norm": 2.2003753185272217, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7253015041351318, + "num_tokens": 426194380.0, + "step": 16706 + }, + { + "epoch": 1.8347243575664396, + "grad_norm": 2.4031646251678467, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7338076829910278, + "num_tokens": 426217176.0, + "step": 16707 + }, + { + "epoch": 1.8348341752690533, + "grad_norm": 2.2006335258483887, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.711899995803833, + "num_tokens": 426245202.0, + "step": 16708 + }, + { + "epoch": 1.834943992971667, + "grad_norm": 2.231130361557007, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7055851221084595, + "num_tokens": 426271799.0, + "step": 16709 + }, + { + "epoch": 1.8350538106742806, + "grad_norm": 2.26764178276062, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7121576070785522, + "num_tokens": 426296692.0, + "step": 16710 + }, + { + "epoch": 1.8351636283768944, + "grad_norm": 2.480776309967041, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7375322580337524, + "num_tokens": 426317574.0, + "step": 16711 + }, + { + "epoch": 1.835273446079508, + "grad_norm": 2.405409812927246, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7239267826080322, + "num_tokens": 426340832.0, + "step": 16712 + }, + { + "epoch": 1.8353832637821217, + "grad_norm": 2.44063138961792, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7222920656204224, + "num_tokens": 426364059.0, + "step": 16713 + }, + { + "epoch": 1.8354930814847354, + "grad_norm": 2.4417777061462402, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7268576622009277, + "num_tokens": 426387186.0, + "step": 16714 + }, + { + "epoch": 1.835602899187349, + "grad_norm": 2.050697088241577, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7224745750427246, + "num_tokens": 426415545.0, + "step": 16715 + }, + { + "epoch": 1.8357127168899625, + "grad_norm": 2.189945697784424, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.716300904750824, + "num_tokens": 426441724.0, + "step": 16716 + }, + { + "epoch": 1.8358225345925763, + "grad_norm": 2.0763044357299805, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.724946916103363, + "num_tokens": 426471789.0, + "step": 16717 + }, + { + "epoch": 1.83593235229519, + "grad_norm": 2.168686866760254, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7325624227523804, + "num_tokens": 426499668.0, + "step": 16718 + }, + { + "epoch": 1.8360421699978038, + "grad_norm": 1.999025821685791, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7018186450004578, + "num_tokens": 426534846.0, + "step": 16719 + }, + { + "epoch": 1.8361519877004173, + "grad_norm": 2.203582763671875, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7376782894134521, + "num_tokens": 426558980.0, + "step": 16720 + }, + { + "epoch": 1.8362618054030309, + "grad_norm": 2.2182886600494385, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7366124391555786, + "num_tokens": 426584471.0, + "step": 16721 + }, + { + "epoch": 1.8363716231056446, + "grad_norm": 2.05755615234375, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.709836483001709, + "num_tokens": 426613269.0, + "step": 16722 + }, + { + "epoch": 1.8364814408082584, + "grad_norm": 2.1816418170928955, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.6987491846084595, + "num_tokens": 426640222.0, + "step": 16723 + }, + { + "epoch": 1.836591258510872, + "grad_norm": 2.1780662536621094, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7145427465438843, + "num_tokens": 426669660.0, + "step": 16724 + }, + { + "epoch": 1.8367010762134857, + "grad_norm": 2.4449572563171387, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7444899678230286, + "num_tokens": 426691542.0, + "step": 16725 + }, + { + "epoch": 1.8368108939160992, + "grad_norm": 2.659388303756714, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7350635528564453, + "num_tokens": 426711029.0, + "step": 16726 + }, + { + "epoch": 1.836920711618713, + "grad_norm": 2.313291549682617, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7267240881919861, + "num_tokens": 426734772.0, + "step": 16727 + }, + { + "epoch": 1.8370305293213267, + "grad_norm": 2.5772500038146973, + "learning_rate": 1e-06, + "loss": 0.7558, + "mean_token_accuracy": 0.7544960975646973, + "num_tokens": 426754244.0, + "step": 16728 + }, + { + "epoch": 1.8371403470239402, + "grad_norm": 2.3666515350341797, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7457847595214844, + "num_tokens": 426778330.0, + "step": 16729 + }, + { + "epoch": 1.8372501647265538, + "grad_norm": 2.078874349594116, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7248448133468628, + "num_tokens": 426810252.0, + "step": 16730 + }, + { + "epoch": 1.8373599824291675, + "grad_norm": 2.5392343997955322, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7273581027984619, + "num_tokens": 426831234.0, + "step": 16731 + }, + { + "epoch": 1.8374698001317813, + "grad_norm": 2.1392009258270264, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7106254696846008, + "num_tokens": 426858265.0, + "step": 16732 + }, + { + "epoch": 1.837579617834395, + "grad_norm": 2.1931979656219482, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7209553718566895, + "num_tokens": 426883076.0, + "step": 16733 + }, + { + "epoch": 1.8376894355370086, + "grad_norm": 2.5423338413238525, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7091635465621948, + "num_tokens": 426904985.0, + "step": 16734 + }, + { + "epoch": 1.8377992532396221, + "grad_norm": 2.1991219520568848, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7241459488868713, + "num_tokens": 426928888.0, + "step": 16735 + }, + { + "epoch": 1.8379090709422359, + "grad_norm": 2.2941877841949463, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7265993356704712, + "num_tokens": 426953779.0, + "step": 16736 + }, + { + "epoch": 1.8380188886448496, + "grad_norm": 2.072049856185913, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7159686088562012, + "num_tokens": 426983710.0, + "step": 16737 + }, + { + "epoch": 1.8381287063474632, + "grad_norm": 2.706731081008911, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.731112539768219, + "num_tokens": 427003818.0, + "step": 16738 + }, + { + "epoch": 1.838238524050077, + "grad_norm": 2.026737928390503, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7196230292320251, + "num_tokens": 427033374.0, + "step": 16739 + }, + { + "epoch": 1.8383483417526905, + "grad_norm": 2.1991727352142334, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7120361328125, + "num_tokens": 427060673.0, + "step": 16740 + }, + { + "epoch": 1.8384581594553042, + "grad_norm": 2.160928249359131, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7259370684623718, + "num_tokens": 427088730.0, + "step": 16741 + }, + { + "epoch": 1.838567977157918, + "grad_norm": 2.359511613845825, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7401827573776245, + "num_tokens": 427109995.0, + "step": 16742 + }, + { + "epoch": 1.8386777948605315, + "grad_norm": 2.4365458488464355, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7052170634269714, + "num_tokens": 427132998.0, + "step": 16743 + }, + { + "epoch": 1.838787612563145, + "grad_norm": 2.449268341064453, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.725318431854248, + "num_tokens": 427154886.0, + "step": 16744 + }, + { + "epoch": 1.8388974302657588, + "grad_norm": 3.1088433265686035, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7553837299346924, + "num_tokens": 427170147.0, + "step": 16745 + }, + { + "epoch": 1.8390072479683726, + "grad_norm": 2.2922608852386475, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.696060299873352, + "num_tokens": 427198816.0, + "step": 16746 + }, + { + "epoch": 1.8391170656709863, + "grad_norm": 2.1445770263671875, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7257282733917236, + "num_tokens": 427224728.0, + "step": 16747 + }, + { + "epoch": 1.8392268833735999, + "grad_norm": 2.09547758102417, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7279942631721497, + "num_tokens": 427255162.0, + "step": 16748 + }, + { + "epoch": 1.8393367010762134, + "grad_norm": 2.1916708946228027, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7329411506652832, + "num_tokens": 427281331.0, + "step": 16749 + }, + { + "epoch": 1.8394465187788271, + "grad_norm": 2.4835524559020996, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7397752404212952, + "num_tokens": 427301000.0, + "step": 16750 + }, + { + "epoch": 1.839556336481441, + "grad_norm": 2.0827062129974365, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.6969712376594543, + "num_tokens": 427331581.0, + "step": 16751 + }, + { + "epoch": 1.8396661541840544, + "grad_norm": 2.121746063232422, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7050343155860901, + "num_tokens": 427359013.0, + "step": 16752 + }, + { + "epoch": 1.839775971886668, + "grad_norm": 2.2747883796691895, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7141934633255005, + "num_tokens": 427384892.0, + "step": 16753 + }, + { + "epoch": 1.8398857895892817, + "grad_norm": 2.245685577392578, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7275012731552124, + "num_tokens": 427410951.0, + "step": 16754 + }, + { + "epoch": 1.8399956072918955, + "grad_norm": 2.614954710006714, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7401180863380432, + "num_tokens": 427429969.0, + "step": 16755 + }, + { + "epoch": 1.8401054249945092, + "grad_norm": 2.30450701713562, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7333612442016602, + "num_tokens": 427452591.0, + "step": 16756 + }, + { + "epoch": 1.8402152426971228, + "grad_norm": 2.0916762351989746, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7252760529518127, + "num_tokens": 427478687.0, + "step": 16757 + }, + { + "epoch": 1.8403250603997363, + "grad_norm": 2.2779958248138428, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7139784693717957, + "num_tokens": 427505861.0, + "step": 16758 + }, + { + "epoch": 1.84043487810235, + "grad_norm": 2.3514387607574463, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7212567329406738, + "num_tokens": 427530040.0, + "step": 16759 + }, + { + "epoch": 1.8405446958049638, + "grad_norm": 2.3421928882598877, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7080178260803223, + "num_tokens": 427553199.0, + "step": 16760 + }, + { + "epoch": 1.8406545135075776, + "grad_norm": 1.9115575551986694, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7135761380195618, + "num_tokens": 427586064.0, + "step": 16761 + }, + { + "epoch": 1.8407643312101911, + "grad_norm": 2.2733168601989746, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7146443128585815, + "num_tokens": 427611835.0, + "step": 16762 + }, + { + "epoch": 1.8408741489128047, + "grad_norm": 2.1629574298858643, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.721260666847229, + "num_tokens": 427638926.0, + "step": 16763 + }, + { + "epoch": 1.8409839666154184, + "grad_norm": 2.2303307056427, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7176846861839294, + "num_tokens": 427665237.0, + "step": 16764 + }, + { + "epoch": 1.8410937843180322, + "grad_norm": 2.470569610595703, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.730285108089447, + "num_tokens": 427687365.0, + "step": 16765 + }, + { + "epoch": 1.8412036020206457, + "grad_norm": 2.1266274452209473, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7230764627456665, + "num_tokens": 427713758.0, + "step": 16766 + }, + { + "epoch": 1.8413134197232592, + "grad_norm": 1.9660723209381104, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7330162525177002, + "num_tokens": 427745894.0, + "step": 16767 + }, + { + "epoch": 1.841423237425873, + "grad_norm": 2.395944833755493, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7264459133148193, + "num_tokens": 427766381.0, + "step": 16768 + }, + { + "epoch": 1.8415330551284868, + "grad_norm": 2.1193368434906006, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7135593891143799, + "num_tokens": 427795835.0, + "step": 16769 + }, + { + "epoch": 1.8416428728311005, + "grad_norm": 2.066539764404297, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7095328569412231, + "num_tokens": 427827251.0, + "step": 16770 + }, + { + "epoch": 1.841752690533714, + "grad_norm": 2.208153009414673, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7272020578384399, + "num_tokens": 427853601.0, + "step": 16771 + }, + { + "epoch": 1.8418625082363276, + "grad_norm": 2.45708966255188, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7185822129249573, + "num_tokens": 427875565.0, + "step": 16772 + }, + { + "epoch": 1.8419723259389413, + "grad_norm": 2.378894329071045, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7062631845474243, + "num_tokens": 427901979.0, + "step": 16773 + }, + { + "epoch": 1.842082143641555, + "grad_norm": 2.0815377235412598, + "learning_rate": 1e-06, + "loss": 0.7618, + "mean_token_accuracy": 0.7503827214241028, + "num_tokens": 427927015.0, + "step": 16774 + }, + { + "epoch": 1.8421919613441686, + "grad_norm": 2.0127291679382324, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7181071043014526, + "num_tokens": 427957191.0, + "step": 16775 + }, + { + "epoch": 1.8423017790467824, + "grad_norm": 2.404576301574707, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.7603296041488647, + "num_tokens": 427980402.0, + "step": 16776 + }, + { + "epoch": 1.842411596749396, + "grad_norm": 2.0076279640197754, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7277675867080688, + "num_tokens": 428011044.0, + "step": 16777 + }, + { + "epoch": 1.8425214144520097, + "grad_norm": 2.1938796043395996, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7323470711708069, + "num_tokens": 428036193.0, + "step": 16778 + }, + { + "epoch": 1.8426312321546234, + "grad_norm": 2.5105457305908203, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7133222818374634, + "num_tokens": 428062023.0, + "step": 16779 + }, + { + "epoch": 1.842741049857237, + "grad_norm": 2.2784738540649414, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7288740277290344, + "num_tokens": 428087478.0, + "step": 16780 + }, + { + "epoch": 1.8428508675598505, + "grad_norm": 2.0279009342193604, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7288910150527954, + "num_tokens": 428115267.0, + "step": 16781 + }, + { + "epoch": 1.8429606852624643, + "grad_norm": 2.0911006927490234, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7046404480934143, + "num_tokens": 428142849.0, + "step": 16782 + }, + { + "epoch": 1.843070502965078, + "grad_norm": 2.2281124591827393, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7204358577728271, + "num_tokens": 428169307.0, + "step": 16783 + }, + { + "epoch": 1.8431803206676918, + "grad_norm": 2.469200372695923, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7349280714988708, + "num_tokens": 428190987.0, + "step": 16784 + }, + { + "epoch": 1.8432901383703053, + "grad_norm": 2.430877685546875, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7304745316505432, + "num_tokens": 428214583.0, + "step": 16785 + }, + { + "epoch": 1.8433999560729188, + "grad_norm": 2.120805501937866, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7218654751777649, + "num_tokens": 428241581.0, + "step": 16786 + }, + { + "epoch": 1.8435097737755326, + "grad_norm": 2.126622200012207, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.6975466012954712, + "num_tokens": 428269413.0, + "step": 16787 + }, + { + "epoch": 1.8436195914781464, + "grad_norm": 2.2679009437561035, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7289963960647583, + "num_tokens": 428294039.0, + "step": 16788 + }, + { + "epoch": 1.84372940918076, + "grad_norm": 2.252279758453369, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.735005795955658, + "num_tokens": 428318744.0, + "step": 16789 + }, + { + "epoch": 1.8438392268833736, + "grad_norm": 2.286423683166504, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7344118356704712, + "num_tokens": 428342846.0, + "step": 16790 + }, + { + "epoch": 1.8439490445859872, + "grad_norm": 2.2421698570251465, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7415232062339783, + "num_tokens": 428368713.0, + "step": 16791 + }, + { + "epoch": 1.844058862288601, + "grad_norm": 2.265056848526001, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7266685366630554, + "num_tokens": 428391944.0, + "step": 16792 + }, + { + "epoch": 1.8441686799912147, + "grad_norm": 2.3932173252105713, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7112345695495605, + "num_tokens": 428415595.0, + "step": 16793 + }, + { + "epoch": 1.8442784976938282, + "grad_norm": 2.3704779148101807, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7365075349807739, + "num_tokens": 428439346.0, + "step": 16794 + }, + { + "epoch": 1.8443883153964418, + "grad_norm": 2.0531041622161865, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7097716927528381, + "num_tokens": 428469500.0, + "step": 16795 + }, + { + "epoch": 1.8444981330990555, + "grad_norm": 2.1568140983581543, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7001047134399414, + "num_tokens": 428496118.0, + "step": 16796 + }, + { + "epoch": 1.8446079508016693, + "grad_norm": 2.008517265319824, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7136794328689575, + "num_tokens": 428525695.0, + "step": 16797 + }, + { + "epoch": 1.844717768504283, + "grad_norm": 2.139467477798462, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7019799947738647, + "num_tokens": 428556298.0, + "step": 16798 + }, + { + "epoch": 1.8448275862068966, + "grad_norm": 2.4502921104431152, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7222704887390137, + "num_tokens": 428577488.0, + "step": 16799 + }, + { + "epoch": 1.84493740390951, + "grad_norm": 2.1564769744873047, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6940422058105469, + "num_tokens": 428605931.0, + "step": 16800 + }, + { + "epoch": 1.8450472216121239, + "grad_norm": 2.494357109069824, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.6971816420555115, + "num_tokens": 428628957.0, + "step": 16801 + }, + { + "epoch": 1.8451570393147376, + "grad_norm": 2.4393320083618164, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7223432064056396, + "num_tokens": 428650473.0, + "step": 16802 + }, + { + "epoch": 1.8452668570173512, + "grad_norm": 2.412670135498047, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7179253101348877, + "num_tokens": 428673072.0, + "step": 16803 + }, + { + "epoch": 1.8453766747199647, + "grad_norm": 2.1744258403778076, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7231538891792297, + "num_tokens": 428699851.0, + "step": 16804 + }, + { + "epoch": 1.8454864924225785, + "grad_norm": 2.1626439094543457, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7118890285491943, + "num_tokens": 428726065.0, + "step": 16805 + }, + { + "epoch": 1.8455963101251922, + "grad_norm": 2.040691614151001, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7161343693733215, + "num_tokens": 428756566.0, + "step": 16806 + }, + { + "epoch": 1.845706127827806, + "grad_norm": 2.3227851390838623, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7328250408172607, + "num_tokens": 428780473.0, + "step": 16807 + }, + { + "epoch": 1.8458159455304195, + "grad_norm": 2.2738986015319824, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7329882383346558, + "num_tokens": 428806495.0, + "step": 16808 + }, + { + "epoch": 1.845925763233033, + "grad_norm": 2.064352035522461, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.724037766456604, + "num_tokens": 428834571.0, + "step": 16809 + }, + { + "epoch": 1.8460355809356468, + "grad_norm": 2.0788164138793945, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7083834409713745, + "num_tokens": 428863648.0, + "step": 16810 + }, + { + "epoch": 1.8461453986382605, + "grad_norm": 2.6967873573303223, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7451087236404419, + "num_tokens": 428882322.0, + "step": 16811 + }, + { + "epoch": 1.8462552163408743, + "grad_norm": 2.119447946548462, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7099963426589966, + "num_tokens": 428909706.0, + "step": 16812 + }, + { + "epoch": 1.8463650340434878, + "grad_norm": 2.301431894302368, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7319518327713013, + "num_tokens": 428931054.0, + "step": 16813 + }, + { + "epoch": 1.8464748517461014, + "grad_norm": 2.30025053024292, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7136940956115723, + "num_tokens": 428956350.0, + "step": 16814 + }, + { + "epoch": 1.8465846694487151, + "grad_norm": 2.1545135974884033, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.723922848701477, + "num_tokens": 428985462.0, + "step": 16815 + }, + { + "epoch": 1.846694487151329, + "grad_norm": 2.1132614612579346, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7269372940063477, + "num_tokens": 429011570.0, + "step": 16816 + }, + { + "epoch": 1.8468043048539424, + "grad_norm": 2.105259895324707, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7296004295349121, + "num_tokens": 429041256.0, + "step": 16817 + }, + { + "epoch": 1.846914122556556, + "grad_norm": 2.744297504425049, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7293013334274292, + "num_tokens": 429058739.0, + "step": 16818 + }, + { + "epoch": 1.8470239402591697, + "grad_norm": 2.1801369190216064, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7258133888244629, + "num_tokens": 429084586.0, + "step": 16819 + }, + { + "epoch": 1.8471337579617835, + "grad_norm": 2.5455965995788574, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7304106950759888, + "num_tokens": 429105449.0, + "step": 16820 + }, + { + "epoch": 1.8472435756643972, + "grad_norm": 2.3269269466400146, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.707868218421936, + "num_tokens": 429131129.0, + "step": 16821 + }, + { + "epoch": 1.8473533933670108, + "grad_norm": 2.182453155517578, + "learning_rate": 1e-06, + "loss": 0.8269, + "mean_token_accuracy": 0.7393736243247986, + "num_tokens": 429155537.0, + "step": 16822 + }, + { + "epoch": 1.8474632110696243, + "grad_norm": 1.9849251508712769, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7214482426643372, + "num_tokens": 429184922.0, + "step": 16823 + }, + { + "epoch": 1.847573028772238, + "grad_norm": 2.1404407024383545, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7365075349807739, + "num_tokens": 429211035.0, + "step": 16824 + }, + { + "epoch": 1.8476828464748518, + "grad_norm": 1.975914478302002, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.703768253326416, + "num_tokens": 429242984.0, + "step": 16825 + }, + { + "epoch": 1.8477926641774656, + "grad_norm": 2.020883798599243, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7387588024139404, + "num_tokens": 429271621.0, + "step": 16826 + }, + { + "epoch": 1.847902481880079, + "grad_norm": 2.1917450428009033, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7200220823287964, + "num_tokens": 429296850.0, + "step": 16827 + }, + { + "epoch": 1.8480122995826926, + "grad_norm": 2.559724807739258, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7338125109672546, + "num_tokens": 429316704.0, + "step": 16828 + }, + { + "epoch": 1.8481221172853064, + "grad_norm": 2.086432933807373, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7331569790840149, + "num_tokens": 429343004.0, + "step": 16829 + }, + { + "epoch": 1.8482319349879202, + "grad_norm": 2.3365797996520996, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7380495071411133, + "num_tokens": 429365869.0, + "step": 16830 + }, + { + "epoch": 1.8483417526905337, + "grad_norm": 1.9728037118911743, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7094584703445435, + "num_tokens": 429397908.0, + "step": 16831 + }, + { + "epoch": 1.8484515703931472, + "grad_norm": 2.2216882705688477, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7203036546707153, + "num_tokens": 429422955.0, + "step": 16832 + }, + { + "epoch": 1.848561388095761, + "grad_norm": 2.11824893951416, + "learning_rate": 1e-06, + "loss": 1.0619, + "mean_token_accuracy": 0.6783528923988342, + "num_tokens": 429453080.0, + "step": 16833 + }, + { + "epoch": 1.8486712057983747, + "grad_norm": 2.168635845184326, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7293412685394287, + "num_tokens": 429479128.0, + "step": 16834 + }, + { + "epoch": 1.8487810235009885, + "grad_norm": 2.2445156574249268, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7192521095275879, + "num_tokens": 429504313.0, + "step": 16835 + }, + { + "epoch": 1.848890841203602, + "grad_norm": 2.042628049850464, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.6985697746276855, + "num_tokens": 429535672.0, + "step": 16836 + }, + { + "epoch": 1.8490006589062156, + "grad_norm": 2.184213161468506, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7034998536109924, + "num_tokens": 429564004.0, + "step": 16837 + }, + { + "epoch": 1.8491104766088293, + "grad_norm": 2.5705857276916504, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7319661974906921, + "num_tokens": 429584734.0, + "step": 16838 + }, + { + "epoch": 1.849220294311443, + "grad_norm": 2.332329034805298, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7255113124847412, + "num_tokens": 429610218.0, + "step": 16839 + }, + { + "epoch": 1.8493301120140566, + "grad_norm": 1.8912051916122437, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7025496959686279, + "num_tokens": 429644881.0, + "step": 16840 + }, + { + "epoch": 1.8494399297166704, + "grad_norm": 2.57867431640625, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7493799328804016, + "num_tokens": 429664220.0, + "step": 16841 + }, + { + "epoch": 1.849549747419284, + "grad_norm": 1.883225917816162, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.6985265016555786, + "num_tokens": 429700842.0, + "step": 16842 + }, + { + "epoch": 1.8496595651218977, + "grad_norm": 2.1855218410491943, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7235537767410278, + "num_tokens": 429725672.0, + "step": 16843 + }, + { + "epoch": 1.8497693828245114, + "grad_norm": 2.5419068336486816, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7370083928108215, + "num_tokens": 429745627.0, + "step": 16844 + }, + { + "epoch": 1.849879200527125, + "grad_norm": 1.9637789726257324, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7274279594421387, + "num_tokens": 429776044.0, + "step": 16845 + }, + { + "epoch": 1.8499890182297385, + "grad_norm": 2.1290271282196045, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.6976004838943481, + "num_tokens": 429803669.0, + "step": 16846 + }, + { + "epoch": 1.8500988359323522, + "grad_norm": 2.2118704319000244, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7010048031806946, + "num_tokens": 429833359.0, + "step": 16847 + }, + { + "epoch": 1.850208653634966, + "grad_norm": 2.2509686946868896, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7173927426338196, + "num_tokens": 429859662.0, + "step": 16848 + }, + { + "epoch": 1.8503184713375798, + "grad_norm": 2.4791321754455566, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7259509563446045, + "num_tokens": 429880973.0, + "step": 16849 + }, + { + "epoch": 1.8504282890401933, + "grad_norm": 2.110322952270508, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7277929782867432, + "num_tokens": 429910104.0, + "step": 16850 + }, + { + "epoch": 1.8505381067428068, + "grad_norm": 2.0581116676330566, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7209571599960327, + "num_tokens": 429938197.0, + "step": 16851 + }, + { + "epoch": 1.8506479244454206, + "grad_norm": 2.390336513519287, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.7400182485580444, + "num_tokens": 429959645.0, + "step": 16852 + }, + { + "epoch": 1.8507577421480343, + "grad_norm": 2.3117334842681885, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7242211103439331, + "num_tokens": 429982467.0, + "step": 16853 + }, + { + "epoch": 1.8508675598506479, + "grad_norm": 2.2998578548431396, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7227615118026733, + "num_tokens": 430007723.0, + "step": 16854 + }, + { + "epoch": 1.8509773775532616, + "grad_norm": 2.385711193084717, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7083988189697266, + "num_tokens": 430032268.0, + "step": 16855 + }, + { + "epoch": 1.8510871952558752, + "grad_norm": 2.175854444503784, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7170230150222778, + "num_tokens": 430060204.0, + "step": 16856 + }, + { + "epoch": 1.851197012958489, + "grad_norm": 2.1396517753601074, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7251479625701904, + "num_tokens": 430086168.0, + "step": 16857 + }, + { + "epoch": 1.8513068306611027, + "grad_norm": 2.2294697761535645, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.6972798705101013, + "num_tokens": 430112781.0, + "step": 16858 + }, + { + "epoch": 1.8514166483637162, + "grad_norm": 2.4353644847869873, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7203013896942139, + "num_tokens": 430135314.0, + "step": 16859 + }, + { + "epoch": 1.8515264660663298, + "grad_norm": 2.3494062423706055, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7375931739807129, + "num_tokens": 430158386.0, + "step": 16860 + }, + { + "epoch": 1.8516362837689435, + "grad_norm": 2.224694013595581, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7386622428894043, + "num_tokens": 430183098.0, + "step": 16861 + }, + { + "epoch": 1.8517461014715573, + "grad_norm": 2.3170878887176514, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7351067066192627, + "num_tokens": 430207302.0, + "step": 16862 + }, + { + "epoch": 1.851855919174171, + "grad_norm": 2.1178719997406006, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7177290916442871, + "num_tokens": 430233925.0, + "step": 16863 + }, + { + "epoch": 1.8519657368767846, + "grad_norm": 2.3666324615478516, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7427673935890198, + "num_tokens": 430255311.0, + "step": 16864 + }, + { + "epoch": 1.852075554579398, + "grad_norm": 1.9490139484405518, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.710185170173645, + "num_tokens": 430286039.0, + "step": 16865 + }, + { + "epoch": 1.8521853722820119, + "grad_norm": 2.2564992904663086, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7074501514434814, + "num_tokens": 430311629.0, + "step": 16866 + }, + { + "epoch": 1.8522951899846256, + "grad_norm": 1.9928531646728516, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.715032160282135, + "num_tokens": 430344015.0, + "step": 16867 + }, + { + "epoch": 1.8524050076872391, + "grad_norm": 2.2216796875, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7234631776809692, + "num_tokens": 430369782.0, + "step": 16868 + }, + { + "epoch": 1.8525148253898527, + "grad_norm": 2.198925733566284, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7188595533370972, + "num_tokens": 430397292.0, + "step": 16869 + }, + { + "epoch": 1.8526246430924664, + "grad_norm": 2.0132486820220947, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7394523620605469, + "num_tokens": 430425556.0, + "step": 16870 + }, + { + "epoch": 1.8527344607950802, + "grad_norm": 2.1250081062316895, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7182961106300354, + "num_tokens": 430453064.0, + "step": 16871 + }, + { + "epoch": 1.852844278497694, + "grad_norm": 2.1476762294769287, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7211849689483643, + "num_tokens": 430482460.0, + "step": 16872 + }, + { + "epoch": 1.8529540962003075, + "grad_norm": 2.2554240226745605, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7164098024368286, + "num_tokens": 430508006.0, + "step": 16873 + }, + { + "epoch": 1.853063913902921, + "grad_norm": 2.354733943939209, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7181602716445923, + "num_tokens": 430532893.0, + "step": 16874 + }, + { + "epoch": 1.8531737316055348, + "grad_norm": 2.2658236026763916, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.716801106929779, + "num_tokens": 430559360.0, + "step": 16875 + }, + { + "epoch": 1.8532835493081485, + "grad_norm": 2.252699375152588, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7314713001251221, + "num_tokens": 430583618.0, + "step": 16876 + }, + { + "epoch": 1.8533933670107623, + "grad_norm": 2.365727424621582, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7058658003807068, + "num_tokens": 430608362.0, + "step": 16877 + }, + { + "epoch": 1.8535031847133758, + "grad_norm": 2.1874430179595947, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.7578318119049072, + "num_tokens": 430633203.0, + "step": 16878 + }, + { + "epoch": 1.8536130024159894, + "grad_norm": 1.9327080249786377, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7272587418556213, + "num_tokens": 430665025.0, + "step": 16879 + }, + { + "epoch": 1.8537228201186031, + "grad_norm": 2.437532424926758, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7479645013809204, + "num_tokens": 430687746.0, + "step": 16880 + }, + { + "epoch": 1.8538326378212169, + "grad_norm": 2.261284589767456, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7038577198982239, + "num_tokens": 430713477.0, + "step": 16881 + }, + { + "epoch": 1.8539424555238304, + "grad_norm": 2.1274704933166504, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7161012887954712, + "num_tokens": 430740923.0, + "step": 16882 + }, + { + "epoch": 1.854052273226444, + "grad_norm": 2.052759885787964, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7170844674110413, + "num_tokens": 430769998.0, + "step": 16883 + }, + { + "epoch": 1.8541620909290577, + "grad_norm": 2.464099884033203, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7392113208770752, + "num_tokens": 430790695.0, + "step": 16884 + }, + { + "epoch": 1.8542719086316715, + "grad_norm": 2.45283579826355, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7467423677444458, + "num_tokens": 430811072.0, + "step": 16885 + }, + { + "epoch": 1.8543817263342852, + "grad_norm": 2.465738296508789, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.740467369556427, + "num_tokens": 430832050.0, + "step": 16886 + }, + { + "epoch": 1.8544915440368988, + "grad_norm": 2.307044744491577, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7146803140640259, + "num_tokens": 430855372.0, + "step": 16887 + }, + { + "epoch": 1.8546013617395123, + "grad_norm": 2.2141072750091553, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.709844172000885, + "num_tokens": 430883741.0, + "step": 16888 + }, + { + "epoch": 1.854711179442126, + "grad_norm": 2.1874947547912598, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.718048095703125, + "num_tokens": 430909492.0, + "step": 16889 + }, + { + "epoch": 1.8548209971447398, + "grad_norm": 2.5786972045898438, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.741635799407959, + "num_tokens": 430929736.0, + "step": 16890 + }, + { + "epoch": 1.8549308148473536, + "grad_norm": 2.2080111503601074, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.6959666609764099, + "num_tokens": 430959194.0, + "step": 16891 + }, + { + "epoch": 1.855040632549967, + "grad_norm": 2.4887216091156006, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7354595065116882, + "num_tokens": 430978525.0, + "step": 16892 + }, + { + "epoch": 1.8551504502525806, + "grad_norm": 2.4894795417785645, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.715049147605896, + "num_tokens": 431000606.0, + "step": 16893 + }, + { + "epoch": 1.8552602679551944, + "grad_norm": 2.1927988529205322, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6910080909729004, + "num_tokens": 431028902.0, + "step": 16894 + }, + { + "epoch": 1.8553700856578081, + "grad_norm": 2.48555850982666, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7291832566261292, + "num_tokens": 431050806.0, + "step": 16895 + }, + { + "epoch": 1.8554799033604217, + "grad_norm": 2.300684928894043, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.726176917552948, + "num_tokens": 431075169.0, + "step": 16896 + }, + { + "epoch": 1.8555897210630352, + "grad_norm": 2.2674500942230225, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7102890014648438, + "num_tokens": 431101343.0, + "step": 16897 + }, + { + "epoch": 1.855699538765649, + "grad_norm": 2.3992767333984375, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7351049184799194, + "num_tokens": 431123478.0, + "step": 16898 + }, + { + "epoch": 1.8558093564682627, + "grad_norm": 2.238013982772827, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.6961448192596436, + "num_tokens": 431152719.0, + "step": 16899 + }, + { + "epoch": 1.8559191741708765, + "grad_norm": 2.3206369876861572, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7091495990753174, + "num_tokens": 431175927.0, + "step": 16900 + }, + { + "epoch": 1.85602899187349, + "grad_norm": 2.184030055999756, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.730431079864502, + "num_tokens": 431203966.0, + "step": 16901 + }, + { + "epoch": 1.8561388095761036, + "grad_norm": 2.341273307800293, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7215219736099243, + "num_tokens": 431226622.0, + "step": 16902 + }, + { + "epoch": 1.8562486272787173, + "grad_norm": 2.487593412399292, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7010631561279297, + "num_tokens": 431250553.0, + "step": 16903 + }, + { + "epoch": 1.856358444981331, + "grad_norm": 2.086784839630127, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7117838263511658, + "num_tokens": 431279074.0, + "step": 16904 + }, + { + "epoch": 1.8564682626839446, + "grad_norm": 2.0674545764923096, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7256783843040466, + "num_tokens": 431309623.0, + "step": 16905 + }, + { + "epoch": 1.8565780803865584, + "grad_norm": 2.2278971672058105, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7103441953659058, + "num_tokens": 431335932.0, + "step": 16906 + }, + { + "epoch": 1.856687898089172, + "grad_norm": 2.3174872398376465, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7313954830169678, + "num_tokens": 431360262.0, + "step": 16907 + }, + { + "epoch": 1.8567977157917857, + "grad_norm": 2.1999385356903076, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.734603762626648, + "num_tokens": 431385833.0, + "step": 16908 + }, + { + "epoch": 1.8569075334943994, + "grad_norm": 2.6376302242279053, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7017975449562073, + "num_tokens": 431408217.0, + "step": 16909 + }, + { + "epoch": 1.857017351197013, + "grad_norm": 1.981292724609375, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7150149941444397, + "num_tokens": 431440208.0, + "step": 16910 + }, + { + "epoch": 1.8571271688996265, + "grad_norm": 2.178521156311035, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7008445262908936, + "num_tokens": 431468391.0, + "step": 16911 + }, + { + "epoch": 1.8572369866022402, + "grad_norm": 2.351465940475464, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7374330759048462, + "num_tokens": 431490116.0, + "step": 16912 + }, + { + "epoch": 1.857346804304854, + "grad_norm": 2.239898443222046, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7059632539749146, + "num_tokens": 431516311.0, + "step": 16913 + }, + { + "epoch": 1.8574566220074678, + "grad_norm": 2.349885940551758, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7246850728988647, + "num_tokens": 431539467.0, + "step": 16914 + }, + { + "epoch": 1.8575664397100813, + "grad_norm": 2.187281608581543, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7432434558868408, + "num_tokens": 431564346.0, + "step": 16915 + }, + { + "epoch": 1.8576762574126948, + "grad_norm": 2.4510295391082764, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7186052799224854, + "num_tokens": 431587485.0, + "step": 16916 + }, + { + "epoch": 1.8577860751153086, + "grad_norm": 2.119950771331787, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6930720210075378, + "num_tokens": 431616460.0, + "step": 16917 + }, + { + "epoch": 1.8578958928179223, + "grad_norm": 2.0483059883117676, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7062766551971436, + "num_tokens": 431647859.0, + "step": 16918 + }, + { + "epoch": 1.8580057105205359, + "grad_norm": 2.331399440765381, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7023699879646301, + "num_tokens": 431671041.0, + "step": 16919 + }, + { + "epoch": 1.8581155282231496, + "grad_norm": 2.607839345932007, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.7364609837532043, + "num_tokens": 431691110.0, + "step": 16920 + }, + { + "epoch": 1.8582253459257632, + "grad_norm": 2.0814826488494873, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6914976835250854, + "num_tokens": 431721037.0, + "step": 16921 + }, + { + "epoch": 1.858335163628377, + "grad_norm": 2.1756629943847656, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7040577530860901, + "num_tokens": 431749790.0, + "step": 16922 + }, + { + "epoch": 1.8584449813309907, + "grad_norm": 2.813551664352417, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7406213283538818, + "num_tokens": 431767474.0, + "step": 16923 + }, + { + "epoch": 1.8585547990336042, + "grad_norm": 2.313178539276123, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7252039909362793, + "num_tokens": 431791833.0, + "step": 16924 + }, + { + "epoch": 1.8586646167362177, + "grad_norm": 2.048819065093994, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7315534353256226, + "num_tokens": 431820061.0, + "step": 16925 + }, + { + "epoch": 1.8587744344388315, + "grad_norm": 1.9985681772232056, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7255474328994751, + "num_tokens": 431849950.0, + "step": 16926 + }, + { + "epoch": 1.8588842521414453, + "grad_norm": 2.208360433578491, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7120994925498962, + "num_tokens": 431875464.0, + "step": 16927 + }, + { + "epoch": 1.858994069844059, + "grad_norm": 2.3787460327148438, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7326400876045227, + "num_tokens": 431898358.0, + "step": 16928 + }, + { + "epoch": 1.8591038875466726, + "grad_norm": 2.123272657394409, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6865593791007996, + "num_tokens": 431928615.0, + "step": 16929 + }, + { + "epoch": 1.859213705249286, + "grad_norm": 2.3292102813720703, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7051777839660645, + "num_tokens": 431953326.0, + "step": 16930 + }, + { + "epoch": 1.8593235229518998, + "grad_norm": 2.752466917037964, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7461740970611572, + "num_tokens": 431972808.0, + "step": 16931 + }, + { + "epoch": 1.8594333406545136, + "grad_norm": 2.5784518718719482, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7254456877708435, + "num_tokens": 431994284.0, + "step": 16932 + }, + { + "epoch": 1.8595431583571271, + "grad_norm": 2.2355291843414307, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7196647524833679, + "num_tokens": 432020340.0, + "step": 16933 + }, + { + "epoch": 1.8596529760597407, + "grad_norm": 2.1495699882507324, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.71351158618927, + "num_tokens": 432047847.0, + "step": 16934 + }, + { + "epoch": 1.8597627937623544, + "grad_norm": 2.239072322845459, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7404643297195435, + "num_tokens": 432071392.0, + "step": 16935 + }, + { + "epoch": 1.8598726114649682, + "grad_norm": 2.195382595062256, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.700289249420166, + "num_tokens": 432097713.0, + "step": 16936 + }, + { + "epoch": 1.859982429167582, + "grad_norm": 2.429643392562866, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7207673788070679, + "num_tokens": 432120466.0, + "step": 16937 + }, + { + "epoch": 1.8600922468701955, + "grad_norm": 2.511148691177368, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7214134931564331, + "num_tokens": 432142351.0, + "step": 16938 + }, + { + "epoch": 1.860202064572809, + "grad_norm": 2.527517080307007, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7508621215820312, + "num_tokens": 432161872.0, + "step": 16939 + }, + { + "epoch": 1.8603118822754228, + "grad_norm": 2.065052032470703, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7183008790016174, + "num_tokens": 432190297.0, + "step": 16940 + }, + { + "epoch": 1.8604216999780365, + "grad_norm": 2.18406081199646, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7169491052627563, + "num_tokens": 432215715.0, + "step": 16941 + }, + { + "epoch": 1.8605315176806503, + "grad_norm": 2.237583875656128, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7321027517318726, + "num_tokens": 432240154.0, + "step": 16942 + }, + { + "epoch": 1.8606413353832638, + "grad_norm": 1.9470876455307007, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7241744995117188, + "num_tokens": 432271797.0, + "step": 16943 + }, + { + "epoch": 1.8607511530858774, + "grad_norm": 2.322155475616455, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7343906164169312, + "num_tokens": 432295510.0, + "step": 16944 + }, + { + "epoch": 1.860860970788491, + "grad_norm": 2.226144313812256, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7189666628837585, + "num_tokens": 432322457.0, + "step": 16945 + }, + { + "epoch": 1.8609707884911049, + "grad_norm": 2.248403549194336, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7182044982910156, + "num_tokens": 432348035.0, + "step": 16946 + }, + { + "epoch": 1.8610806061937184, + "grad_norm": 2.2321059703826904, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.7028623819351196, + "num_tokens": 432375807.0, + "step": 16947 + }, + { + "epoch": 1.861190423896332, + "grad_norm": 2.2395243644714355, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7245146036148071, + "num_tokens": 432400823.0, + "step": 16948 + }, + { + "epoch": 1.8613002415989457, + "grad_norm": 1.9024466276168823, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7202945947647095, + "num_tokens": 432435703.0, + "step": 16949 + }, + { + "epoch": 1.8614100593015594, + "grad_norm": 1.9213920831680298, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.6923367977142334, + "num_tokens": 432469823.0, + "step": 16950 + }, + { + "epoch": 1.8615198770041732, + "grad_norm": 2.247147560119629, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7207659482955933, + "num_tokens": 432494893.0, + "step": 16951 + }, + { + "epoch": 1.8616296947067867, + "grad_norm": 2.400059223175049, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7279070615768433, + "num_tokens": 432516650.0, + "step": 16952 + }, + { + "epoch": 1.8617395124094003, + "grad_norm": 2.314316987991333, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7031905651092529, + "num_tokens": 432540569.0, + "step": 16953 + }, + { + "epoch": 1.861849330112014, + "grad_norm": 2.213593006134033, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7139620184898376, + "num_tokens": 432566566.0, + "step": 16954 + }, + { + "epoch": 1.8619591478146278, + "grad_norm": 2.3282811641693115, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7331175804138184, + "num_tokens": 432589527.0, + "step": 16955 + }, + { + "epoch": 1.8620689655172413, + "grad_norm": 2.1003589630126953, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.6988571286201477, + "num_tokens": 432618217.0, + "step": 16956 + }, + { + "epoch": 1.862178783219855, + "grad_norm": 2.04498291015625, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7016386985778809, + "num_tokens": 432646091.0, + "step": 16957 + }, + { + "epoch": 1.8622886009224686, + "grad_norm": 2.211376905441284, + "learning_rate": 1e-06, + "loss": 1.0307, + "mean_token_accuracy": 0.6964215040206909, + "num_tokens": 432675505.0, + "step": 16958 + }, + { + "epoch": 1.8623984186250824, + "grad_norm": 2.1210994720458984, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7093607187271118, + "num_tokens": 432705076.0, + "step": 16959 + }, + { + "epoch": 1.8625082363276961, + "grad_norm": 1.8846638202667236, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7129451036453247, + "num_tokens": 432737813.0, + "step": 16960 + }, + { + "epoch": 1.8626180540303097, + "grad_norm": 2.189262628555298, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7208985090255737, + "num_tokens": 432762571.0, + "step": 16961 + }, + { + "epoch": 1.8627278717329232, + "grad_norm": 2.475613832473755, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7148925065994263, + "num_tokens": 432784573.0, + "step": 16962 + }, + { + "epoch": 1.862837689435537, + "grad_norm": 2.199664354324341, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6913597583770752, + "num_tokens": 432811650.0, + "step": 16963 + }, + { + "epoch": 1.8629475071381507, + "grad_norm": 2.4828131198883057, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7111105918884277, + "num_tokens": 432835192.0, + "step": 16964 + }, + { + "epoch": 1.8630573248407645, + "grad_norm": 2.0791878700256348, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7072548270225525, + "num_tokens": 432865994.0, + "step": 16965 + }, + { + "epoch": 1.863167142543378, + "grad_norm": 2.1165449619293213, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7328154444694519, + "num_tokens": 432893234.0, + "step": 16966 + }, + { + "epoch": 1.8632769602459915, + "grad_norm": 2.694293260574341, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7277549505233765, + "num_tokens": 432912366.0, + "step": 16967 + }, + { + "epoch": 1.8633867779486053, + "grad_norm": 2.242192506790161, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.7088488340377808, + "num_tokens": 432940089.0, + "step": 16968 + }, + { + "epoch": 1.863496595651219, + "grad_norm": 1.9935989379882812, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.694943904876709, + "num_tokens": 432970869.0, + "step": 16969 + }, + { + "epoch": 1.8636064133538326, + "grad_norm": 2.512803792953491, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6962601542472839, + "num_tokens": 433001189.0, + "step": 16970 + }, + { + "epoch": 1.8637162310564463, + "grad_norm": 2.4242944717407227, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.719179093837738, + "num_tokens": 433023436.0, + "step": 16971 + }, + { + "epoch": 1.8638260487590599, + "grad_norm": 2.24131441116333, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7285251021385193, + "num_tokens": 433048103.0, + "step": 16972 + }, + { + "epoch": 1.8639358664616736, + "grad_norm": 2.300260066986084, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7300119400024414, + "num_tokens": 433071074.0, + "step": 16973 + }, + { + "epoch": 1.8640456841642874, + "grad_norm": 2.0895676612854004, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7078736424446106, + "num_tokens": 433100611.0, + "step": 16974 + }, + { + "epoch": 1.864155501866901, + "grad_norm": 2.0732059478759766, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7180628776550293, + "num_tokens": 433129525.0, + "step": 16975 + }, + { + "epoch": 1.8642653195695145, + "grad_norm": 2.034517526626587, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7202418446540833, + "num_tokens": 433157242.0, + "step": 16976 + }, + { + "epoch": 1.8643751372721282, + "grad_norm": 2.269906997680664, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7236521244049072, + "num_tokens": 433180459.0, + "step": 16977 + }, + { + "epoch": 1.864484954974742, + "grad_norm": 2.2349493503570557, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7228569984436035, + "num_tokens": 433207276.0, + "step": 16978 + }, + { + "epoch": 1.8645947726773557, + "grad_norm": 2.7304906845092773, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7424007058143616, + "num_tokens": 433225182.0, + "step": 16979 + }, + { + "epoch": 1.8647045903799693, + "grad_norm": 2.193087100982666, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7037571668624878, + "num_tokens": 433251842.0, + "step": 16980 + }, + { + "epoch": 1.8648144080825828, + "grad_norm": 2.0733602046966553, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7270553112030029, + "num_tokens": 433278482.0, + "step": 16981 + }, + { + "epoch": 1.8649242257851966, + "grad_norm": 2.352278709411621, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7077836990356445, + "num_tokens": 433303437.0, + "step": 16982 + }, + { + "epoch": 1.8650340434878103, + "grad_norm": 2.4459784030914307, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.712475597858429, + "num_tokens": 433326652.0, + "step": 16983 + }, + { + "epoch": 1.8651438611904239, + "grad_norm": 2.2434558868408203, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.726820707321167, + "num_tokens": 433351385.0, + "step": 16984 + }, + { + "epoch": 1.8652536788930374, + "grad_norm": 2.2672009468078613, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7238650918006897, + "num_tokens": 433375399.0, + "step": 16985 + }, + { + "epoch": 1.8653634965956511, + "grad_norm": 2.1953420639038086, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.6979701519012451, + "num_tokens": 433404795.0, + "step": 16986 + }, + { + "epoch": 1.865473314298265, + "grad_norm": 2.172574520111084, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7187987565994263, + "num_tokens": 433433250.0, + "step": 16987 + }, + { + "epoch": 1.8655831320008787, + "grad_norm": 2.327669143676758, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7092725038528442, + "num_tokens": 433459084.0, + "step": 16988 + }, + { + "epoch": 1.8656929497034922, + "grad_norm": 2.366250514984131, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7200692892074585, + "num_tokens": 433483318.0, + "step": 16989 + }, + { + "epoch": 1.8658027674061057, + "grad_norm": 2.480905771255493, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7221845984458923, + "num_tokens": 433504987.0, + "step": 16990 + }, + { + "epoch": 1.8659125851087195, + "grad_norm": 2.3711695671081543, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7162911891937256, + "num_tokens": 433530417.0, + "step": 16991 + }, + { + "epoch": 1.8660224028113332, + "grad_norm": 2.2091474533081055, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.7025669813156128, + "num_tokens": 433558838.0, + "step": 16992 + }, + { + "epoch": 1.866132220513947, + "grad_norm": 2.527662754058838, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7360232472419739, + "num_tokens": 433580190.0, + "step": 16993 + }, + { + "epoch": 1.8662420382165605, + "grad_norm": 2.1647889614105225, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7283639907836914, + "num_tokens": 433606967.0, + "step": 16994 + }, + { + "epoch": 1.866351855919174, + "grad_norm": 2.5422861576080322, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7437744140625, + "num_tokens": 433626938.0, + "step": 16995 + }, + { + "epoch": 1.8664616736217878, + "grad_norm": 2.389188051223755, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7316619157791138, + "num_tokens": 433649961.0, + "step": 16996 + }, + { + "epoch": 1.8665714913244016, + "grad_norm": 2.4173085689544678, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7393909096717834, + "num_tokens": 433671302.0, + "step": 16997 + }, + { + "epoch": 1.8666813090270151, + "grad_norm": 2.1911675930023193, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7181214094161987, + "num_tokens": 433698521.0, + "step": 16998 + }, + { + "epoch": 1.8667911267296287, + "grad_norm": 2.5221216678619385, + "learning_rate": 1e-06, + "loss": 0.7814, + "mean_token_accuracy": 0.7500786781311035, + "num_tokens": 433719640.0, + "step": 16999 + }, + { + "epoch": 1.8669009444322424, + "grad_norm": 2.059692144393921, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7524992227554321, + "num_tokens": 433747753.0, + "step": 17000 + }, + { + "epoch": 1.8670107621348562, + "grad_norm": 2.091099977493286, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7128263711929321, + "num_tokens": 433775398.0, + "step": 17001 + }, + { + "epoch": 1.86712057983747, + "grad_norm": 2.3500161170959473, + "learning_rate": 1e-06, + "loss": 0.7817, + "mean_token_accuracy": 0.7533066272735596, + "num_tokens": 433796638.0, + "step": 17002 + }, + { + "epoch": 1.8672303975400835, + "grad_norm": 2.220780849456787, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7386542558670044, + "num_tokens": 433821363.0, + "step": 17003 + }, + { + "epoch": 1.867340215242697, + "grad_norm": 2.191297769546509, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7164111137390137, + "num_tokens": 433846499.0, + "step": 17004 + }, + { + "epoch": 1.8674500329453108, + "grad_norm": 2.2051188945770264, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7375778555870056, + "num_tokens": 433870934.0, + "step": 17005 + }, + { + "epoch": 1.8675598506479245, + "grad_norm": 2.228687047958374, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7158156633377075, + "num_tokens": 433894791.0, + "step": 17006 + }, + { + "epoch": 1.8676696683505383, + "grad_norm": 2.4096014499664307, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7197821140289307, + "num_tokens": 433919152.0, + "step": 17007 + }, + { + "epoch": 1.8677794860531518, + "grad_norm": 2.415397882461548, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7226988077163696, + "num_tokens": 433941956.0, + "step": 17008 + }, + { + "epoch": 1.8678893037557653, + "grad_norm": 2.566117763519287, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7174068689346313, + "num_tokens": 433962216.0, + "step": 17009 + }, + { + "epoch": 1.867999121458379, + "grad_norm": 2.3414599895477295, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7150008082389832, + "num_tokens": 433986275.0, + "step": 17010 + }, + { + "epoch": 1.8681089391609929, + "grad_norm": 2.3544647693634033, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.73085618019104, + "num_tokens": 434010177.0, + "step": 17011 + }, + { + "epoch": 1.8682187568636064, + "grad_norm": 2.240708827972412, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7266268730163574, + "num_tokens": 434036445.0, + "step": 17012 + }, + { + "epoch": 1.86832857456622, + "grad_norm": 2.158433675765991, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7195360064506531, + "num_tokens": 434064149.0, + "step": 17013 + }, + { + "epoch": 1.8684383922688337, + "grad_norm": 2.395434856414795, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.717751145362854, + "num_tokens": 434088896.0, + "step": 17014 + }, + { + "epoch": 1.8685482099714474, + "grad_norm": 2.387127161026001, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7116434574127197, + "num_tokens": 434111769.0, + "step": 17015 + }, + { + "epoch": 1.8686580276740612, + "grad_norm": 2.2449474334716797, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7076590061187744, + "num_tokens": 434137994.0, + "step": 17016 + }, + { + "epoch": 1.8687678453766747, + "grad_norm": 2.001737117767334, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.71134352684021, + "num_tokens": 434168378.0, + "step": 17017 + }, + { + "epoch": 1.8688776630792883, + "grad_norm": 2.092294216156006, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7216726541519165, + "num_tokens": 434196657.0, + "step": 17018 + }, + { + "epoch": 1.868987480781902, + "grad_norm": 1.9530788660049438, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7031221389770508, + "num_tokens": 434230106.0, + "step": 17019 + }, + { + "epoch": 1.8690972984845158, + "grad_norm": 2.3701815605163574, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7533739805221558, + "num_tokens": 434252571.0, + "step": 17020 + }, + { + "epoch": 1.8692071161871293, + "grad_norm": 2.176018476486206, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7226378321647644, + "num_tokens": 434278049.0, + "step": 17021 + }, + { + "epoch": 1.869316933889743, + "grad_norm": 2.086827516555786, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7253013849258423, + "num_tokens": 434303727.0, + "step": 17022 + }, + { + "epoch": 1.8694267515923566, + "grad_norm": 2.1612088680267334, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7298905849456787, + "num_tokens": 434329929.0, + "step": 17023 + }, + { + "epoch": 1.8695365692949704, + "grad_norm": 2.2131593227386475, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.716873049736023, + "num_tokens": 434356373.0, + "step": 17024 + }, + { + "epoch": 1.8696463869975841, + "grad_norm": 2.4176807403564453, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.7402077913284302, + "num_tokens": 434379452.0, + "step": 17025 + }, + { + "epoch": 1.8697562047001977, + "grad_norm": 2.2701408863067627, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7363214492797852, + "num_tokens": 434402413.0, + "step": 17026 + }, + { + "epoch": 1.8698660224028112, + "grad_norm": 2.306650161743164, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.699954092502594, + "num_tokens": 434427875.0, + "step": 17027 + }, + { + "epoch": 1.869975840105425, + "grad_norm": 2.388153076171875, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7280813455581665, + "num_tokens": 434450070.0, + "step": 17028 + }, + { + "epoch": 1.8700856578080387, + "grad_norm": 2.2596633434295654, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.700606107711792, + "num_tokens": 434476143.0, + "step": 17029 + }, + { + "epoch": 1.8701954755106525, + "grad_norm": 2.2562453746795654, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.718464732170105, + "num_tokens": 434503372.0, + "step": 17030 + }, + { + "epoch": 1.870305293213266, + "grad_norm": 2.951237201690674, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.7512754797935486, + "num_tokens": 434518617.0, + "step": 17031 + }, + { + "epoch": 1.8704151109158795, + "grad_norm": 2.1916112899780273, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7280101776123047, + "num_tokens": 434543445.0, + "step": 17032 + }, + { + "epoch": 1.8705249286184933, + "grad_norm": 2.450019598007202, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7310073375701904, + "num_tokens": 434565733.0, + "step": 17033 + }, + { + "epoch": 1.870634746321107, + "grad_norm": 2.0785088539123535, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7225531935691833, + "num_tokens": 434594870.0, + "step": 17034 + }, + { + "epoch": 1.8707445640237206, + "grad_norm": 2.2855706214904785, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7212474942207336, + "num_tokens": 434620040.0, + "step": 17035 + }, + { + "epoch": 1.8708543817263343, + "grad_norm": 2.1809985637664795, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7059453129768372, + "num_tokens": 434647846.0, + "step": 17036 + }, + { + "epoch": 1.8709641994289479, + "grad_norm": 2.153357982635498, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6958452463150024, + "num_tokens": 434674902.0, + "step": 17037 + }, + { + "epoch": 1.8710740171315616, + "grad_norm": 2.435774803161621, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.7402621507644653, + "num_tokens": 434697483.0, + "step": 17038 + }, + { + "epoch": 1.8711838348341754, + "grad_norm": 2.163025379180908, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7251816391944885, + "num_tokens": 434722926.0, + "step": 17039 + }, + { + "epoch": 1.871293652536789, + "grad_norm": 2.2336647510528564, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7012671232223511, + "num_tokens": 434749065.0, + "step": 17040 + }, + { + "epoch": 1.8714034702394025, + "grad_norm": 2.4130802154541016, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7328523993492126, + "num_tokens": 434771340.0, + "step": 17041 + }, + { + "epoch": 1.8715132879420162, + "grad_norm": 2.2396280765533447, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7363574504852295, + "num_tokens": 434796254.0, + "step": 17042 + }, + { + "epoch": 1.87162310564463, + "grad_norm": 2.159048080444336, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7405658960342407, + "num_tokens": 434821845.0, + "step": 17043 + }, + { + "epoch": 1.8717329233472437, + "grad_norm": 2.1383633613586426, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7164487838745117, + "num_tokens": 434847771.0, + "step": 17044 + }, + { + "epoch": 1.8718427410498573, + "grad_norm": 2.3362350463867188, + "learning_rate": 1e-06, + "loss": 0.7698, + "mean_token_accuracy": 0.7552659511566162, + "num_tokens": 434870971.0, + "step": 17045 + }, + { + "epoch": 1.8719525587524708, + "grad_norm": 2.0555248260498047, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7036377191543579, + "num_tokens": 434905260.0, + "step": 17046 + }, + { + "epoch": 1.8720623764550846, + "grad_norm": 2.3489110469818115, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7227192521095276, + "num_tokens": 434926825.0, + "step": 17047 + }, + { + "epoch": 1.8721721941576983, + "grad_norm": 2.190213203430176, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7262013554573059, + "num_tokens": 434952275.0, + "step": 17048 + }, + { + "epoch": 1.8722820118603118, + "grad_norm": 2.2632641792297363, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7012724280357361, + "num_tokens": 434978620.0, + "step": 17049 + }, + { + "epoch": 1.8723918295629254, + "grad_norm": 2.3077805042266846, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7232100963592529, + "num_tokens": 435004728.0, + "step": 17050 + }, + { + "epoch": 1.8725016472655391, + "grad_norm": 2.2869668006896973, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7295515537261963, + "num_tokens": 435029876.0, + "step": 17051 + }, + { + "epoch": 1.872611464968153, + "grad_norm": 2.160808563232422, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7008704543113708, + "num_tokens": 435057914.0, + "step": 17052 + }, + { + "epoch": 1.8727212826707667, + "grad_norm": 2.3988022804260254, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7249699234962463, + "num_tokens": 435083661.0, + "step": 17053 + }, + { + "epoch": 1.8728311003733802, + "grad_norm": 2.1761858463287354, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7036019563674927, + "num_tokens": 435111331.0, + "step": 17054 + }, + { + "epoch": 1.8729409180759937, + "grad_norm": 2.9418563842773438, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7067781686782837, + "num_tokens": 435139172.0, + "step": 17055 + }, + { + "epoch": 1.8730507357786075, + "grad_norm": 2.289306163787842, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7194771766662598, + "num_tokens": 435163240.0, + "step": 17056 + }, + { + "epoch": 1.8731605534812212, + "grad_norm": 2.1622154712677, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.720977246761322, + "num_tokens": 435191436.0, + "step": 17057 + }, + { + "epoch": 1.873270371183835, + "grad_norm": 2.8131794929504395, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7480947971343994, + "num_tokens": 435208724.0, + "step": 17058 + }, + { + "epoch": 1.8733801888864485, + "grad_norm": 2.6712517738342285, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7327340841293335, + "num_tokens": 435227975.0, + "step": 17059 + }, + { + "epoch": 1.873490006589062, + "grad_norm": 2.0067286491394043, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7100058197975159, + "num_tokens": 435260023.0, + "step": 17060 + }, + { + "epoch": 1.8735998242916758, + "grad_norm": 1.9915910959243774, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.6875677704811096, + "num_tokens": 435294943.0, + "step": 17061 + }, + { + "epoch": 1.8737096419942896, + "grad_norm": 2.387902021408081, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7361951470375061, + "num_tokens": 435316858.0, + "step": 17062 + }, + { + "epoch": 1.8738194596969031, + "grad_norm": 2.3376104831695557, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7136578559875488, + "num_tokens": 435341637.0, + "step": 17063 + }, + { + "epoch": 1.8739292773995166, + "grad_norm": 2.1034140586853027, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7068692445755005, + "num_tokens": 435371260.0, + "step": 17064 + }, + { + "epoch": 1.8740390951021304, + "grad_norm": 2.193143367767334, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7468865513801575, + "num_tokens": 435396382.0, + "step": 17065 + }, + { + "epoch": 1.8741489128047442, + "grad_norm": 2.519568920135498, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7345136404037476, + "num_tokens": 435416703.0, + "step": 17066 + }, + { + "epoch": 1.874258730507358, + "grad_norm": 2.167438268661499, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.6965929269790649, + "num_tokens": 435443303.0, + "step": 17067 + }, + { + "epoch": 1.8743685482099715, + "grad_norm": 2.1113948822021484, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7274113893508911, + "num_tokens": 435472602.0, + "step": 17068 + }, + { + "epoch": 1.874478365912585, + "grad_norm": 2.157905340194702, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7297685146331787, + "num_tokens": 435498882.0, + "step": 17069 + }, + { + "epoch": 1.8745881836151987, + "grad_norm": 2.1607823371887207, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7303464412689209, + "num_tokens": 435524945.0, + "step": 17070 + }, + { + "epoch": 1.8746980013178125, + "grad_norm": 2.322279691696167, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7212795615196228, + "num_tokens": 435551159.0, + "step": 17071 + }, + { + "epoch": 1.8748078190204263, + "grad_norm": 2.06382417678833, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.707222580909729, + "num_tokens": 435582271.0, + "step": 17072 + }, + { + "epoch": 1.8749176367230398, + "grad_norm": 2.2392163276672363, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7208579778671265, + "num_tokens": 435607271.0, + "step": 17073 + }, + { + "epoch": 1.8750274544256533, + "grad_norm": 2.8238413333892822, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7161698341369629, + "num_tokens": 435626156.0, + "step": 17074 + }, + { + "epoch": 1.875137272128267, + "grad_norm": 2.3500401973724365, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7276104688644409, + "num_tokens": 435649211.0, + "step": 17075 + }, + { + "epoch": 1.8752470898308808, + "grad_norm": 2.248176097869873, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7151191830635071, + "num_tokens": 435674262.0, + "step": 17076 + }, + { + "epoch": 1.8753569075334944, + "grad_norm": 2.281440258026123, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7186987996101379, + "num_tokens": 435699127.0, + "step": 17077 + }, + { + "epoch": 1.875466725236108, + "grad_norm": 2.287081241607666, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7143547534942627, + "num_tokens": 435723860.0, + "step": 17078 + }, + { + "epoch": 1.8755765429387217, + "grad_norm": 2.0378448963165283, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7295308113098145, + "num_tokens": 435755305.0, + "step": 17079 + }, + { + "epoch": 1.8756863606413354, + "grad_norm": 2.405824661254883, + "learning_rate": 1e-06, + "loss": 0.8287, + "mean_token_accuracy": 0.7409191131591797, + "num_tokens": 435779939.0, + "step": 17080 + }, + { + "epoch": 1.8757961783439492, + "grad_norm": 2.418621063232422, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7272188067436218, + "num_tokens": 435806653.0, + "step": 17081 + }, + { + "epoch": 1.8759059960465627, + "grad_norm": 2.230955123901367, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7278280258178711, + "num_tokens": 435834265.0, + "step": 17082 + }, + { + "epoch": 1.8760158137491763, + "grad_norm": 2.575237989425659, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7243449091911316, + "num_tokens": 435854977.0, + "step": 17083 + }, + { + "epoch": 1.87612563145179, + "grad_norm": 2.1761043071746826, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7266430854797363, + "num_tokens": 435880205.0, + "step": 17084 + }, + { + "epoch": 1.8762354491544038, + "grad_norm": 2.1582589149475098, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7291895151138306, + "num_tokens": 435907523.0, + "step": 17085 + }, + { + "epoch": 1.8763452668570173, + "grad_norm": 2.0460920333862305, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7107380628585815, + "num_tokens": 435939408.0, + "step": 17086 + }, + { + "epoch": 1.876455084559631, + "grad_norm": 2.3165876865386963, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7281252145767212, + "num_tokens": 435963413.0, + "step": 17087 + }, + { + "epoch": 1.8765649022622446, + "grad_norm": 2.462347984313965, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7247324585914612, + "num_tokens": 435984115.0, + "step": 17088 + }, + { + "epoch": 1.8766747199648584, + "grad_norm": 2.1530120372772217, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7156022787094116, + "num_tokens": 436011505.0, + "step": 17089 + }, + { + "epoch": 1.876784537667472, + "grad_norm": 2.206465005874634, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7167379856109619, + "num_tokens": 436036152.0, + "step": 17090 + }, + { + "epoch": 1.8768943553700856, + "grad_norm": 2.2810311317443848, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7436518669128418, + "num_tokens": 436059169.0, + "step": 17091 + }, + { + "epoch": 1.8770041730726992, + "grad_norm": 2.154585361480713, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7173147201538086, + "num_tokens": 436086323.0, + "step": 17092 + }, + { + "epoch": 1.877113990775313, + "grad_norm": 2.1399102210998535, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.690910279750824, + "num_tokens": 436113402.0, + "step": 17093 + }, + { + "epoch": 1.8772238084779267, + "grad_norm": 2.664418935775757, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7210132479667664, + "num_tokens": 436134776.0, + "step": 17094 + }, + { + "epoch": 1.8773336261805404, + "grad_norm": 2.1084043979644775, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7433561086654663, + "num_tokens": 436161239.0, + "step": 17095 + }, + { + "epoch": 1.877443443883154, + "grad_norm": 2.211860179901123, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.726548433303833, + "num_tokens": 436186497.0, + "step": 17096 + }, + { + "epoch": 1.8775532615857675, + "grad_norm": 2.0899598598480225, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7297763228416443, + "num_tokens": 436213183.0, + "step": 17097 + }, + { + "epoch": 1.8776630792883813, + "grad_norm": 2.350954055786133, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7382843494415283, + "num_tokens": 436235080.0, + "step": 17098 + }, + { + "epoch": 1.877772896990995, + "grad_norm": 2.2256906032562256, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7311424016952515, + "num_tokens": 436259559.0, + "step": 17099 + }, + { + "epoch": 1.8778827146936086, + "grad_norm": 1.9114651679992676, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7069822549819946, + "num_tokens": 436294082.0, + "step": 17100 + }, + { + "epoch": 1.8779925323962223, + "grad_norm": 2.3991971015930176, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7087007761001587, + "num_tokens": 436318915.0, + "step": 17101 + }, + { + "epoch": 1.8781023500988359, + "grad_norm": 2.287911891937256, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7154449224472046, + "num_tokens": 436344073.0, + "step": 17102 + }, + { + "epoch": 1.8782121678014496, + "grad_norm": 2.249387264251709, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.6963294744491577, + "num_tokens": 436370393.0, + "step": 17103 + }, + { + "epoch": 1.8783219855040634, + "grad_norm": 2.5805792808532715, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7453525066375732, + "num_tokens": 436389960.0, + "step": 17104 + }, + { + "epoch": 1.878431803206677, + "grad_norm": 2.1356256008148193, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7058320641517639, + "num_tokens": 436418388.0, + "step": 17105 + }, + { + "epoch": 1.8785416209092904, + "grad_norm": 1.910971999168396, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7219635844230652, + "num_tokens": 436452689.0, + "step": 17106 + }, + { + "epoch": 1.8786514386119042, + "grad_norm": 2.438918352127075, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7225950956344604, + "num_tokens": 436477202.0, + "step": 17107 + }, + { + "epoch": 1.878761256314518, + "grad_norm": 2.5153205394744873, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7291134595870972, + "num_tokens": 436496787.0, + "step": 17108 + }, + { + "epoch": 1.8788710740171317, + "grad_norm": 2.1242079734802246, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7219383716583252, + "num_tokens": 436524466.0, + "step": 17109 + }, + { + "epoch": 1.8789808917197452, + "grad_norm": 1.89910089969635, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7131043672561646, + "num_tokens": 436558112.0, + "step": 17110 + }, + { + "epoch": 1.8790907094223588, + "grad_norm": 2.0245862007141113, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7047638297080994, + "num_tokens": 436587078.0, + "step": 17111 + }, + { + "epoch": 1.8792005271249725, + "grad_norm": 2.359683036804199, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7293590307235718, + "num_tokens": 436611112.0, + "step": 17112 + }, + { + "epoch": 1.8793103448275863, + "grad_norm": 2.292466640472412, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7328634858131409, + "num_tokens": 436635680.0, + "step": 17113 + }, + { + "epoch": 1.8794201625301998, + "grad_norm": 2.2907185554504395, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7007555961608887, + "num_tokens": 436662112.0, + "step": 17114 + }, + { + "epoch": 1.8795299802328134, + "grad_norm": 2.568315267562866, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7238974571228027, + "num_tokens": 436681956.0, + "step": 17115 + }, + { + "epoch": 1.8796397979354271, + "grad_norm": 2.082131862640381, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7227886915206909, + "num_tokens": 436710122.0, + "step": 17116 + }, + { + "epoch": 1.8797496156380409, + "grad_norm": 2.421419143676758, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7108799815177917, + "num_tokens": 436732404.0, + "step": 17117 + }, + { + "epoch": 1.8798594333406546, + "grad_norm": 2.3979978561401367, + "learning_rate": 1e-06, + "loss": 0.8057, + "mean_token_accuracy": 0.7505478858947754, + "num_tokens": 436753714.0, + "step": 17118 + }, + { + "epoch": 1.8799692510432682, + "grad_norm": 2.2458367347717285, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7040574550628662, + "num_tokens": 436780453.0, + "step": 17119 + }, + { + "epoch": 1.8800790687458817, + "grad_norm": 2.405611753463745, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7239127159118652, + "num_tokens": 436803085.0, + "step": 17120 + }, + { + "epoch": 1.8801888864484955, + "grad_norm": 2.579559564590454, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7439734935760498, + "num_tokens": 436821609.0, + "step": 17121 + }, + { + "epoch": 1.8802987041511092, + "grad_norm": 1.9758039712905884, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7276434302330017, + "num_tokens": 436851046.0, + "step": 17122 + }, + { + "epoch": 1.880408521853723, + "grad_norm": 2.2397937774658203, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7113522291183472, + "num_tokens": 436877779.0, + "step": 17123 + }, + { + "epoch": 1.8805183395563365, + "grad_norm": 2.2025935649871826, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7135164737701416, + "num_tokens": 436903474.0, + "step": 17124 + }, + { + "epoch": 1.88062815725895, + "grad_norm": 2.070122241973877, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7079445123672485, + "num_tokens": 436932768.0, + "step": 17125 + }, + { + "epoch": 1.8807379749615638, + "grad_norm": 1.966577172279358, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7158206701278687, + "num_tokens": 436963804.0, + "step": 17126 + }, + { + "epoch": 1.8808477926641776, + "grad_norm": 2.3046553134918213, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7337787747383118, + "num_tokens": 436987883.0, + "step": 17127 + }, + { + "epoch": 1.880957610366791, + "grad_norm": 2.05779767036438, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7182778120040894, + "num_tokens": 437019842.0, + "step": 17128 + }, + { + "epoch": 1.8810674280694046, + "grad_norm": 2.282470464706421, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7346363067626953, + "num_tokens": 437045087.0, + "step": 17129 + }, + { + "epoch": 1.8811772457720184, + "grad_norm": 2.3698155879974365, + "learning_rate": 1e-06, + "loss": 0.8085, + "mean_token_accuracy": 0.7402086853981018, + "num_tokens": 437066069.0, + "step": 17130 + }, + { + "epoch": 1.8812870634746321, + "grad_norm": 2.288623332977295, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7191131114959717, + "num_tokens": 437092363.0, + "step": 17131 + }, + { + "epoch": 1.881396881177246, + "grad_norm": 2.3195884227752686, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7161470651626587, + "num_tokens": 437116811.0, + "step": 17132 + }, + { + "epoch": 1.8815066988798594, + "grad_norm": 2.1097240447998047, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7456702589988708, + "num_tokens": 437143321.0, + "step": 17133 + }, + { + "epoch": 1.881616516582473, + "grad_norm": 2.123507499694824, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7141894698143005, + "num_tokens": 437171153.0, + "step": 17134 + }, + { + "epoch": 1.8817263342850867, + "grad_norm": 2.3038361072540283, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7368386387825012, + "num_tokens": 437195180.0, + "step": 17135 + }, + { + "epoch": 1.8818361519877005, + "grad_norm": 2.4684627056121826, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7400836944580078, + "num_tokens": 437216334.0, + "step": 17136 + }, + { + "epoch": 1.881945969690314, + "grad_norm": 2.4051902294158936, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7248055934906006, + "num_tokens": 437240442.0, + "step": 17137 + }, + { + "epoch": 1.8820557873929278, + "grad_norm": 2.5375168323516846, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.7483378648757935, + "num_tokens": 437260475.0, + "step": 17138 + }, + { + "epoch": 1.8821656050955413, + "grad_norm": 2.3014707565307617, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7427772879600525, + "num_tokens": 437283332.0, + "step": 17139 + }, + { + "epoch": 1.882275422798155, + "grad_norm": 2.370447874069214, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.7003858089447021, + "num_tokens": 437309161.0, + "step": 17140 + }, + { + "epoch": 1.8823852405007688, + "grad_norm": 2.355178117752075, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7273184657096863, + "num_tokens": 437333894.0, + "step": 17141 + }, + { + "epoch": 1.8824950582033824, + "grad_norm": 1.9030025005340576, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7408241629600525, + "num_tokens": 437365317.0, + "step": 17142 + }, + { + "epoch": 1.882604875905996, + "grad_norm": 2.2407901287078857, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7228759527206421, + "num_tokens": 437390884.0, + "step": 17143 + }, + { + "epoch": 1.8827146936086097, + "grad_norm": 2.3731322288513184, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7257436513900757, + "num_tokens": 437415846.0, + "step": 17144 + }, + { + "epoch": 1.8828245113112234, + "grad_norm": 2.2412831783294678, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7196671962738037, + "num_tokens": 437441374.0, + "step": 17145 + }, + { + "epoch": 1.8829343290138372, + "grad_norm": 2.109938859939575, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7347343564033508, + "num_tokens": 437469371.0, + "step": 17146 + }, + { + "epoch": 1.8830441467164507, + "grad_norm": 2.354923963546753, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7045043706893921, + "num_tokens": 437494951.0, + "step": 17147 + }, + { + "epoch": 1.8831539644190642, + "grad_norm": 2.0473058223724365, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.706620991230011, + "num_tokens": 437525475.0, + "step": 17148 + }, + { + "epoch": 1.883263782121678, + "grad_norm": 2.2091128826141357, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7377587556838989, + "num_tokens": 437548419.0, + "step": 17149 + }, + { + "epoch": 1.8833735998242918, + "grad_norm": 2.1238324642181396, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7156602144241333, + "num_tokens": 437576100.0, + "step": 17150 + }, + { + "epoch": 1.8834834175269053, + "grad_norm": 2.296015977859497, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.72873455286026, + "num_tokens": 437600123.0, + "step": 17151 + }, + { + "epoch": 1.883593235229519, + "grad_norm": 2.24847149848938, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7413926124572754, + "num_tokens": 437624189.0, + "step": 17152 + }, + { + "epoch": 1.8837030529321326, + "grad_norm": 2.5381252765655518, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.723601222038269, + "num_tokens": 437644811.0, + "step": 17153 + }, + { + "epoch": 1.8838128706347463, + "grad_norm": 2.278653621673584, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7280162572860718, + "num_tokens": 437668203.0, + "step": 17154 + }, + { + "epoch": 1.88392268833736, + "grad_norm": 2.2390518188476562, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7249107360839844, + "num_tokens": 437692694.0, + "step": 17155 + }, + { + "epoch": 1.8840325060399736, + "grad_norm": 2.288723945617676, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7365591526031494, + "num_tokens": 437715004.0, + "step": 17156 + }, + { + "epoch": 1.8841423237425872, + "grad_norm": 2.313103437423706, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7116032838821411, + "num_tokens": 437739351.0, + "step": 17157 + }, + { + "epoch": 1.884252141445201, + "grad_norm": 2.2084662914276123, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7253717184066772, + "num_tokens": 437764677.0, + "step": 17158 + }, + { + "epoch": 1.8843619591478147, + "grad_norm": 2.640167713165283, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7373429536819458, + "num_tokens": 437784430.0, + "step": 17159 + }, + { + "epoch": 1.8844717768504284, + "grad_norm": 2.1888504028320312, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.734673023223877, + "num_tokens": 437810499.0, + "step": 17160 + }, + { + "epoch": 1.884581594553042, + "grad_norm": 2.474215269088745, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7204186916351318, + "num_tokens": 437831548.0, + "step": 17161 + }, + { + "epoch": 1.8846914122556555, + "grad_norm": 2.321850299835205, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7329494953155518, + "num_tokens": 437853871.0, + "step": 17162 + }, + { + "epoch": 1.8848012299582693, + "grad_norm": 2.174067497253418, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7372989058494568, + "num_tokens": 437880154.0, + "step": 17163 + }, + { + "epoch": 1.884911047660883, + "grad_norm": 2.1096208095550537, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7230006456375122, + "num_tokens": 437908093.0, + "step": 17164 + }, + { + "epoch": 1.8850208653634966, + "grad_norm": 2.030911445617676, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7090228796005249, + "num_tokens": 437939904.0, + "step": 17165 + }, + { + "epoch": 1.88513068306611, + "grad_norm": 2.174786329269409, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7165971994400024, + "num_tokens": 437966605.0, + "step": 17166 + }, + { + "epoch": 1.8852405007687238, + "grad_norm": 2.326751470565796, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7244662046432495, + "num_tokens": 437991000.0, + "step": 17167 + }, + { + "epoch": 1.8853503184713376, + "grad_norm": 2.2547383308410645, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7378963232040405, + "num_tokens": 438015413.0, + "step": 17168 + }, + { + "epoch": 1.8854601361739514, + "grad_norm": 2.2149734497070312, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7286432981491089, + "num_tokens": 438041330.0, + "step": 17169 + }, + { + "epoch": 1.885569953876565, + "grad_norm": 2.0266995429992676, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7110567092895508, + "num_tokens": 438069773.0, + "step": 17170 + }, + { + "epoch": 1.8856797715791784, + "grad_norm": 2.065770149230957, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7109056115150452, + "num_tokens": 438100117.0, + "step": 17171 + }, + { + "epoch": 1.8857895892817922, + "grad_norm": 2.3742618560791016, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7343059778213501, + "num_tokens": 438122490.0, + "step": 17172 + }, + { + "epoch": 1.885899406984406, + "grad_norm": 2.1347384452819824, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7028248310089111, + "num_tokens": 438148480.0, + "step": 17173 + }, + { + "epoch": 1.8860092246870197, + "grad_norm": 2.67972469329834, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7239825129508972, + "num_tokens": 438169242.0, + "step": 17174 + }, + { + "epoch": 1.8861190423896332, + "grad_norm": 2.069751024246216, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6896817684173584, + "num_tokens": 438199746.0, + "step": 17175 + }, + { + "epoch": 1.8862288600922468, + "grad_norm": 2.1973414421081543, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7070733308792114, + "num_tokens": 438226475.0, + "step": 17176 + }, + { + "epoch": 1.8863386777948605, + "grad_norm": 2.126230239868164, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7136572003364563, + "num_tokens": 438253134.0, + "step": 17177 + }, + { + "epoch": 1.8864484954974743, + "grad_norm": 2.215054512023926, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7137372493743896, + "num_tokens": 438279438.0, + "step": 17178 + }, + { + "epoch": 1.8865583132000878, + "grad_norm": 2.16017484664917, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7239981889724731, + "num_tokens": 438305863.0, + "step": 17179 + }, + { + "epoch": 1.8866681309027014, + "grad_norm": 1.8944711685180664, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7249757051467896, + "num_tokens": 438337719.0, + "step": 17180 + }, + { + "epoch": 1.8867779486053151, + "grad_norm": 2.0629544258117676, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6909999847412109, + "num_tokens": 438370151.0, + "step": 17181 + }, + { + "epoch": 1.8868877663079289, + "grad_norm": 1.9748961925506592, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7016536593437195, + "num_tokens": 438403063.0, + "step": 17182 + }, + { + "epoch": 1.8869975840105426, + "grad_norm": 2.568671226501465, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.7309920191764832, + "num_tokens": 438423290.0, + "step": 17183 + }, + { + "epoch": 1.8871074017131562, + "grad_norm": 2.156304121017456, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7408414483070374, + "num_tokens": 438449884.0, + "step": 17184 + }, + { + "epoch": 1.8872172194157697, + "grad_norm": 2.263322114944458, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6955276131629944, + "num_tokens": 438476807.0, + "step": 17185 + }, + { + "epoch": 1.8873270371183835, + "grad_norm": 2.5966153144836426, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7285900115966797, + "num_tokens": 438496891.0, + "step": 17186 + }, + { + "epoch": 1.8874368548209972, + "grad_norm": 2.2412662506103516, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7190127372741699, + "num_tokens": 438522528.0, + "step": 17187 + }, + { + "epoch": 1.887546672523611, + "grad_norm": 2.1036200523376465, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7304874062538147, + "num_tokens": 438549535.0, + "step": 17188 + }, + { + "epoch": 1.8876564902262245, + "grad_norm": 2.485189199447632, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7369186878204346, + "num_tokens": 438571104.0, + "step": 17189 + }, + { + "epoch": 1.887766307928838, + "grad_norm": 2.393731117248535, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6883437633514404, + "num_tokens": 438595778.0, + "step": 17190 + }, + { + "epoch": 1.8878761256314518, + "grad_norm": 2.1836910247802734, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7306478023529053, + "num_tokens": 438622452.0, + "step": 17191 + }, + { + "epoch": 1.8879859433340656, + "grad_norm": 1.9693020582199097, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7155486941337585, + "num_tokens": 438654916.0, + "step": 17192 + }, + { + "epoch": 1.888095761036679, + "grad_norm": 2.2874057292938232, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7325981259346008, + "num_tokens": 438679822.0, + "step": 17193 + }, + { + "epoch": 1.8882055787392926, + "grad_norm": 2.1351840496063232, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7198938131332397, + "num_tokens": 438707496.0, + "step": 17194 + }, + { + "epoch": 1.8883153964419064, + "grad_norm": 2.257467746734619, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.7116768956184387, + "num_tokens": 438732143.0, + "step": 17195 + }, + { + "epoch": 1.8884252141445201, + "grad_norm": 2.129988193511963, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7282666563987732, + "num_tokens": 438758493.0, + "step": 17196 + }, + { + "epoch": 1.888535031847134, + "grad_norm": 2.438225507736206, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7331311106681824, + "num_tokens": 438778576.0, + "step": 17197 + }, + { + "epoch": 1.8886448495497474, + "grad_norm": 2.317061424255371, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7101020216941833, + "num_tokens": 438803526.0, + "step": 17198 + }, + { + "epoch": 1.888754667252361, + "grad_norm": 2.2881879806518555, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.6904606819152832, + "num_tokens": 438832978.0, + "step": 17199 + }, + { + "epoch": 1.8888644849549747, + "grad_norm": 2.089550018310547, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.717278003692627, + "num_tokens": 438861299.0, + "step": 17200 + }, + { + "epoch": 1.8889743026575885, + "grad_norm": 2.4071550369262695, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7102440595626831, + "num_tokens": 438885011.0, + "step": 17201 + }, + { + "epoch": 1.889084120360202, + "grad_norm": 2.421098232269287, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.741753876209259, + "num_tokens": 438905956.0, + "step": 17202 + }, + { + "epoch": 1.8891939380628158, + "grad_norm": 2.1474249362945557, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6918508410453796, + "num_tokens": 438935235.0, + "step": 17203 + }, + { + "epoch": 1.8893037557654293, + "grad_norm": 2.339423179626465, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7388467788696289, + "num_tokens": 438958265.0, + "step": 17204 + }, + { + "epoch": 1.889413573468043, + "grad_norm": 2.0655715465545654, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.716338574886322, + "num_tokens": 438989313.0, + "step": 17205 + }, + { + "epoch": 1.8895233911706568, + "grad_norm": 2.5029969215393066, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7181059718132019, + "num_tokens": 439011430.0, + "step": 17206 + }, + { + "epoch": 1.8896332088732704, + "grad_norm": 2.159608840942383, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7418575286865234, + "num_tokens": 439036917.0, + "step": 17207 + }, + { + "epoch": 1.8897430265758839, + "grad_norm": 2.19865083694458, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7291656732559204, + "num_tokens": 439062534.0, + "step": 17208 + }, + { + "epoch": 1.8898528442784976, + "grad_norm": 2.2734744548797607, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7420138120651245, + "num_tokens": 439087629.0, + "step": 17209 + }, + { + "epoch": 1.8899626619811114, + "grad_norm": 1.9305808544158936, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7083693742752075, + "num_tokens": 439119659.0, + "step": 17210 + }, + { + "epoch": 1.8900724796837252, + "grad_norm": 2.3789732456207275, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7250060439109802, + "num_tokens": 439144559.0, + "step": 17211 + }, + { + "epoch": 1.8901822973863387, + "grad_norm": 2.1614038944244385, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7336809039115906, + "num_tokens": 439170302.0, + "step": 17212 + }, + { + "epoch": 1.8902921150889522, + "grad_norm": 2.497753620147705, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.735562801361084, + "num_tokens": 439191616.0, + "step": 17213 + }, + { + "epoch": 1.890401932791566, + "grad_norm": 2.539294958114624, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7341997623443604, + "num_tokens": 439213437.0, + "step": 17214 + }, + { + "epoch": 1.8905117504941797, + "grad_norm": 2.338121175765991, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.717333972454071, + "num_tokens": 439236257.0, + "step": 17215 + }, + { + "epoch": 1.8906215681967933, + "grad_norm": 2.3580853939056396, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.738211452960968, + "num_tokens": 439258658.0, + "step": 17216 + }, + { + "epoch": 1.890731385899407, + "grad_norm": 2.3433971405029297, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7224829792976379, + "num_tokens": 439284076.0, + "step": 17217 + }, + { + "epoch": 1.8908412036020206, + "grad_norm": 2.3750221729278564, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7294845581054688, + "num_tokens": 439305543.0, + "step": 17218 + }, + { + "epoch": 1.8909510213046343, + "grad_norm": 2.273790121078491, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7223355174064636, + "num_tokens": 439329160.0, + "step": 17219 + }, + { + "epoch": 1.891060839007248, + "grad_norm": 2.221247673034668, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.723873496055603, + "num_tokens": 439354061.0, + "step": 17220 + }, + { + "epoch": 1.8911706567098616, + "grad_norm": 2.1162679195404053, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7209795117378235, + "num_tokens": 439381394.0, + "step": 17221 + }, + { + "epoch": 1.8912804744124752, + "grad_norm": 2.2698237895965576, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7275534868240356, + "num_tokens": 439405582.0, + "step": 17222 + }, + { + "epoch": 1.891390292115089, + "grad_norm": 2.2933757305145264, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7239781022071838, + "num_tokens": 439430586.0, + "step": 17223 + }, + { + "epoch": 1.8915001098177027, + "grad_norm": 2.0477135181427, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.6991667747497559, + "num_tokens": 439459271.0, + "step": 17224 + }, + { + "epoch": 1.8916099275203164, + "grad_norm": 2.297272205352783, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7089841961860657, + "num_tokens": 439487696.0, + "step": 17225 + }, + { + "epoch": 1.89171974522293, + "grad_norm": 2.4206418991088867, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7160274386405945, + "num_tokens": 439510224.0, + "step": 17226 + }, + { + "epoch": 1.8918295629255435, + "grad_norm": 2.1287853717803955, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7086522579193115, + "num_tokens": 439537852.0, + "step": 17227 + }, + { + "epoch": 1.8919393806281573, + "grad_norm": 2.285670042037964, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7134431600570679, + "num_tokens": 439561013.0, + "step": 17228 + }, + { + "epoch": 1.892049198330771, + "grad_norm": 2.189781665802002, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7122575044631958, + "num_tokens": 439586361.0, + "step": 17229 + }, + { + "epoch": 1.8921590160333845, + "grad_norm": 2.4207546710968018, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7364896535873413, + "num_tokens": 439608870.0, + "step": 17230 + }, + { + "epoch": 1.892268833735998, + "grad_norm": 2.524773597717285, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7265428304672241, + "num_tokens": 439629941.0, + "step": 17231 + }, + { + "epoch": 1.8923786514386118, + "grad_norm": 2.049565076828003, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7241449356079102, + "num_tokens": 439659357.0, + "step": 17232 + }, + { + "epoch": 1.8924884691412256, + "grad_norm": 2.210293769836426, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7419887185096741, + "num_tokens": 439683179.0, + "step": 17233 + }, + { + "epoch": 1.8925982868438394, + "grad_norm": 2.0526580810546875, + "learning_rate": 1e-06, + "loss": 0.7905, + "mean_token_accuracy": 0.7481257319450378, + "num_tokens": 439709746.0, + "step": 17234 + }, + { + "epoch": 1.8927081045464529, + "grad_norm": 2.1580615043640137, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7233917713165283, + "num_tokens": 439739218.0, + "step": 17235 + }, + { + "epoch": 1.8928179222490664, + "grad_norm": 2.197148323059082, + "learning_rate": 1e-06, + "loss": 0.8075, + "mean_token_accuracy": 0.7401752471923828, + "num_tokens": 439763609.0, + "step": 17236 + }, + { + "epoch": 1.8929277399516802, + "grad_norm": 2.139294147491455, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7387566566467285, + "num_tokens": 439791913.0, + "step": 17237 + }, + { + "epoch": 1.893037557654294, + "grad_norm": 2.2395100593566895, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7130358815193176, + "num_tokens": 439817651.0, + "step": 17238 + }, + { + "epoch": 1.8931473753569077, + "grad_norm": 2.3182168006896973, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7265322804450989, + "num_tokens": 439842201.0, + "step": 17239 + }, + { + "epoch": 1.8932571930595212, + "grad_norm": 2.009474992752075, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7009630799293518, + "num_tokens": 439874016.0, + "step": 17240 + }, + { + "epoch": 1.8933670107621348, + "grad_norm": 2.2618942260742188, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.7421402931213379, + "num_tokens": 439896853.0, + "step": 17241 + }, + { + "epoch": 1.8934768284647485, + "grad_norm": 2.2190098762512207, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7064638137817383, + "num_tokens": 439925444.0, + "step": 17242 + }, + { + "epoch": 1.8935866461673623, + "grad_norm": 2.3828017711639404, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7415622472763062, + "num_tokens": 439949243.0, + "step": 17243 + }, + { + "epoch": 1.8936964638699758, + "grad_norm": 2.2045609951019287, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7280938625335693, + "num_tokens": 439973543.0, + "step": 17244 + }, + { + "epoch": 1.8938062815725893, + "grad_norm": 2.3698649406433105, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.715746283531189, + "num_tokens": 439997493.0, + "step": 17245 + }, + { + "epoch": 1.893916099275203, + "grad_norm": 2.417093276977539, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7341353893280029, + "num_tokens": 440020418.0, + "step": 17246 + }, + { + "epoch": 1.8940259169778169, + "grad_norm": 2.130310535430908, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.701915442943573, + "num_tokens": 440049514.0, + "step": 17247 + }, + { + "epoch": 1.8941357346804306, + "grad_norm": 2.2239129543304443, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7124741077423096, + "num_tokens": 440077204.0, + "step": 17248 + }, + { + "epoch": 1.8942455523830442, + "grad_norm": 2.3153374195098877, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.7366499900817871, + "num_tokens": 440101311.0, + "step": 17249 + }, + { + "epoch": 1.8943553700856577, + "grad_norm": 2.1206064224243164, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7204633951187134, + "num_tokens": 440126939.0, + "step": 17250 + }, + { + "epoch": 1.8944651877882714, + "grad_norm": 2.050861120223999, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7200984954833984, + "num_tokens": 440156798.0, + "step": 17251 + }, + { + "epoch": 1.8945750054908852, + "grad_norm": 2.1043202877044678, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.70250403881073, + "num_tokens": 440186152.0, + "step": 17252 + }, + { + "epoch": 1.894684823193499, + "grad_norm": 2.0555167198181152, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7207189202308655, + "num_tokens": 440216001.0, + "step": 17253 + }, + { + "epoch": 1.8947946408961125, + "grad_norm": 2.4012081623077393, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7684520483016968, + "num_tokens": 440236936.0, + "step": 17254 + }, + { + "epoch": 1.894904458598726, + "grad_norm": 2.39298415184021, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7217831611633301, + "num_tokens": 440259423.0, + "step": 17255 + }, + { + "epoch": 1.8950142763013398, + "grad_norm": 2.0541601181030273, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7243046164512634, + "num_tokens": 440289000.0, + "step": 17256 + }, + { + "epoch": 1.8951240940039535, + "grad_norm": 2.3238766193389893, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7177100777626038, + "num_tokens": 440312326.0, + "step": 17257 + }, + { + "epoch": 1.895233911706567, + "grad_norm": 2.169006586074829, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7230014801025391, + "num_tokens": 440338306.0, + "step": 17258 + }, + { + "epoch": 1.8953437294091806, + "grad_norm": 2.3990955352783203, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7241771221160889, + "num_tokens": 440360943.0, + "step": 17259 + }, + { + "epoch": 1.8954535471117944, + "grad_norm": 2.249370574951172, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7306057214736938, + "num_tokens": 440386168.0, + "step": 17260 + }, + { + "epoch": 1.8955633648144081, + "grad_norm": 2.153776168823242, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7089311480522156, + "num_tokens": 440414046.0, + "step": 17261 + }, + { + "epoch": 1.8956731825170219, + "grad_norm": 2.327254056930542, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7485607862472534, + "num_tokens": 440435848.0, + "step": 17262 + }, + { + "epoch": 1.8957830002196354, + "grad_norm": 2.2881224155426025, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.696864664554596, + "num_tokens": 440460508.0, + "step": 17263 + }, + { + "epoch": 1.895892817922249, + "grad_norm": 2.355820894241333, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7134819626808167, + "num_tokens": 440484818.0, + "step": 17264 + }, + { + "epoch": 1.8960026356248627, + "grad_norm": 2.1084015369415283, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7208338975906372, + "num_tokens": 440512545.0, + "step": 17265 + }, + { + "epoch": 1.8961124533274765, + "grad_norm": 2.0298688411712646, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7107568979263306, + "num_tokens": 440541453.0, + "step": 17266 + }, + { + "epoch": 1.89622227103009, + "grad_norm": 2.370553731918335, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7192875146865845, + "num_tokens": 440563998.0, + "step": 17267 + }, + { + "epoch": 1.8963320887327038, + "grad_norm": 2.205491065979004, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7225056290626526, + "num_tokens": 440589541.0, + "step": 17268 + }, + { + "epoch": 1.8964419064353173, + "grad_norm": 2.4024510383605957, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7067685127258301, + "num_tokens": 440614180.0, + "step": 17269 + }, + { + "epoch": 1.896551724137931, + "grad_norm": 2.1342153549194336, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7386008501052856, + "num_tokens": 440642868.0, + "step": 17270 + }, + { + "epoch": 1.8966615418405448, + "grad_norm": 2.371250867843628, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7005532383918762, + "num_tokens": 440667136.0, + "step": 17271 + }, + { + "epoch": 1.8967713595431583, + "grad_norm": 2.2786567211151123, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7317286133766174, + "num_tokens": 440691498.0, + "step": 17272 + }, + { + "epoch": 1.8968811772457719, + "grad_norm": 2.162618637084961, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7263268828392029, + "num_tokens": 440718013.0, + "step": 17273 + }, + { + "epoch": 1.8969909949483856, + "grad_norm": 2.2804510593414307, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7301144599914551, + "num_tokens": 440742059.0, + "step": 17274 + }, + { + "epoch": 1.8971008126509994, + "grad_norm": 2.5781354904174805, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7208492755889893, + "num_tokens": 440764066.0, + "step": 17275 + }, + { + "epoch": 1.8972106303536131, + "grad_norm": 2.460994005203247, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7158859372138977, + "num_tokens": 440786612.0, + "step": 17276 + }, + { + "epoch": 1.8973204480562267, + "grad_norm": 2.5908896923065186, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7131586074829102, + "num_tokens": 440807869.0, + "step": 17277 + }, + { + "epoch": 1.8974302657588402, + "grad_norm": 1.9875410795211792, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.6989678740501404, + "num_tokens": 440838320.0, + "step": 17278 + }, + { + "epoch": 1.897540083461454, + "grad_norm": 2.2631070613861084, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7341911196708679, + "num_tokens": 440863978.0, + "step": 17279 + }, + { + "epoch": 1.8976499011640677, + "grad_norm": 2.0858798027038574, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7442619800567627, + "num_tokens": 440890402.0, + "step": 17280 + }, + { + "epoch": 1.8977597188666813, + "grad_norm": 2.597223997116089, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7339362502098083, + "num_tokens": 440909906.0, + "step": 17281 + }, + { + "epoch": 1.897869536569295, + "grad_norm": 2.239536762237549, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7192128300666809, + "num_tokens": 440935807.0, + "step": 17282 + }, + { + "epoch": 1.8979793542719086, + "grad_norm": 2.3147778511047363, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7168410420417786, + "num_tokens": 440960393.0, + "step": 17283 + }, + { + "epoch": 1.8980891719745223, + "grad_norm": 2.3092246055603027, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.710573136806488, + "num_tokens": 440988654.0, + "step": 17284 + }, + { + "epoch": 1.898198989677136, + "grad_norm": 2.2731194496154785, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7160903215408325, + "num_tokens": 441013691.0, + "step": 17285 + }, + { + "epoch": 1.8983088073797496, + "grad_norm": 2.1210827827453613, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.72650146484375, + "num_tokens": 441041565.0, + "step": 17286 + }, + { + "epoch": 1.8984186250823631, + "grad_norm": 2.0929222106933594, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7331883907318115, + "num_tokens": 441070067.0, + "step": 17287 + }, + { + "epoch": 1.898528442784977, + "grad_norm": 2.3151400089263916, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7229571342468262, + "num_tokens": 441095796.0, + "step": 17288 + }, + { + "epoch": 1.8986382604875907, + "grad_norm": 2.0701284408569336, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7211673259735107, + "num_tokens": 441127058.0, + "step": 17289 + }, + { + "epoch": 1.8987480781902044, + "grad_norm": 2.149251937866211, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7066943049430847, + "num_tokens": 441157397.0, + "step": 17290 + }, + { + "epoch": 1.898857895892818, + "grad_norm": 2.455888271331787, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7415770888328552, + "num_tokens": 441179773.0, + "step": 17291 + }, + { + "epoch": 1.8989677135954315, + "grad_norm": 2.4925758838653564, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7365175485610962, + "num_tokens": 441200304.0, + "step": 17292 + }, + { + "epoch": 1.8990775312980452, + "grad_norm": 2.180208206176758, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7123100161552429, + "num_tokens": 441227982.0, + "step": 17293 + }, + { + "epoch": 1.899187349000659, + "grad_norm": 2.853004217147827, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.734512984752655, + "num_tokens": 441246309.0, + "step": 17294 + }, + { + "epoch": 1.8992971667032725, + "grad_norm": 2.060370922088623, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7158825397491455, + "num_tokens": 441277482.0, + "step": 17295 + }, + { + "epoch": 1.899406984405886, + "grad_norm": 2.488314151763916, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.72318434715271, + "num_tokens": 441299481.0, + "step": 17296 + }, + { + "epoch": 1.8995168021084998, + "grad_norm": 2.6745781898498535, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7439531087875366, + "num_tokens": 441317848.0, + "step": 17297 + }, + { + "epoch": 1.8996266198111136, + "grad_norm": 2.557182550430298, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7382776737213135, + "num_tokens": 441337326.0, + "step": 17298 + }, + { + "epoch": 1.8997364375137273, + "grad_norm": 2.8094515800476074, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7278491258621216, + "num_tokens": 441355782.0, + "step": 17299 + }, + { + "epoch": 1.8998462552163409, + "grad_norm": 2.085395574569702, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7022671699523926, + "num_tokens": 441385873.0, + "step": 17300 + }, + { + "epoch": 1.8999560729189544, + "grad_norm": 2.2512831687927246, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7198558449745178, + "num_tokens": 441410124.0, + "step": 17301 + }, + { + "epoch": 1.9000658906215682, + "grad_norm": 2.323395252227783, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7105595469474792, + "num_tokens": 441434340.0, + "step": 17302 + }, + { + "epoch": 1.900175708324182, + "grad_norm": 2.1024463176727295, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7226436734199524, + "num_tokens": 441463469.0, + "step": 17303 + }, + { + "epoch": 1.9002855260267957, + "grad_norm": 2.077936887741089, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7232880592346191, + "num_tokens": 441492181.0, + "step": 17304 + }, + { + "epoch": 1.9003953437294092, + "grad_norm": 2.223494291305542, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7244784832000732, + "num_tokens": 441517754.0, + "step": 17305 + }, + { + "epoch": 1.9005051614320227, + "grad_norm": 2.3564343452453613, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7101891040802002, + "num_tokens": 441541635.0, + "step": 17306 + }, + { + "epoch": 1.9006149791346365, + "grad_norm": 2.190063953399658, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.7485978603363037, + "num_tokens": 441564925.0, + "step": 17307 + }, + { + "epoch": 1.9007247968372503, + "grad_norm": 2.167149066925049, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7487819194793701, + "num_tokens": 441590180.0, + "step": 17308 + }, + { + "epoch": 1.9008346145398638, + "grad_norm": 2.1949565410614014, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.70745450258255, + "num_tokens": 441617419.0, + "step": 17309 + }, + { + "epoch": 1.9009444322424773, + "grad_norm": 2.107051372528076, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7294890284538269, + "num_tokens": 441645491.0, + "step": 17310 + }, + { + "epoch": 1.901054249945091, + "grad_norm": 2.114328384399414, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7084412574768066, + "num_tokens": 441674095.0, + "step": 17311 + }, + { + "epoch": 1.9011640676477048, + "grad_norm": 2.199230670928955, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7393978834152222, + "num_tokens": 441699422.0, + "step": 17312 + }, + { + "epoch": 1.9012738853503186, + "grad_norm": 2.349575996398926, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7192671895027161, + "num_tokens": 441722242.0, + "step": 17313 + }, + { + "epoch": 1.9013837030529321, + "grad_norm": 2.189289093017578, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7156785726547241, + "num_tokens": 441750764.0, + "step": 17314 + }, + { + "epoch": 1.9014935207555457, + "grad_norm": 2.432353973388672, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7323475480079651, + "num_tokens": 441772786.0, + "step": 17315 + }, + { + "epoch": 1.9016033384581594, + "grad_norm": 2.5843496322631836, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7442247867584229, + "num_tokens": 441792181.0, + "step": 17316 + }, + { + "epoch": 1.9017131561607732, + "grad_norm": 2.1303772926330566, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7190173864364624, + "num_tokens": 441820418.0, + "step": 17317 + }, + { + "epoch": 1.9018229738633867, + "grad_norm": 2.352165937423706, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7143651247024536, + "num_tokens": 441843724.0, + "step": 17318 + }, + { + "epoch": 1.9019327915660005, + "grad_norm": 2.4675750732421875, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.729387640953064, + "num_tokens": 441865284.0, + "step": 17319 + }, + { + "epoch": 1.902042609268614, + "grad_norm": 2.0815155506134033, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.723803699016571, + "num_tokens": 441893945.0, + "step": 17320 + }, + { + "epoch": 1.9021524269712278, + "grad_norm": 2.3606882095336914, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.727989673614502, + "num_tokens": 441916847.0, + "step": 17321 + }, + { + "epoch": 1.9022622446738415, + "grad_norm": 2.4047303199768066, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7206791639328003, + "num_tokens": 441939973.0, + "step": 17322 + }, + { + "epoch": 1.902372062376455, + "grad_norm": 2.030388832092285, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.701768159866333, + "num_tokens": 441970986.0, + "step": 17323 + }, + { + "epoch": 1.9024818800790686, + "grad_norm": 2.2896342277526855, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7343650460243225, + "num_tokens": 441995792.0, + "step": 17324 + }, + { + "epoch": 1.9025916977816824, + "grad_norm": 2.2256462574005127, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.741710901260376, + "num_tokens": 442019491.0, + "step": 17325 + }, + { + "epoch": 1.9027015154842961, + "grad_norm": 2.353166103363037, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7232680320739746, + "num_tokens": 442042437.0, + "step": 17326 + }, + { + "epoch": 1.9028113331869099, + "grad_norm": 2.053109884262085, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7175129652023315, + "num_tokens": 442072128.0, + "step": 17327 + }, + { + "epoch": 1.9029211508895234, + "grad_norm": 2.1588542461395264, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7190554141998291, + "num_tokens": 442098887.0, + "step": 17328 + }, + { + "epoch": 1.903030968592137, + "grad_norm": 2.655869960784912, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7039048671722412, + "num_tokens": 442121265.0, + "step": 17329 + }, + { + "epoch": 1.9031407862947507, + "grad_norm": 2.107715368270874, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7333037853240967, + "num_tokens": 442147254.0, + "step": 17330 + }, + { + "epoch": 1.9032506039973645, + "grad_norm": 2.0085394382476807, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7115416526794434, + "num_tokens": 442179803.0, + "step": 17331 + }, + { + "epoch": 1.903360421699978, + "grad_norm": 2.29940128326416, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7200068235397339, + "num_tokens": 442204833.0, + "step": 17332 + }, + { + "epoch": 1.9034702394025917, + "grad_norm": 2.0972630977630615, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.6962818503379822, + "num_tokens": 442234133.0, + "step": 17333 + }, + { + "epoch": 1.9035800571052053, + "grad_norm": 2.1392037868499756, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7260571122169495, + "num_tokens": 442263170.0, + "step": 17334 + }, + { + "epoch": 1.903689874807819, + "grad_norm": 2.2006914615631104, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7201070785522461, + "num_tokens": 442287888.0, + "step": 17335 + }, + { + "epoch": 1.9037996925104328, + "grad_norm": 2.203012228012085, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7046999931335449, + "num_tokens": 442313065.0, + "step": 17336 + }, + { + "epoch": 1.9039095102130463, + "grad_norm": 2.2489304542541504, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7221556901931763, + "num_tokens": 442340138.0, + "step": 17337 + }, + { + "epoch": 1.9040193279156599, + "grad_norm": 2.446777582168579, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7446829676628113, + "num_tokens": 442362098.0, + "step": 17338 + }, + { + "epoch": 1.9041291456182736, + "grad_norm": 1.8838213682174683, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7077984809875488, + "num_tokens": 442398009.0, + "step": 17339 + }, + { + "epoch": 1.9042389633208874, + "grad_norm": 2.3681864738464355, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7235357761383057, + "num_tokens": 442421356.0, + "step": 17340 + }, + { + "epoch": 1.9043487810235011, + "grad_norm": 2.3967084884643555, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7117874622344971, + "num_tokens": 442449210.0, + "step": 17341 + }, + { + "epoch": 1.9044585987261147, + "grad_norm": 1.8680884838104248, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7365021705627441, + "num_tokens": 442484796.0, + "step": 17342 + }, + { + "epoch": 1.9045684164287282, + "grad_norm": 2.241377592086792, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7247962355613708, + "num_tokens": 442509244.0, + "step": 17343 + }, + { + "epoch": 1.904678234131342, + "grad_norm": 2.060746908187866, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7315738201141357, + "num_tokens": 442535604.0, + "step": 17344 + }, + { + "epoch": 1.9047880518339557, + "grad_norm": 2.187675714492798, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7324135899543762, + "num_tokens": 442561148.0, + "step": 17345 + }, + { + "epoch": 1.9048978695365693, + "grad_norm": 2.15358304977417, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7183359861373901, + "num_tokens": 442588356.0, + "step": 17346 + }, + { + "epoch": 1.9050076872391828, + "grad_norm": 2.427133083343506, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7351294755935669, + "num_tokens": 442609419.0, + "step": 17347 + }, + { + "epoch": 1.9051175049417965, + "grad_norm": 2.597677707672119, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7178171277046204, + "num_tokens": 442629915.0, + "step": 17348 + }, + { + "epoch": 1.9052273226444103, + "grad_norm": 2.254850149154663, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7176580429077148, + "num_tokens": 442654828.0, + "step": 17349 + }, + { + "epoch": 1.905337140347024, + "grad_norm": 2.2006990909576416, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7196134328842163, + "num_tokens": 442679941.0, + "step": 17350 + }, + { + "epoch": 1.9054469580496376, + "grad_norm": 2.4743638038635254, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7311038970947266, + "num_tokens": 442700481.0, + "step": 17351 + }, + { + "epoch": 1.9055567757522511, + "grad_norm": 2.051321268081665, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7255836129188538, + "num_tokens": 442731005.0, + "step": 17352 + }, + { + "epoch": 1.9056665934548649, + "grad_norm": 2.2893123626708984, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7204691171646118, + "num_tokens": 442757046.0, + "step": 17353 + }, + { + "epoch": 1.9057764111574786, + "grad_norm": 2.40303635597229, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7285268306732178, + "num_tokens": 442779917.0, + "step": 17354 + }, + { + "epoch": 1.9058862288600924, + "grad_norm": 1.9988116025924683, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.686018705368042, + "num_tokens": 442811996.0, + "step": 17355 + }, + { + "epoch": 1.905996046562706, + "grad_norm": 2.527143716812134, + "learning_rate": 1e-06, + "loss": 0.7706, + "mean_token_accuracy": 0.7508439421653748, + "num_tokens": 442832928.0, + "step": 17356 + }, + { + "epoch": 1.9061058642653195, + "grad_norm": 2.4019012451171875, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.739237368106842, + "num_tokens": 442853939.0, + "step": 17357 + }, + { + "epoch": 1.9062156819679332, + "grad_norm": 2.2335565090179443, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.7010730504989624, + "num_tokens": 442881308.0, + "step": 17358 + }, + { + "epoch": 1.906325499670547, + "grad_norm": 2.213979721069336, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7400416731834412, + "num_tokens": 442905107.0, + "step": 17359 + }, + { + "epoch": 1.9064353173731605, + "grad_norm": 2.171501636505127, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7268490791320801, + "num_tokens": 442931764.0, + "step": 17360 + }, + { + "epoch": 1.906545135075774, + "grad_norm": 2.2682223320007324, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7407869100570679, + "num_tokens": 442957634.0, + "step": 17361 + }, + { + "epoch": 1.9066549527783878, + "grad_norm": 2.3400042057037354, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7211229801177979, + "num_tokens": 442983297.0, + "step": 17362 + }, + { + "epoch": 1.9067647704810016, + "grad_norm": 2.5150437355041504, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.725838303565979, + "num_tokens": 443005191.0, + "step": 17363 + }, + { + "epoch": 1.9068745881836153, + "grad_norm": 2.38127064704895, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7338401079177856, + "num_tokens": 443027282.0, + "step": 17364 + }, + { + "epoch": 1.9069844058862289, + "grad_norm": 2.329594850540161, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7145674228668213, + "num_tokens": 443053095.0, + "step": 17365 + }, + { + "epoch": 1.9070942235888424, + "grad_norm": 2.1655335426330566, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7506707906723022, + "num_tokens": 443077476.0, + "step": 17366 + }, + { + "epoch": 1.9072040412914562, + "grad_norm": 2.1357932090759277, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7027989029884338, + "num_tokens": 443106934.0, + "step": 17367 + }, + { + "epoch": 1.90731385899407, + "grad_norm": 2.1871438026428223, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7221399545669556, + "num_tokens": 443134270.0, + "step": 17368 + }, + { + "epoch": 1.9074236766966837, + "grad_norm": 2.2221691608428955, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.7390105128288269, + "num_tokens": 443158501.0, + "step": 17369 + }, + { + "epoch": 1.9075334943992972, + "grad_norm": 2.152036190032959, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7052424550056458, + "num_tokens": 443186366.0, + "step": 17370 + }, + { + "epoch": 1.9076433121019107, + "grad_norm": 1.8788431882858276, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7220489978790283, + "num_tokens": 443221856.0, + "step": 17371 + }, + { + "epoch": 1.9077531298045245, + "grad_norm": 2.086163282394409, + "learning_rate": 1e-06, + "loss": 1.0452, + "mean_token_accuracy": 0.68695068359375, + "num_tokens": 443252428.0, + "step": 17372 + }, + { + "epoch": 1.9078629475071383, + "grad_norm": 2.1606643199920654, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7017694711685181, + "num_tokens": 443279616.0, + "step": 17373 + }, + { + "epoch": 1.9079727652097518, + "grad_norm": 2.220226287841797, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7335244417190552, + "num_tokens": 443303645.0, + "step": 17374 + }, + { + "epoch": 1.9080825829123653, + "grad_norm": 2.3898212909698486, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.718582272529602, + "num_tokens": 443326619.0, + "step": 17375 + }, + { + "epoch": 1.908192400614979, + "grad_norm": 2.160111904144287, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7312830686569214, + "num_tokens": 443351813.0, + "step": 17376 + }, + { + "epoch": 1.9083022183175928, + "grad_norm": 2.2290496826171875, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7158277630805969, + "num_tokens": 443378179.0, + "step": 17377 + }, + { + "epoch": 1.9084120360202066, + "grad_norm": 2.257438898086548, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7407188415527344, + "num_tokens": 443403235.0, + "step": 17378 + }, + { + "epoch": 1.9085218537228201, + "grad_norm": 2.6215362548828125, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7301503419876099, + "num_tokens": 443423730.0, + "step": 17379 + }, + { + "epoch": 1.9086316714254337, + "grad_norm": 2.179793119430542, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7163020372390747, + "num_tokens": 443450362.0, + "step": 17380 + }, + { + "epoch": 1.9087414891280474, + "grad_norm": 2.31807279586792, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7253876328468323, + "num_tokens": 443473842.0, + "step": 17381 + }, + { + "epoch": 1.9088513068306612, + "grad_norm": 2.480489492416382, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7209054231643677, + "num_tokens": 443495715.0, + "step": 17382 + }, + { + "epoch": 1.9089611245332747, + "grad_norm": 2.2197341918945312, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7342629432678223, + "num_tokens": 443520272.0, + "step": 17383 + }, + { + "epoch": 1.9090709422358885, + "grad_norm": 2.6309027671813965, + "learning_rate": 1e-06, + "loss": 0.7981, + "mean_token_accuracy": 0.7441093325614929, + "num_tokens": 443538086.0, + "step": 17384 + }, + { + "epoch": 1.909180759938502, + "grad_norm": 2.2346200942993164, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7164849638938904, + "num_tokens": 443561688.0, + "step": 17385 + }, + { + "epoch": 1.9092905776411158, + "grad_norm": 2.081415891647339, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.70174241065979, + "num_tokens": 443592086.0, + "step": 17386 + }, + { + "epoch": 1.9094003953437295, + "grad_norm": 2.305205821990967, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.723547101020813, + "num_tokens": 443617154.0, + "step": 17387 + }, + { + "epoch": 1.909510213046343, + "grad_norm": 2.347297430038452, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7163739204406738, + "num_tokens": 443641116.0, + "step": 17388 + }, + { + "epoch": 1.9096200307489566, + "grad_norm": 2.214893102645874, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.716587245464325, + "num_tokens": 443667576.0, + "step": 17389 + }, + { + "epoch": 1.9097298484515703, + "grad_norm": 2.4931480884552, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7380175590515137, + "num_tokens": 443689059.0, + "step": 17390 + }, + { + "epoch": 1.909839666154184, + "grad_norm": 2.4114272594451904, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7196439504623413, + "num_tokens": 443712660.0, + "step": 17391 + }, + { + "epoch": 1.9099494838567979, + "grad_norm": 1.9865641593933105, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7150155305862427, + "num_tokens": 443743314.0, + "step": 17392 + }, + { + "epoch": 1.9100593015594114, + "grad_norm": 2.6278817653656006, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7150123119354248, + "num_tokens": 443762393.0, + "step": 17393 + }, + { + "epoch": 1.910169119262025, + "grad_norm": 2.234253406524658, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7128031849861145, + "num_tokens": 443789195.0, + "step": 17394 + }, + { + "epoch": 1.9102789369646387, + "grad_norm": 2.183046340942383, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7317937016487122, + "num_tokens": 443814167.0, + "step": 17395 + }, + { + "epoch": 1.9103887546672524, + "grad_norm": 2.038296699523926, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7161869406700134, + "num_tokens": 443844227.0, + "step": 17396 + }, + { + "epoch": 1.910498572369866, + "grad_norm": 2.3582208156585693, + "learning_rate": 1e-06, + "loss": 0.7531, + "mean_token_accuracy": 0.7586467862129211, + "num_tokens": 443865544.0, + "step": 17397 + }, + { + "epoch": 1.9106083900724797, + "grad_norm": 2.0437960624694824, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7109938263893127, + "num_tokens": 443896244.0, + "step": 17398 + }, + { + "epoch": 1.9107182077750933, + "grad_norm": 2.3366003036499023, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7375462651252747, + "num_tokens": 443920662.0, + "step": 17399 + }, + { + "epoch": 1.910828025477707, + "grad_norm": 2.141744375228882, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7325191497802734, + "num_tokens": 443945790.0, + "step": 17400 + }, + { + "epoch": 1.9109378431803208, + "grad_norm": 2.166038990020752, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7220567464828491, + "num_tokens": 443972197.0, + "step": 17401 + }, + { + "epoch": 1.9110476608829343, + "grad_norm": 2.2226171493530273, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7201493382453918, + "num_tokens": 443996619.0, + "step": 17402 + }, + { + "epoch": 1.9111574785855479, + "grad_norm": 2.721207857131958, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7249187231063843, + "num_tokens": 444015214.0, + "step": 17403 + }, + { + "epoch": 1.9112672962881616, + "grad_norm": 2.1452462673187256, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7203966379165649, + "num_tokens": 444044505.0, + "step": 17404 + }, + { + "epoch": 1.9113771139907754, + "grad_norm": 2.5266401767730713, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7298842668533325, + "num_tokens": 444064883.0, + "step": 17405 + }, + { + "epoch": 1.9114869316933891, + "grad_norm": 2.49389386177063, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.735593318939209, + "num_tokens": 444087720.0, + "step": 17406 + }, + { + "epoch": 1.9115967493960027, + "grad_norm": 2.440004587173462, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7361339926719666, + "num_tokens": 444107579.0, + "step": 17407 + }, + { + "epoch": 1.9117065670986162, + "grad_norm": 2.3471176624298096, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7270227670669556, + "num_tokens": 444130373.0, + "step": 17408 + }, + { + "epoch": 1.91181638480123, + "grad_norm": 2.2153990268707275, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7203677892684937, + "num_tokens": 444158875.0, + "step": 17409 + }, + { + "epoch": 1.9119262025038437, + "grad_norm": 2.104602575302124, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7174187898635864, + "num_tokens": 444186148.0, + "step": 17410 + }, + { + "epoch": 1.9120360202064572, + "grad_norm": 2.361830472946167, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7098361849784851, + "num_tokens": 444208520.0, + "step": 17411 + }, + { + "epoch": 1.9121458379090708, + "grad_norm": 2.326857089996338, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7340241074562073, + "num_tokens": 444234678.0, + "step": 17412 + }, + { + "epoch": 1.9122556556116845, + "grad_norm": 2.0379624366760254, + "learning_rate": 1e-06, + "loss": 0.8085, + "mean_token_accuracy": 0.7534601092338562, + "num_tokens": 444261766.0, + "step": 17413 + }, + { + "epoch": 1.9123654733142983, + "grad_norm": 2.3751094341278076, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.733435869216919, + "num_tokens": 444285517.0, + "step": 17414 + }, + { + "epoch": 1.912475291016912, + "grad_norm": 2.245953321456909, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.699447751045227, + "num_tokens": 444312046.0, + "step": 17415 + }, + { + "epoch": 1.9125851087195256, + "grad_norm": 2.3395919799804688, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7224928736686707, + "num_tokens": 444336265.0, + "step": 17416 + }, + { + "epoch": 1.9126949264221391, + "grad_norm": 2.133965015411377, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7388731241226196, + "num_tokens": 444362648.0, + "step": 17417 + }, + { + "epoch": 1.9128047441247529, + "grad_norm": 2.335495710372925, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7362892627716064, + "num_tokens": 444386617.0, + "step": 17418 + }, + { + "epoch": 1.9129145618273666, + "grad_norm": 2.1568901538848877, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7434139847755432, + "num_tokens": 444412379.0, + "step": 17419 + }, + { + "epoch": 1.9130243795299804, + "grad_norm": 2.498502254486084, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7310662865638733, + "num_tokens": 444434194.0, + "step": 17420 + }, + { + "epoch": 1.913134197232594, + "grad_norm": 2.4372777938842773, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7372127175331116, + "num_tokens": 444455917.0, + "step": 17421 + }, + { + "epoch": 1.9132440149352075, + "grad_norm": 2.284804105758667, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7156651020050049, + "num_tokens": 444483168.0, + "step": 17422 + }, + { + "epoch": 1.9133538326378212, + "grad_norm": 2.484912157058716, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7117177248001099, + "num_tokens": 444508219.0, + "step": 17423 + }, + { + "epoch": 1.913463650340435, + "grad_norm": 2.443763256072998, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7227997779846191, + "num_tokens": 444532409.0, + "step": 17424 + }, + { + "epoch": 1.9135734680430485, + "grad_norm": 2.204582929611206, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.720455527305603, + "num_tokens": 444558583.0, + "step": 17425 + }, + { + "epoch": 1.913683285745662, + "grad_norm": 2.263634443283081, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.731410026550293, + "num_tokens": 444582976.0, + "step": 17426 + }, + { + "epoch": 1.9137931034482758, + "grad_norm": 2.404762029647827, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7190598249435425, + "num_tokens": 444606523.0, + "step": 17427 + }, + { + "epoch": 1.9139029211508896, + "grad_norm": 1.8585726022720337, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7146055102348328, + "num_tokens": 444641658.0, + "step": 17428 + }, + { + "epoch": 1.9140127388535033, + "grad_norm": 2.5112602710723877, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.722133457660675, + "num_tokens": 444662850.0, + "step": 17429 + }, + { + "epoch": 1.9141225565561168, + "grad_norm": 2.221334218978882, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.709966242313385, + "num_tokens": 444689986.0, + "step": 17430 + }, + { + "epoch": 1.9142323742587304, + "grad_norm": 2.3908660411834717, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7197052240371704, + "num_tokens": 444713313.0, + "step": 17431 + }, + { + "epoch": 1.9143421919613441, + "grad_norm": 2.1702966690063477, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7175374031066895, + "num_tokens": 444739835.0, + "step": 17432 + }, + { + "epoch": 1.914452009663958, + "grad_norm": 2.0776920318603516, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7481334209442139, + "num_tokens": 444767518.0, + "step": 17433 + }, + { + "epoch": 1.9145618273665717, + "grad_norm": 2.409590244293213, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7364252805709839, + "num_tokens": 444789357.0, + "step": 17434 + }, + { + "epoch": 1.9146716450691852, + "grad_norm": 2.4354028701782227, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.7333393096923828, + "num_tokens": 444810505.0, + "step": 17435 + }, + { + "epoch": 1.9147814627717987, + "grad_norm": 2.304356575012207, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7018164396286011, + "num_tokens": 444835576.0, + "step": 17436 + }, + { + "epoch": 1.9148912804744125, + "grad_norm": 2.3630847930908203, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.6982828378677368, + "num_tokens": 444861481.0, + "step": 17437 + }, + { + "epoch": 1.9150010981770262, + "grad_norm": 2.3040106296539307, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7056674957275391, + "num_tokens": 444887745.0, + "step": 17438 + }, + { + "epoch": 1.9151109158796398, + "grad_norm": 2.2029919624328613, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7043758034706116, + "num_tokens": 444916486.0, + "step": 17439 + }, + { + "epoch": 1.9152207335822533, + "grad_norm": 2.4767696857452393, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7319563031196594, + "num_tokens": 444938805.0, + "step": 17440 + }, + { + "epoch": 1.915330551284867, + "grad_norm": 2.065678596496582, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7285338640213013, + "num_tokens": 444967351.0, + "step": 17441 + }, + { + "epoch": 1.9154403689874808, + "grad_norm": 2.271559953689575, + "learning_rate": 1e-06, + "loss": 0.8038, + "mean_token_accuracy": 0.7457974553108215, + "num_tokens": 444991536.0, + "step": 17442 + }, + { + "epoch": 1.9155501866900946, + "grad_norm": 2.0958354473114014, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7094919681549072, + "num_tokens": 445023852.0, + "step": 17443 + }, + { + "epoch": 1.9156600043927081, + "grad_norm": 2.3265440464019775, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.725192666053772, + "num_tokens": 445048209.0, + "step": 17444 + }, + { + "epoch": 1.9157698220953217, + "grad_norm": 2.217670202255249, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7167593240737915, + "num_tokens": 445074336.0, + "step": 17445 + }, + { + "epoch": 1.9158796397979354, + "grad_norm": 2.417546272277832, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7376686334609985, + "num_tokens": 445096358.0, + "step": 17446 + }, + { + "epoch": 1.9159894575005492, + "grad_norm": 2.200615167617798, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7045199871063232, + "num_tokens": 445123495.0, + "step": 17447 + }, + { + "epoch": 1.9160992752031627, + "grad_norm": 2.0655179023742676, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7043709754943848, + "num_tokens": 445153148.0, + "step": 17448 + }, + { + "epoch": 1.9162090929057765, + "grad_norm": 2.076310396194458, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7293910384178162, + "num_tokens": 445182421.0, + "step": 17449 + }, + { + "epoch": 1.91631891060839, + "grad_norm": 2.102191925048828, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7122693061828613, + "num_tokens": 445211712.0, + "step": 17450 + }, + { + "epoch": 1.9164287283110037, + "grad_norm": 2.0723447799682617, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7245399951934814, + "num_tokens": 445240281.0, + "step": 17451 + }, + { + "epoch": 1.9165385460136175, + "grad_norm": 2.3135838508605957, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.6972382664680481, + "num_tokens": 445266332.0, + "step": 17452 + }, + { + "epoch": 1.916648363716231, + "grad_norm": 2.2217907905578613, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7154382467269897, + "num_tokens": 445290972.0, + "step": 17453 + }, + { + "epoch": 1.9167581814188446, + "grad_norm": 2.246788501739502, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7241518497467041, + "num_tokens": 445316587.0, + "step": 17454 + }, + { + "epoch": 1.9168679991214583, + "grad_norm": 2.1847124099731445, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7197257876396179, + "num_tokens": 445343554.0, + "step": 17455 + }, + { + "epoch": 1.916977816824072, + "grad_norm": 2.1835837364196777, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7139434814453125, + "num_tokens": 445369362.0, + "step": 17456 + }, + { + "epoch": 1.9170876345266858, + "grad_norm": 2.382521629333496, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7380592823028564, + "num_tokens": 445391336.0, + "step": 17457 + }, + { + "epoch": 1.9171974522292994, + "grad_norm": 2.444918394088745, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7032397985458374, + "num_tokens": 445416698.0, + "step": 17458 + }, + { + "epoch": 1.917307269931913, + "grad_norm": 2.1485815048217773, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7153712511062622, + "num_tokens": 445444373.0, + "step": 17459 + }, + { + "epoch": 1.9174170876345267, + "grad_norm": 1.9204398393630981, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.711540162563324, + "num_tokens": 445477850.0, + "step": 17460 + }, + { + "epoch": 1.9175269053371404, + "grad_norm": 2.377101421356201, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7367739677429199, + "num_tokens": 445502157.0, + "step": 17461 + }, + { + "epoch": 1.917636723039754, + "grad_norm": 2.165435314178467, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7224911451339722, + "num_tokens": 445529542.0, + "step": 17462 + }, + { + "epoch": 1.9177465407423677, + "grad_norm": 2.1928274631500244, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7309675812721252, + "num_tokens": 445554367.0, + "step": 17463 + }, + { + "epoch": 1.9178563584449813, + "grad_norm": 2.4068081378936768, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7116906642913818, + "num_tokens": 445577217.0, + "step": 17464 + }, + { + "epoch": 1.917966176147595, + "grad_norm": 2.136107921600342, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7381763458251953, + "num_tokens": 445602399.0, + "step": 17465 + }, + { + "epoch": 1.9180759938502088, + "grad_norm": 2.051175355911255, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.6983788013458252, + "num_tokens": 445633604.0, + "step": 17466 + }, + { + "epoch": 1.9181858115528223, + "grad_norm": 2.142289638519287, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7269822955131531, + "num_tokens": 445660636.0, + "step": 17467 + }, + { + "epoch": 1.9182956292554358, + "grad_norm": 2.3067727088928223, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7104092836380005, + "num_tokens": 445685897.0, + "step": 17468 + }, + { + "epoch": 1.9184054469580496, + "grad_norm": 2.39198637008667, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7238705158233643, + "num_tokens": 445708672.0, + "step": 17469 + }, + { + "epoch": 1.9185152646606634, + "grad_norm": 2.270960569381714, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7141218781471252, + "num_tokens": 445732770.0, + "step": 17470 + }, + { + "epoch": 1.9186250823632771, + "grad_norm": 2.202465534210205, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.6956559419631958, + "num_tokens": 445759280.0, + "step": 17471 + }, + { + "epoch": 1.9187349000658906, + "grad_norm": 2.590479850769043, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7086224555969238, + "num_tokens": 445781161.0, + "step": 17472 + }, + { + "epoch": 1.9188447177685042, + "grad_norm": 2.2769038677215576, + "learning_rate": 1e-06, + "loss": 0.7448, + "mean_token_accuracy": 0.7626356482505798, + "num_tokens": 445803370.0, + "step": 17473 + }, + { + "epoch": 1.918954535471118, + "grad_norm": 2.2051455974578857, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7196490168571472, + "num_tokens": 445830061.0, + "step": 17474 + }, + { + "epoch": 1.9190643531737317, + "grad_norm": 2.089232921600342, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7358453869819641, + "num_tokens": 445857673.0, + "step": 17475 + }, + { + "epoch": 1.9191741708763452, + "grad_norm": 1.909358263015747, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7159475088119507, + "num_tokens": 445890028.0, + "step": 17476 + }, + { + "epoch": 1.9192839885789588, + "grad_norm": 2.420316219329834, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7377370595932007, + "num_tokens": 445911688.0, + "step": 17477 + }, + { + "epoch": 1.9193938062815725, + "grad_norm": 2.013127565383911, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6993163228034973, + "num_tokens": 445942372.0, + "step": 17478 + }, + { + "epoch": 1.9195036239841863, + "grad_norm": 1.9280366897583008, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.721245527267456, + "num_tokens": 445972989.0, + "step": 17479 + }, + { + "epoch": 1.9196134416868, + "grad_norm": 2.5212507247924805, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7305117845535278, + "num_tokens": 445992467.0, + "step": 17480 + }, + { + "epoch": 1.9197232593894136, + "grad_norm": 2.680765151977539, + "learning_rate": 1e-06, + "loss": 0.7791, + "mean_token_accuracy": 0.7444554567337036, + "num_tokens": 446011257.0, + "step": 17481 + }, + { + "epoch": 1.919833077092027, + "grad_norm": 2.0815844535827637, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7115781307220459, + "num_tokens": 446040542.0, + "step": 17482 + }, + { + "epoch": 1.9199428947946409, + "grad_norm": 2.3504159450531006, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7415018081665039, + "num_tokens": 446062262.0, + "step": 17483 + }, + { + "epoch": 1.9200527124972546, + "grad_norm": 2.1219098567962646, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7057616114616394, + "num_tokens": 446090953.0, + "step": 17484 + }, + { + "epoch": 1.9201625301998684, + "grad_norm": 2.058647394180298, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7045564651489258, + "num_tokens": 446119930.0, + "step": 17485 + }, + { + "epoch": 1.920272347902482, + "grad_norm": 2.1520283222198486, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7054719924926758, + "num_tokens": 446148695.0, + "step": 17486 + }, + { + "epoch": 1.9203821656050954, + "grad_norm": 2.495400905609131, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7567213177680969, + "num_tokens": 446167271.0, + "step": 17487 + }, + { + "epoch": 1.9204919833077092, + "grad_norm": 2.3423662185668945, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7033298015594482, + "num_tokens": 446190405.0, + "step": 17488 + }, + { + "epoch": 1.920601801010323, + "grad_norm": 2.0228421688079834, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7232367992401123, + "num_tokens": 446222893.0, + "step": 17489 + }, + { + "epoch": 1.9207116187129365, + "grad_norm": 2.479278802871704, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7163002490997314, + "num_tokens": 446247165.0, + "step": 17490 + }, + { + "epoch": 1.92082143641555, + "grad_norm": 2.4380905628204346, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7080400586128235, + "num_tokens": 446272210.0, + "step": 17491 + }, + { + "epoch": 1.9209312541181638, + "grad_norm": 2.5084450244903564, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7267681956291199, + "num_tokens": 446291984.0, + "step": 17492 + }, + { + "epoch": 1.9210410718207775, + "grad_norm": 2.152812957763672, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7172263860702515, + "num_tokens": 446321939.0, + "step": 17493 + }, + { + "epoch": 1.9211508895233913, + "grad_norm": 2.621709108352661, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.7463823556900024, + "num_tokens": 446342541.0, + "step": 17494 + }, + { + "epoch": 1.9212607072260048, + "grad_norm": 2.203583240509033, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7301799058914185, + "num_tokens": 446367839.0, + "step": 17495 + }, + { + "epoch": 1.9213705249286184, + "grad_norm": 2.13033127784729, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7225091457366943, + "num_tokens": 446399389.0, + "step": 17496 + }, + { + "epoch": 1.9214803426312321, + "grad_norm": 2.0788280963897705, + "learning_rate": 1e-06, + "loss": 0.7909, + "mean_token_accuracy": 0.7491986155509949, + "num_tokens": 446427267.0, + "step": 17497 + }, + { + "epoch": 1.9215901603338459, + "grad_norm": 2.058363199234009, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7204996943473816, + "num_tokens": 446456510.0, + "step": 17498 + }, + { + "epoch": 1.9216999780364594, + "grad_norm": 2.1389544010162354, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7067359685897827, + "num_tokens": 446485627.0, + "step": 17499 + }, + { + "epoch": 1.9218097957390732, + "grad_norm": 2.2513887882232666, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.6973780393600464, + "num_tokens": 446512809.0, + "step": 17500 + }, + { + "epoch": 1.9219196134416867, + "grad_norm": 2.2842836380004883, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7027024030685425, + "num_tokens": 446538107.0, + "step": 17501 + }, + { + "epoch": 1.9220294311443005, + "grad_norm": 2.6027896404266357, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7113141417503357, + "num_tokens": 446557978.0, + "step": 17502 + }, + { + "epoch": 1.9221392488469142, + "grad_norm": 2.419006109237671, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7368694543838501, + "num_tokens": 446580631.0, + "step": 17503 + }, + { + "epoch": 1.9222490665495278, + "grad_norm": 2.1992411613464355, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.7078816890716553, + "num_tokens": 446607916.0, + "step": 17504 + }, + { + "epoch": 1.9223588842521413, + "grad_norm": 2.3778634071350098, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7314373254776001, + "num_tokens": 446630037.0, + "step": 17505 + }, + { + "epoch": 1.922468701954755, + "grad_norm": 2.0701489448547363, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6900304555892944, + "num_tokens": 446660553.0, + "step": 17506 + }, + { + "epoch": 1.9225785196573688, + "grad_norm": 2.3964734077453613, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.726109504699707, + "num_tokens": 446681800.0, + "step": 17507 + }, + { + "epoch": 1.9226883373599826, + "grad_norm": 2.581645965576172, + "learning_rate": 1e-06, + "loss": 0.7784, + "mean_token_accuracy": 0.7578811049461365, + "num_tokens": 446700401.0, + "step": 17508 + }, + { + "epoch": 1.922798155062596, + "grad_norm": 2.5847761631011963, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7255040407180786, + "num_tokens": 446720564.0, + "step": 17509 + }, + { + "epoch": 1.9229079727652096, + "grad_norm": 2.1645119190216064, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7224987745285034, + "num_tokens": 446748011.0, + "step": 17510 + }, + { + "epoch": 1.9230177904678234, + "grad_norm": 2.5660510063171387, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7431779503822327, + "num_tokens": 446769188.0, + "step": 17511 + }, + { + "epoch": 1.9231276081704372, + "grad_norm": 2.006819486618042, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7039515376091003, + "num_tokens": 446802681.0, + "step": 17512 + }, + { + "epoch": 1.9232374258730507, + "grad_norm": 2.064349412918091, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7128124833106995, + "num_tokens": 446831914.0, + "step": 17513 + }, + { + "epoch": 1.9233472435756644, + "grad_norm": 2.3840177059173584, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7136279940605164, + "num_tokens": 446855268.0, + "step": 17514 + }, + { + "epoch": 1.923457061278278, + "grad_norm": 2.219271421432495, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.6973057985305786, + "num_tokens": 446883026.0, + "step": 17515 + }, + { + "epoch": 1.9235668789808917, + "grad_norm": 2.2507483959198, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7329567074775696, + "num_tokens": 446907020.0, + "step": 17516 + }, + { + "epoch": 1.9236766966835055, + "grad_norm": 2.018808126449585, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7279282808303833, + "num_tokens": 446937133.0, + "step": 17517 + }, + { + "epoch": 1.923786514386119, + "grad_norm": 2.6910736560821533, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7277617454528809, + "num_tokens": 446955403.0, + "step": 17518 + }, + { + "epoch": 1.9238963320887326, + "grad_norm": 2.235292673110962, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.707578718662262, + "num_tokens": 446981035.0, + "step": 17519 + }, + { + "epoch": 1.9240061497913463, + "grad_norm": 2.5335323810577393, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.730553388595581, + "num_tokens": 447001578.0, + "step": 17520 + }, + { + "epoch": 1.92411596749396, + "grad_norm": 2.213550329208374, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7098866701126099, + "num_tokens": 447028012.0, + "step": 17521 + }, + { + "epoch": 1.9242257851965738, + "grad_norm": 2.2149558067321777, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7052842378616333, + "num_tokens": 447056240.0, + "step": 17522 + }, + { + "epoch": 1.9243356028991874, + "grad_norm": 2.2927637100219727, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.719984769821167, + "num_tokens": 447081279.0, + "step": 17523 + }, + { + "epoch": 1.924445420601801, + "grad_norm": 2.263141393661499, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7086690664291382, + "num_tokens": 447107059.0, + "step": 17524 + }, + { + "epoch": 1.9245552383044147, + "grad_norm": 2.236968755722046, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7224666476249695, + "num_tokens": 447131919.0, + "step": 17525 + }, + { + "epoch": 1.9246650560070284, + "grad_norm": 2.349036693572998, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7337691187858582, + "num_tokens": 447155410.0, + "step": 17526 + }, + { + "epoch": 1.924774873709642, + "grad_norm": 2.626206874847412, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7214741110801697, + "num_tokens": 447175092.0, + "step": 17527 + }, + { + "epoch": 1.9248846914122557, + "grad_norm": 2.386821985244751, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7268944382667542, + "num_tokens": 447198755.0, + "step": 17528 + }, + { + "epoch": 1.9249945091148692, + "grad_norm": 2.298896312713623, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7255938053131104, + "num_tokens": 447224905.0, + "step": 17529 + }, + { + "epoch": 1.925104326817483, + "grad_norm": 2.5899126529693604, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7159600853919983, + "num_tokens": 447245470.0, + "step": 17530 + }, + { + "epoch": 1.9252141445200968, + "grad_norm": 2.3945488929748535, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7204751968383789, + "num_tokens": 447269956.0, + "step": 17531 + }, + { + "epoch": 1.9253239622227103, + "grad_norm": 2.197450876235962, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.757078230381012, + "num_tokens": 447297312.0, + "step": 17532 + }, + { + "epoch": 1.9254337799253238, + "grad_norm": 1.8939647674560547, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7043344974517822, + "num_tokens": 447329258.0, + "step": 17533 + }, + { + "epoch": 1.9255435976279376, + "grad_norm": 2.4386613368988037, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7377753257751465, + "num_tokens": 447350597.0, + "step": 17534 + }, + { + "epoch": 1.9256534153305513, + "grad_norm": 2.304751396179199, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7082546353340149, + "num_tokens": 447376725.0, + "step": 17535 + }, + { + "epoch": 1.925763233033165, + "grad_norm": 2.1927318572998047, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.707116961479187, + "num_tokens": 447403363.0, + "step": 17536 + }, + { + "epoch": 1.9258730507357786, + "grad_norm": 2.191898822784424, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7222417593002319, + "num_tokens": 447431174.0, + "step": 17537 + }, + { + "epoch": 1.9259828684383922, + "grad_norm": 2.108936071395874, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7009601593017578, + "num_tokens": 447460931.0, + "step": 17538 + }, + { + "epoch": 1.926092686141006, + "grad_norm": 2.1607964038848877, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7389691472053528, + "num_tokens": 447487353.0, + "step": 17539 + }, + { + "epoch": 1.9262025038436197, + "grad_norm": 2.243476390838623, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7279671430587769, + "num_tokens": 447513848.0, + "step": 17540 + }, + { + "epoch": 1.9263123215462332, + "grad_norm": 2.2592275142669678, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7273783683776855, + "num_tokens": 447538705.0, + "step": 17541 + }, + { + "epoch": 1.9264221392488468, + "grad_norm": 2.353250026702881, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7395696640014648, + "num_tokens": 447564206.0, + "step": 17542 + }, + { + "epoch": 1.9265319569514605, + "grad_norm": 2.184136152267456, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7006345391273499, + "num_tokens": 447594784.0, + "step": 17543 + }, + { + "epoch": 1.9266417746540743, + "grad_norm": 2.471886157989502, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7148281335830688, + "num_tokens": 447616833.0, + "step": 17544 + }, + { + "epoch": 1.926751592356688, + "grad_norm": 2.456033706665039, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7355633974075317, + "num_tokens": 447638237.0, + "step": 17545 + }, + { + "epoch": 1.9268614100593016, + "grad_norm": 2.0277624130249023, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7138469219207764, + "num_tokens": 447668104.0, + "step": 17546 + }, + { + "epoch": 1.926971227761915, + "grad_norm": 2.385472536087036, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7420310974121094, + "num_tokens": 447691200.0, + "step": 17547 + }, + { + "epoch": 1.9270810454645289, + "grad_norm": 2.073086738586426, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6908950209617615, + "num_tokens": 447721830.0, + "step": 17548 + }, + { + "epoch": 1.9271908631671426, + "grad_norm": 2.258857488632202, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7019672393798828, + "num_tokens": 447748276.0, + "step": 17549 + }, + { + "epoch": 1.9273006808697564, + "grad_norm": 2.267137289047241, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7193963527679443, + "num_tokens": 447773766.0, + "step": 17550 + }, + { + "epoch": 1.92741049857237, + "grad_norm": 2.0037639141082764, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7228282690048218, + "num_tokens": 447805095.0, + "step": 17551 + }, + { + "epoch": 1.9275203162749834, + "grad_norm": 2.196258783340454, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.6951320767402649, + "num_tokens": 447832277.0, + "step": 17552 + }, + { + "epoch": 1.9276301339775972, + "grad_norm": 2.236696481704712, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7209094762802124, + "num_tokens": 447858980.0, + "step": 17553 + }, + { + "epoch": 1.927739951680211, + "grad_norm": 2.369030475616455, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7283318042755127, + "num_tokens": 447880646.0, + "step": 17554 + }, + { + "epoch": 1.9278497693828245, + "grad_norm": 2.1072449684143066, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6976745128631592, + "num_tokens": 447908059.0, + "step": 17555 + }, + { + "epoch": 1.927959587085438, + "grad_norm": 2.343876838684082, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7189295291900635, + "num_tokens": 447932575.0, + "step": 17556 + }, + { + "epoch": 1.9280694047880518, + "grad_norm": 2.14426851272583, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7075760364532471, + "num_tokens": 447959591.0, + "step": 17557 + }, + { + "epoch": 1.9281792224906655, + "grad_norm": 2.242805004119873, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7298113107681274, + "num_tokens": 447985823.0, + "step": 17558 + }, + { + "epoch": 1.9282890401932793, + "grad_norm": 2.2023701667785645, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7095515131950378, + "num_tokens": 448012782.0, + "step": 17559 + }, + { + "epoch": 1.9283988578958928, + "grad_norm": 2.610694169998169, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7386187314987183, + "num_tokens": 448032220.0, + "step": 17560 + }, + { + "epoch": 1.9285086755985064, + "grad_norm": 2.071563243865967, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6984580159187317, + "num_tokens": 448064228.0, + "step": 17561 + }, + { + "epoch": 1.9286184933011201, + "grad_norm": 2.0214741230010986, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6888328194618225, + "num_tokens": 448094012.0, + "step": 17562 + }, + { + "epoch": 1.9287283110037339, + "grad_norm": 2.241706132888794, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.711240828037262, + "num_tokens": 448121414.0, + "step": 17563 + }, + { + "epoch": 1.9288381287063474, + "grad_norm": 2.4810678958892822, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7235223054885864, + "num_tokens": 448145485.0, + "step": 17564 + }, + { + "epoch": 1.9289479464089612, + "grad_norm": 2.1125991344451904, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7333337068557739, + "num_tokens": 448172786.0, + "step": 17565 + }, + { + "epoch": 1.9290577641115747, + "grad_norm": 2.2296664714813232, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7169134616851807, + "num_tokens": 448196133.0, + "step": 17566 + }, + { + "epoch": 1.9291675818141885, + "grad_norm": 2.2346346378326416, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.702926754951477, + "num_tokens": 448223206.0, + "step": 17567 + }, + { + "epoch": 1.9292773995168022, + "grad_norm": 2.4034266471862793, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.716783881187439, + "num_tokens": 448244397.0, + "step": 17568 + }, + { + "epoch": 1.9293872172194158, + "grad_norm": 2.371037244796753, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7183161973953247, + "num_tokens": 448265952.0, + "step": 17569 + }, + { + "epoch": 1.9294970349220293, + "grad_norm": 2.3440439701080322, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.7057442665100098, + "num_tokens": 448291205.0, + "step": 17570 + }, + { + "epoch": 1.929606852624643, + "grad_norm": 2.347024440765381, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7229773998260498, + "num_tokens": 448315567.0, + "step": 17571 + }, + { + "epoch": 1.9297166703272568, + "grad_norm": 2.3051671981811523, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7028066515922546, + "num_tokens": 448341805.0, + "step": 17572 + }, + { + "epoch": 1.9298264880298706, + "grad_norm": 2.410660743713379, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7310658097267151, + "num_tokens": 448364077.0, + "step": 17573 + }, + { + "epoch": 1.929936305732484, + "grad_norm": 2.6129086017608643, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.743220329284668, + "num_tokens": 448383981.0, + "step": 17574 + }, + { + "epoch": 1.9300461234350976, + "grad_norm": 2.4308595657348633, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7053498029708862, + "num_tokens": 448406674.0, + "step": 17575 + }, + { + "epoch": 1.9301559411377114, + "grad_norm": 2.4488327503204346, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7291136980056763, + "num_tokens": 448427565.0, + "step": 17576 + }, + { + "epoch": 1.9302657588403251, + "grad_norm": 2.594741106033325, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7224404811859131, + "num_tokens": 448448449.0, + "step": 17577 + }, + { + "epoch": 1.9303755765429387, + "grad_norm": 2.2678792476654053, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7321237921714783, + "num_tokens": 448474504.0, + "step": 17578 + }, + { + "epoch": 1.9304853942455524, + "grad_norm": 2.1541404724121094, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7197808027267456, + "num_tokens": 448500208.0, + "step": 17579 + }, + { + "epoch": 1.930595211948166, + "grad_norm": 2.48287034034729, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.717376172542572, + "num_tokens": 448522105.0, + "step": 17580 + }, + { + "epoch": 1.9307050296507797, + "grad_norm": 2.1654229164123535, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.706244945526123, + "num_tokens": 448550415.0, + "step": 17581 + }, + { + "epoch": 1.9308148473533935, + "grad_norm": 2.3600151538848877, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.727350652217865, + "num_tokens": 448574583.0, + "step": 17582 + }, + { + "epoch": 1.930924665056007, + "grad_norm": 2.3430047035217285, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.705436110496521, + "num_tokens": 448598539.0, + "step": 17583 + }, + { + "epoch": 1.9310344827586206, + "grad_norm": 2.5168182849884033, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7311050295829773, + "num_tokens": 448619492.0, + "step": 17584 + }, + { + "epoch": 1.9311443004612343, + "grad_norm": 2.3717057704925537, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.6946998834609985, + "num_tokens": 448644340.0, + "step": 17585 + }, + { + "epoch": 1.931254118163848, + "grad_norm": 2.161677122116089, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7403891086578369, + "num_tokens": 448670447.0, + "step": 17586 + }, + { + "epoch": 1.9313639358664618, + "grad_norm": 2.1482858657836914, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7221965789794922, + "num_tokens": 448695437.0, + "step": 17587 + }, + { + "epoch": 1.9314737535690754, + "grad_norm": 2.2038722038269043, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7137013077735901, + "num_tokens": 448722680.0, + "step": 17588 + }, + { + "epoch": 1.931583571271689, + "grad_norm": 2.5578770637512207, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7321494817733765, + "num_tokens": 448743265.0, + "step": 17589 + }, + { + "epoch": 1.9316933889743026, + "grad_norm": 2.3114914894104004, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7253044843673706, + "num_tokens": 448768088.0, + "step": 17590 + }, + { + "epoch": 1.9318032066769164, + "grad_norm": 2.288738965988159, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7137421369552612, + "num_tokens": 448793948.0, + "step": 17591 + }, + { + "epoch": 1.93191302437953, + "grad_norm": 2.0602989196777344, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7251782417297363, + "num_tokens": 448824642.0, + "step": 17592 + }, + { + "epoch": 1.9320228420821435, + "grad_norm": 2.1464412212371826, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7244493961334229, + "num_tokens": 448850975.0, + "step": 17593 + }, + { + "epoch": 1.9321326597847572, + "grad_norm": 2.61142635345459, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7231428623199463, + "num_tokens": 448871109.0, + "step": 17594 + }, + { + "epoch": 1.932242477487371, + "grad_norm": 2.140469789505005, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.717542827129364, + "num_tokens": 448897166.0, + "step": 17595 + }, + { + "epoch": 1.9323522951899847, + "grad_norm": 1.9951542615890503, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7453302145004272, + "num_tokens": 448926557.0, + "step": 17596 + }, + { + "epoch": 1.9324621128925983, + "grad_norm": 2.2167012691497803, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7354644536972046, + "num_tokens": 448953415.0, + "step": 17597 + }, + { + "epoch": 1.9325719305952118, + "grad_norm": 2.354444980621338, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7499802708625793, + "num_tokens": 448975984.0, + "step": 17598 + }, + { + "epoch": 1.9326817482978256, + "grad_norm": 1.9587305784225464, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7145013809204102, + "num_tokens": 449008834.0, + "step": 17599 + }, + { + "epoch": 1.9327915660004393, + "grad_norm": 2.0643301010131836, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7179126143455505, + "num_tokens": 449037746.0, + "step": 17600 + }, + { + "epoch": 1.932901383703053, + "grad_norm": 2.4813148975372314, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7341631650924683, + "num_tokens": 449059223.0, + "step": 17601 + }, + { + "epoch": 1.9330112014056666, + "grad_norm": 2.2628273963928223, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7272022366523743, + "num_tokens": 449084645.0, + "step": 17602 + }, + { + "epoch": 1.9331210191082802, + "grad_norm": 2.1387031078338623, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7336760759353638, + "num_tokens": 449113122.0, + "step": 17603 + }, + { + "epoch": 1.933230836810894, + "grad_norm": 2.4228968620300293, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7136164903640747, + "num_tokens": 449137156.0, + "step": 17604 + }, + { + "epoch": 1.9333406545135077, + "grad_norm": 1.972898244857788, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7063598036766052, + "num_tokens": 449169934.0, + "step": 17605 + }, + { + "epoch": 1.9334504722161212, + "grad_norm": 2.356802463531494, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7042760252952576, + "num_tokens": 449194132.0, + "step": 17606 + }, + { + "epoch": 1.9335602899187347, + "grad_norm": 2.246992588043213, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7015321254730225, + "num_tokens": 449222378.0, + "step": 17607 + }, + { + "epoch": 1.9336701076213485, + "grad_norm": 2.5811593532562256, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7327300310134888, + "num_tokens": 449243362.0, + "step": 17608 + }, + { + "epoch": 1.9337799253239623, + "grad_norm": 2.1886239051818848, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7141200304031372, + "num_tokens": 449270585.0, + "step": 17609 + }, + { + "epoch": 1.933889743026576, + "grad_norm": 2.1372880935668945, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7090746164321899, + "num_tokens": 449300488.0, + "step": 17610 + }, + { + "epoch": 1.9339995607291895, + "grad_norm": 2.2597014904022217, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7111891508102417, + "num_tokens": 449327620.0, + "step": 17611 + }, + { + "epoch": 1.934109378431803, + "grad_norm": 2.385849714279175, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7273730039596558, + "num_tokens": 449349433.0, + "step": 17612 + }, + { + "epoch": 1.9342191961344168, + "grad_norm": 2.3865768909454346, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7280229330062866, + "num_tokens": 449370877.0, + "step": 17613 + }, + { + "epoch": 1.9343290138370306, + "grad_norm": 2.281432867050171, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7371571063995361, + "num_tokens": 449394033.0, + "step": 17614 + }, + { + "epoch": 1.9344388315396444, + "grad_norm": 2.1562302112579346, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.749333381652832, + "num_tokens": 449419099.0, + "step": 17615 + }, + { + "epoch": 1.934548649242258, + "grad_norm": 2.349540948867798, + "learning_rate": 1e-06, + "loss": 0.8189, + "mean_token_accuracy": 0.7442769408226013, + "num_tokens": 449441479.0, + "step": 17616 + }, + { + "epoch": 1.9346584669448714, + "grad_norm": 2.103511095046997, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7108869552612305, + "num_tokens": 449469532.0, + "step": 17617 + }, + { + "epoch": 1.9347682846474852, + "grad_norm": 2.254528284072876, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.712376594543457, + "num_tokens": 449496450.0, + "step": 17618 + }, + { + "epoch": 1.934878102350099, + "grad_norm": 2.2213327884674072, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7184667587280273, + "num_tokens": 449523201.0, + "step": 17619 + }, + { + "epoch": 1.9349879200527125, + "grad_norm": 2.4679341316223145, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.732049822807312, + "num_tokens": 449544194.0, + "step": 17620 + }, + { + "epoch": 1.935097737755326, + "grad_norm": 2.2920966148376465, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7110753655433655, + "num_tokens": 449570767.0, + "step": 17621 + }, + { + "epoch": 1.9352075554579398, + "grad_norm": 2.146005868911743, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.7018736600875854, + "num_tokens": 449599178.0, + "step": 17622 + }, + { + "epoch": 1.9353173731605535, + "grad_norm": 2.4588966369628906, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7228357791900635, + "num_tokens": 449619835.0, + "step": 17623 + }, + { + "epoch": 1.9354271908631673, + "grad_norm": 2.597783088684082, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7332148551940918, + "num_tokens": 449639618.0, + "step": 17624 + }, + { + "epoch": 1.9355370085657808, + "grad_norm": 2.131376266479492, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7114761471748352, + "num_tokens": 449666705.0, + "step": 17625 + }, + { + "epoch": 1.9356468262683943, + "grad_norm": 2.2921953201293945, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7351721525192261, + "num_tokens": 449690880.0, + "step": 17626 + }, + { + "epoch": 1.935756643971008, + "grad_norm": 2.244603157043457, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.6985251903533936, + "num_tokens": 449716948.0, + "step": 17627 + }, + { + "epoch": 1.9358664616736219, + "grad_norm": 2.153824806213379, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7197425365447998, + "num_tokens": 449744247.0, + "step": 17628 + }, + { + "epoch": 1.9359762793762354, + "grad_norm": 2.1577866077423096, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7286244630813599, + "num_tokens": 449769895.0, + "step": 17629 + }, + { + "epoch": 1.9360860970788492, + "grad_norm": 2.322249174118042, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7335867881774902, + "num_tokens": 449794638.0, + "step": 17630 + }, + { + "epoch": 1.9361959147814627, + "grad_norm": 2.0030670166015625, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.6991937756538391, + "num_tokens": 449826538.0, + "step": 17631 + }, + { + "epoch": 1.9363057324840764, + "grad_norm": 2.4976720809936523, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7116296291351318, + "num_tokens": 449849811.0, + "step": 17632 + }, + { + "epoch": 1.9364155501866902, + "grad_norm": 2.213327169418335, + "learning_rate": 1e-06, + "loss": 0.7759, + "mean_token_accuracy": 0.7526751160621643, + "num_tokens": 449872536.0, + "step": 17633 + }, + { + "epoch": 1.9365253678893037, + "grad_norm": 2.0080840587615967, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7233017086982727, + "num_tokens": 449905337.0, + "step": 17634 + }, + { + "epoch": 1.9366351855919173, + "grad_norm": 1.8565436601638794, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7130112648010254, + "num_tokens": 449937078.0, + "step": 17635 + }, + { + "epoch": 1.936745003294531, + "grad_norm": 2.1035916805267334, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7171527147293091, + "num_tokens": 449963477.0, + "step": 17636 + }, + { + "epoch": 1.9368548209971448, + "grad_norm": 2.3288826942443848, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7181275486946106, + "num_tokens": 449985791.0, + "step": 17637 + }, + { + "epoch": 1.9369646386997585, + "grad_norm": 2.556018114089966, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.7468152046203613, + "num_tokens": 450004844.0, + "step": 17638 + }, + { + "epoch": 1.937074456402372, + "grad_norm": 2.071986675262451, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7380380630493164, + "num_tokens": 450030835.0, + "step": 17639 + }, + { + "epoch": 1.9371842741049856, + "grad_norm": 2.367151975631714, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7059605121612549, + "num_tokens": 450056716.0, + "step": 17640 + }, + { + "epoch": 1.9372940918075994, + "grad_norm": 2.2487258911132812, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6976245641708374, + "num_tokens": 450084783.0, + "step": 17641 + }, + { + "epoch": 1.9374039095102131, + "grad_norm": 2.1464505195617676, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7090476751327515, + "num_tokens": 450111006.0, + "step": 17642 + }, + { + "epoch": 1.9375137272128267, + "grad_norm": 2.4251649379730225, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.699994683265686, + "num_tokens": 450134466.0, + "step": 17643 + }, + { + "epoch": 1.9376235449154404, + "grad_norm": 2.172093629837036, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7150568962097168, + "num_tokens": 450161873.0, + "step": 17644 + }, + { + "epoch": 1.937733362618054, + "grad_norm": 2.2289345264434814, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7307252883911133, + "num_tokens": 450186479.0, + "step": 17645 + }, + { + "epoch": 1.9378431803206677, + "grad_norm": 1.8811554908752441, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7131766676902771, + "num_tokens": 450219585.0, + "step": 17646 + }, + { + "epoch": 1.9379529980232815, + "grad_norm": 2.1710712909698486, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7319484949111938, + "num_tokens": 450246435.0, + "step": 17647 + }, + { + "epoch": 1.938062815725895, + "grad_norm": 2.2552571296691895, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7227151393890381, + "num_tokens": 450272375.0, + "step": 17648 + }, + { + "epoch": 1.9381726334285085, + "grad_norm": 2.2961738109588623, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7217214107513428, + "num_tokens": 450295681.0, + "step": 17649 + }, + { + "epoch": 1.9382824511311223, + "grad_norm": 2.266679048538208, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7354556322097778, + "num_tokens": 450319728.0, + "step": 17650 + }, + { + "epoch": 1.938392268833736, + "grad_norm": 2.3315324783325195, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.724544107913971, + "num_tokens": 450341401.0, + "step": 17651 + }, + { + "epoch": 1.9385020865363498, + "grad_norm": 2.450068235397339, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7242411375045776, + "num_tokens": 450364796.0, + "step": 17652 + }, + { + "epoch": 1.9386119042389633, + "grad_norm": 2.3784708976745605, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7165906429290771, + "num_tokens": 450387941.0, + "step": 17653 + }, + { + "epoch": 1.9387217219415769, + "grad_norm": 2.2004802227020264, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7338711023330688, + "num_tokens": 450413681.0, + "step": 17654 + }, + { + "epoch": 1.9388315396441906, + "grad_norm": 2.169159412384033, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7041691541671753, + "num_tokens": 450441979.0, + "step": 17655 + }, + { + "epoch": 1.9389413573468044, + "grad_norm": 2.6068756580352783, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7406867146492004, + "num_tokens": 450462412.0, + "step": 17656 + }, + { + "epoch": 1.939051175049418, + "grad_norm": 2.2263729572296143, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7109758853912354, + "num_tokens": 450487660.0, + "step": 17657 + }, + { + "epoch": 1.9391609927520315, + "grad_norm": 2.501621723175049, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7095317840576172, + "num_tokens": 450509004.0, + "step": 17658 + }, + { + "epoch": 1.9392708104546452, + "grad_norm": 1.8325444459915161, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7226120829582214, + "num_tokens": 450540967.0, + "step": 17659 + }, + { + "epoch": 1.939380628157259, + "grad_norm": 2.1382102966308594, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.700559139251709, + "num_tokens": 450570340.0, + "step": 17660 + }, + { + "epoch": 1.9394904458598727, + "grad_norm": 2.3106141090393066, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7419828176498413, + "num_tokens": 450593075.0, + "step": 17661 + }, + { + "epoch": 1.9396002635624863, + "grad_norm": 2.4419703483581543, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7230788469314575, + "num_tokens": 450614414.0, + "step": 17662 + }, + { + "epoch": 1.9397100812650998, + "grad_norm": 2.4572882652282715, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7729604244232178, + "num_tokens": 450633547.0, + "step": 17663 + }, + { + "epoch": 1.9398198989677136, + "grad_norm": 2.4434549808502197, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7559431195259094, + "num_tokens": 450652906.0, + "step": 17664 + }, + { + "epoch": 1.9399297166703273, + "grad_norm": 2.306203603744507, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7041078209877014, + "num_tokens": 450676916.0, + "step": 17665 + }, + { + "epoch": 1.940039534372941, + "grad_norm": 2.0686092376708984, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7366377115249634, + "num_tokens": 450704809.0, + "step": 17666 + }, + { + "epoch": 1.9401493520755546, + "grad_norm": 2.182778835296631, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7118876576423645, + "num_tokens": 450732361.0, + "step": 17667 + }, + { + "epoch": 1.9402591697781681, + "grad_norm": 2.4486560821533203, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7226293087005615, + "num_tokens": 450754805.0, + "step": 17668 + }, + { + "epoch": 1.940368987480782, + "grad_norm": 2.455573558807373, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7437615990638733, + "num_tokens": 450776984.0, + "step": 17669 + }, + { + "epoch": 1.9404788051833957, + "grad_norm": 2.6706418991088867, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7264745831489563, + "num_tokens": 450796681.0, + "step": 17670 + }, + { + "epoch": 1.9405886228860092, + "grad_norm": 2.2099204063415527, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7055473327636719, + "num_tokens": 450823506.0, + "step": 17671 + }, + { + "epoch": 1.9406984405886227, + "grad_norm": 2.2220637798309326, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7194756269454956, + "num_tokens": 450847502.0, + "step": 17672 + }, + { + "epoch": 1.9408082582912365, + "grad_norm": 2.172966480255127, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7042128443717957, + "num_tokens": 450873647.0, + "step": 17673 + }, + { + "epoch": 1.9409180759938502, + "grad_norm": 2.0210025310516357, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.704730749130249, + "num_tokens": 450905345.0, + "step": 17674 + }, + { + "epoch": 1.941027893696464, + "grad_norm": 2.125777244567871, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7028010487556458, + "num_tokens": 450933536.0, + "step": 17675 + }, + { + "epoch": 1.9411377113990775, + "grad_norm": 2.3786680698394775, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7271906733512878, + "num_tokens": 450955172.0, + "step": 17676 + }, + { + "epoch": 1.941247529101691, + "grad_norm": 2.189850330352783, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7212351560592651, + "num_tokens": 450981104.0, + "step": 17677 + }, + { + "epoch": 1.9413573468043048, + "grad_norm": 2.141411304473877, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6928320527076721, + "num_tokens": 451010585.0, + "step": 17678 + }, + { + "epoch": 1.9414671645069186, + "grad_norm": 2.4539690017700195, + "learning_rate": 1e-06, + "loss": 0.7612, + "mean_token_accuracy": 0.7557581067085266, + "num_tokens": 451029701.0, + "step": 17679 + }, + { + "epoch": 1.9415769822095323, + "grad_norm": 2.1203699111938477, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7254844307899475, + "num_tokens": 451056270.0, + "step": 17680 + }, + { + "epoch": 1.9416867999121459, + "grad_norm": 2.461679458618164, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7303498983383179, + "num_tokens": 451077361.0, + "step": 17681 + }, + { + "epoch": 1.9417966176147594, + "grad_norm": 2.400240182876587, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7320067882537842, + "num_tokens": 451097956.0, + "step": 17682 + }, + { + "epoch": 1.9419064353173732, + "grad_norm": 2.016761064529419, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7161829471588135, + "num_tokens": 451127710.0, + "step": 17683 + }, + { + "epoch": 1.942016253019987, + "grad_norm": 2.3329553604125977, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7350015640258789, + "num_tokens": 451150788.0, + "step": 17684 + }, + { + "epoch": 1.9421260707226005, + "grad_norm": 2.202617645263672, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.749886155128479, + "num_tokens": 451175800.0, + "step": 17685 + }, + { + "epoch": 1.942235888425214, + "grad_norm": 2.2217609882354736, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7315807342529297, + "num_tokens": 451204059.0, + "step": 17686 + }, + { + "epoch": 1.9423457061278278, + "grad_norm": 2.393714189529419, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.72459876537323, + "num_tokens": 451228040.0, + "step": 17687 + }, + { + "epoch": 1.9424555238304415, + "grad_norm": 2.0296342372894287, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7095134258270264, + "num_tokens": 451257609.0, + "step": 17688 + }, + { + "epoch": 1.9425653415330553, + "grad_norm": 2.6197967529296875, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7334291934967041, + "num_tokens": 451277531.0, + "step": 17689 + }, + { + "epoch": 1.9426751592356688, + "grad_norm": 2.1027097702026367, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7245011925697327, + "num_tokens": 451307530.0, + "step": 17690 + }, + { + "epoch": 1.9427849769382823, + "grad_norm": 2.39587664604187, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7342193126678467, + "num_tokens": 451330114.0, + "step": 17691 + }, + { + "epoch": 1.942894794640896, + "grad_norm": 2.0111632347106934, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7218831777572632, + "num_tokens": 451361490.0, + "step": 17692 + }, + { + "epoch": 1.9430046123435099, + "grad_norm": 2.0836689472198486, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7099418640136719, + "num_tokens": 451390774.0, + "step": 17693 + }, + { + "epoch": 1.9431144300461234, + "grad_norm": 1.9349876642227173, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7271690964698792, + "num_tokens": 451422673.0, + "step": 17694 + }, + { + "epoch": 1.9432242477487371, + "grad_norm": 2.3556318283081055, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7319009304046631, + "num_tokens": 451443658.0, + "step": 17695 + }, + { + "epoch": 1.9433340654513507, + "grad_norm": 2.688953161239624, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7245212197303772, + "num_tokens": 451462135.0, + "step": 17696 + }, + { + "epoch": 1.9434438831539644, + "grad_norm": 2.3247015476226807, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7227513194084167, + "num_tokens": 451486512.0, + "step": 17697 + }, + { + "epoch": 1.9435537008565782, + "grad_norm": 2.284868001937866, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7136775255203247, + "num_tokens": 451512541.0, + "step": 17698 + }, + { + "epoch": 1.9436635185591917, + "grad_norm": 2.190551280975342, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7429252862930298, + "num_tokens": 451536288.0, + "step": 17699 + }, + { + "epoch": 1.9437733362618053, + "grad_norm": 2.0761754512786865, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7038470506668091, + "num_tokens": 451566517.0, + "step": 17700 + }, + { + "epoch": 1.943883153964419, + "grad_norm": 2.329573631286621, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7205053567886353, + "num_tokens": 451589194.0, + "step": 17701 + }, + { + "epoch": 1.9439929716670328, + "grad_norm": 2.0725574493408203, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7504959106445312, + "num_tokens": 451615585.0, + "step": 17702 + }, + { + "epoch": 1.9441027893696465, + "grad_norm": 2.1105422973632812, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7092036604881287, + "num_tokens": 451644055.0, + "step": 17703 + }, + { + "epoch": 1.94421260707226, + "grad_norm": 2.2892003059387207, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7072259187698364, + "num_tokens": 451670741.0, + "step": 17704 + }, + { + "epoch": 1.9443224247748736, + "grad_norm": 2.1607654094696045, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.725592851638794, + "num_tokens": 451696414.0, + "step": 17705 + }, + { + "epoch": 1.9444322424774874, + "grad_norm": 2.5699563026428223, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7334875464439392, + "num_tokens": 451715490.0, + "step": 17706 + }, + { + "epoch": 1.9445420601801011, + "grad_norm": 2.6111888885498047, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7310681939125061, + "num_tokens": 451735789.0, + "step": 17707 + }, + { + "epoch": 1.9446518778827147, + "grad_norm": 2.1492719650268555, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7019274234771729, + "num_tokens": 451766015.0, + "step": 17708 + }, + { + "epoch": 1.9447616955853284, + "grad_norm": 2.408599615097046, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7366637587547302, + "num_tokens": 451787432.0, + "step": 17709 + }, + { + "epoch": 1.944871513287942, + "grad_norm": 2.190877676010132, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7119036316871643, + "num_tokens": 451813736.0, + "step": 17710 + }, + { + "epoch": 1.9449813309905557, + "grad_norm": 2.275845527648926, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7205643653869629, + "num_tokens": 451838398.0, + "step": 17711 + }, + { + "epoch": 1.9450911486931695, + "grad_norm": 2.302276849746704, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.719172477722168, + "num_tokens": 451863183.0, + "step": 17712 + }, + { + "epoch": 1.945200966395783, + "grad_norm": 2.2529540061950684, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7073442935943604, + "num_tokens": 451888957.0, + "step": 17713 + }, + { + "epoch": 1.9453107840983965, + "grad_norm": 1.926013708114624, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7157468795776367, + "num_tokens": 451920526.0, + "step": 17714 + }, + { + "epoch": 1.9454206018010103, + "grad_norm": 2.2816684246063232, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7147800922393799, + "num_tokens": 451944005.0, + "step": 17715 + }, + { + "epoch": 1.945530419503624, + "grad_norm": 2.4419987201690674, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.72230064868927, + "num_tokens": 451966553.0, + "step": 17716 + }, + { + "epoch": 1.9456402372062378, + "grad_norm": 2.016399621963501, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7071016430854797, + "num_tokens": 451997484.0, + "step": 17717 + }, + { + "epoch": 1.9457500549088513, + "grad_norm": 2.7491166591644287, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7289714217185974, + "num_tokens": 452016089.0, + "step": 17718 + }, + { + "epoch": 1.9458598726114649, + "grad_norm": 2.073639154434204, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7110689878463745, + "num_tokens": 452045232.0, + "step": 17719 + }, + { + "epoch": 1.9459696903140786, + "grad_norm": 2.5808215141296387, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7266397476196289, + "num_tokens": 452064874.0, + "step": 17720 + }, + { + "epoch": 1.9460795080166924, + "grad_norm": 2.2748165130615234, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7078608870506287, + "num_tokens": 452091165.0, + "step": 17721 + }, + { + "epoch": 1.946189325719306, + "grad_norm": 1.9854665994644165, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7194710969924927, + "num_tokens": 452121671.0, + "step": 17722 + }, + { + "epoch": 1.9462991434219195, + "grad_norm": 2.703186273574829, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7354540824890137, + "num_tokens": 452139833.0, + "step": 17723 + }, + { + "epoch": 1.9464089611245332, + "grad_norm": 2.056278705596924, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7433865070343018, + "num_tokens": 452166507.0, + "step": 17724 + }, + { + "epoch": 1.946518778827147, + "grad_norm": 2.4816272258758545, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7384047508239746, + "num_tokens": 452187883.0, + "step": 17725 + }, + { + "epoch": 1.9466285965297607, + "grad_norm": 2.385741949081421, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7272815108299255, + "num_tokens": 452210880.0, + "step": 17726 + }, + { + "epoch": 1.9467384142323743, + "grad_norm": 2.1570849418640137, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7436790466308594, + "num_tokens": 452235750.0, + "step": 17727 + }, + { + "epoch": 1.9468482319349878, + "grad_norm": 2.2514283657073975, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7192555665969849, + "num_tokens": 452261181.0, + "step": 17728 + }, + { + "epoch": 1.9469580496376016, + "grad_norm": 2.2217893600463867, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7348408699035645, + "num_tokens": 452286787.0, + "step": 17729 + }, + { + "epoch": 1.9470678673402153, + "grad_norm": 2.8819010257720947, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7367085814476013, + "num_tokens": 452303245.0, + "step": 17730 + }, + { + "epoch": 1.947177685042829, + "grad_norm": 2.405135154724121, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7044656276702881, + "num_tokens": 452327799.0, + "step": 17731 + }, + { + "epoch": 1.9472875027454426, + "grad_norm": 2.4788818359375, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.730217456817627, + "num_tokens": 452349466.0, + "step": 17732 + }, + { + "epoch": 1.9473973204480561, + "grad_norm": 2.313441276550293, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7247300148010254, + "num_tokens": 452374845.0, + "step": 17733 + }, + { + "epoch": 1.94750713815067, + "grad_norm": 2.4041266441345215, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7688628435134888, + "num_tokens": 452395808.0, + "step": 17734 + }, + { + "epoch": 1.9476169558532836, + "grad_norm": 2.585073947906494, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.7611055374145508, + "num_tokens": 452413773.0, + "step": 17735 + }, + { + "epoch": 1.9477267735558972, + "grad_norm": 2.2178282737731934, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7127249240875244, + "num_tokens": 452438577.0, + "step": 17736 + }, + { + "epoch": 1.9478365912585107, + "grad_norm": 2.551516056060791, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7234506011009216, + "num_tokens": 452459985.0, + "step": 17737 + }, + { + "epoch": 1.9479464089611245, + "grad_norm": 1.9555346965789795, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.708017110824585, + "num_tokens": 452490602.0, + "step": 17738 + }, + { + "epoch": 1.9480562266637382, + "grad_norm": 2.285305976867676, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7253898978233337, + "num_tokens": 452515829.0, + "step": 17739 + }, + { + "epoch": 1.948166044366352, + "grad_norm": 2.4312937259674072, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7144376039505005, + "num_tokens": 452538146.0, + "step": 17740 + }, + { + "epoch": 1.9482758620689655, + "grad_norm": 2.16707706451416, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.734360933303833, + "num_tokens": 452562935.0, + "step": 17741 + }, + { + "epoch": 1.948385679771579, + "grad_norm": 2.1699917316436768, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7183218002319336, + "num_tokens": 452588823.0, + "step": 17742 + }, + { + "epoch": 1.9484954974741928, + "grad_norm": 2.407902479171753, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7425050735473633, + "num_tokens": 452610929.0, + "step": 17743 + }, + { + "epoch": 1.9486053151768066, + "grad_norm": 2.222440719604492, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7401163578033447, + "num_tokens": 452638938.0, + "step": 17744 + }, + { + "epoch": 1.94871513287942, + "grad_norm": 2.198413848876953, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7360092401504517, + "num_tokens": 452663042.0, + "step": 17745 + }, + { + "epoch": 1.9488249505820339, + "grad_norm": 2.152925968170166, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7283921241760254, + "num_tokens": 452690331.0, + "step": 17746 + }, + { + "epoch": 1.9489347682846474, + "grad_norm": 2.121384620666504, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7295945882797241, + "num_tokens": 452718295.0, + "step": 17747 + }, + { + "epoch": 1.9490445859872612, + "grad_norm": 2.3173282146453857, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7228025197982788, + "num_tokens": 452742118.0, + "step": 17748 + }, + { + "epoch": 1.949154403689875, + "grad_norm": 2.4518587589263916, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.729544997215271, + "num_tokens": 452763854.0, + "step": 17749 + }, + { + "epoch": 1.9492642213924884, + "grad_norm": 1.9199100732803345, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7372888922691345, + "num_tokens": 452796392.0, + "step": 17750 + }, + { + "epoch": 1.949374039095102, + "grad_norm": 2.223052501678467, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7247335910797119, + "num_tokens": 452823791.0, + "step": 17751 + }, + { + "epoch": 1.9494838567977157, + "grad_norm": 2.585157632827759, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7441216707229614, + "num_tokens": 452843236.0, + "step": 17752 + }, + { + "epoch": 1.9495936745003295, + "grad_norm": 2.0985212326049805, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7067450881004333, + "num_tokens": 452873852.0, + "step": 17753 + }, + { + "epoch": 1.9497034922029433, + "grad_norm": 2.4544663429260254, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7415422797203064, + "num_tokens": 452893970.0, + "step": 17754 + }, + { + "epoch": 1.9498133099055568, + "grad_norm": 2.169550895690918, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.71589595079422, + "num_tokens": 452919007.0, + "step": 17755 + }, + { + "epoch": 1.9499231276081703, + "grad_norm": 1.897752285003662, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7074738144874573, + "num_tokens": 452952976.0, + "step": 17756 + }, + { + "epoch": 1.950032945310784, + "grad_norm": 2.5613043308258057, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7439485788345337, + "num_tokens": 452971884.0, + "step": 17757 + }, + { + "epoch": 1.9501427630133978, + "grad_norm": 2.038377285003662, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.725225567817688, + "num_tokens": 453001733.0, + "step": 17758 + }, + { + "epoch": 1.9502525807160114, + "grad_norm": 2.3500306606292725, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7077154517173767, + "num_tokens": 453024568.0, + "step": 17759 + }, + { + "epoch": 1.9503623984186251, + "grad_norm": 2.2868902683258057, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6900405883789062, + "num_tokens": 453050719.0, + "step": 17760 + }, + { + "epoch": 1.9504722161212387, + "grad_norm": 2.4752604961395264, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7368417978286743, + "num_tokens": 453072187.0, + "step": 17761 + }, + { + "epoch": 1.9505820338238524, + "grad_norm": 2.2357211112976074, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7054470777511597, + "num_tokens": 453097069.0, + "step": 17762 + }, + { + "epoch": 1.9506918515264662, + "grad_norm": 1.9343876838684082, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7067326903343201, + "num_tokens": 453129532.0, + "step": 17763 + }, + { + "epoch": 1.9508016692290797, + "grad_norm": 2.162321090698242, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7194607257843018, + "num_tokens": 453157252.0, + "step": 17764 + }, + { + "epoch": 1.9509114869316933, + "grad_norm": 2.076333522796631, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7095273733139038, + "num_tokens": 453185363.0, + "step": 17765 + }, + { + "epoch": 1.951021304634307, + "grad_norm": 2.2305381298065186, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7424299716949463, + "num_tokens": 453210561.0, + "step": 17766 + }, + { + "epoch": 1.9511311223369208, + "grad_norm": 2.371181011199951, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7411346435546875, + "num_tokens": 453233535.0, + "step": 17767 + }, + { + "epoch": 1.9512409400395345, + "grad_norm": 2.5854599475860596, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7202138900756836, + "num_tokens": 453254644.0, + "step": 17768 + }, + { + "epoch": 1.951350757742148, + "grad_norm": 2.370936393737793, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7355071306228638, + "num_tokens": 453277528.0, + "step": 17769 + }, + { + "epoch": 1.9514605754447616, + "grad_norm": 2.1781105995178223, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7189961671829224, + "num_tokens": 453305206.0, + "step": 17770 + }, + { + "epoch": 1.9515703931473753, + "grad_norm": 2.38535475730896, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7232828140258789, + "num_tokens": 453329374.0, + "step": 17771 + }, + { + "epoch": 1.951680210849989, + "grad_norm": 1.9978832006454468, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.723724901676178, + "num_tokens": 453361774.0, + "step": 17772 + }, + { + "epoch": 1.9517900285526026, + "grad_norm": 2.3980343341827393, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7597284317016602, + "num_tokens": 453381976.0, + "step": 17773 + }, + { + "epoch": 1.9518998462552162, + "grad_norm": 2.0045735836029053, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7437576055526733, + "num_tokens": 453410347.0, + "step": 17774 + }, + { + "epoch": 1.95200966395783, + "grad_norm": 2.2212774753570557, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7283217906951904, + "num_tokens": 453435561.0, + "step": 17775 + }, + { + "epoch": 1.9521194816604437, + "grad_norm": 1.979385256767273, + "learning_rate": 1e-06, + "loss": 1.0217, + "mean_token_accuracy": 0.6888646483421326, + "num_tokens": 453467806.0, + "step": 17776 + }, + { + "epoch": 1.9522292993630574, + "grad_norm": 2.830414295196533, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7513218522071838, + "num_tokens": 453484545.0, + "step": 17777 + }, + { + "epoch": 1.952339117065671, + "grad_norm": 2.0296897888183594, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7403939962387085, + "num_tokens": 453514329.0, + "step": 17778 + }, + { + "epoch": 1.9524489347682845, + "grad_norm": 2.0870518684387207, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7233119010925293, + "num_tokens": 453541898.0, + "step": 17779 + }, + { + "epoch": 1.9525587524708983, + "grad_norm": 2.15812611579895, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7241624593734741, + "num_tokens": 453567812.0, + "step": 17780 + }, + { + "epoch": 1.952668570173512, + "grad_norm": 2.537928342819214, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7257338762283325, + "num_tokens": 453588981.0, + "step": 17781 + }, + { + "epoch": 1.9527783878761258, + "grad_norm": 2.2169017791748047, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7370465993881226, + "num_tokens": 453615492.0, + "step": 17782 + }, + { + "epoch": 1.9528882055787393, + "grad_norm": 2.558556079864502, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.7455729246139526, + "num_tokens": 453636689.0, + "step": 17783 + }, + { + "epoch": 1.9529980232813529, + "grad_norm": 2.0617282390594482, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7140448689460754, + "num_tokens": 453667906.0, + "step": 17784 + }, + { + "epoch": 1.9531078409839666, + "grad_norm": 2.4597420692443848, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.7624267339706421, + "num_tokens": 453688888.0, + "step": 17785 + }, + { + "epoch": 1.9532176586865804, + "grad_norm": 2.306093454360962, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7084024548530579, + "num_tokens": 453715094.0, + "step": 17786 + }, + { + "epoch": 1.953327476389194, + "grad_norm": 2.2483463287353516, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7402856945991516, + "num_tokens": 453740984.0, + "step": 17787 + }, + { + "epoch": 1.9534372940918074, + "grad_norm": 2.069716691970825, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7212080359458923, + "num_tokens": 453771023.0, + "step": 17788 + }, + { + "epoch": 1.9535471117944212, + "grad_norm": 2.018993377685547, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7112802267074585, + "num_tokens": 453801229.0, + "step": 17789 + }, + { + "epoch": 1.953656929497035, + "grad_norm": 2.028411388397217, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7053607702255249, + "num_tokens": 453830664.0, + "step": 17790 + }, + { + "epoch": 1.9537667471996487, + "grad_norm": 2.435467004776001, + "learning_rate": 1e-06, + "loss": 0.831, + "mean_token_accuracy": 0.7397589683532715, + "num_tokens": 453853237.0, + "step": 17791 + }, + { + "epoch": 1.9538765649022622, + "grad_norm": 2.1859612464904785, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.6989086866378784, + "num_tokens": 453881822.0, + "step": 17792 + }, + { + "epoch": 1.9539863826048758, + "grad_norm": 2.204176902770996, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7214996218681335, + "num_tokens": 453909439.0, + "step": 17793 + }, + { + "epoch": 1.9540962003074895, + "grad_norm": 2.252056121826172, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7398755550384521, + "num_tokens": 453931297.0, + "step": 17794 + }, + { + "epoch": 1.9542060180101033, + "grad_norm": 2.2972817420959473, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7405884861946106, + "num_tokens": 453955573.0, + "step": 17795 + }, + { + "epoch": 1.954315835712717, + "grad_norm": 1.9995920658111572, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7226633429527283, + "num_tokens": 453987199.0, + "step": 17796 + }, + { + "epoch": 1.9544256534153306, + "grad_norm": 2.1587836742401123, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.7030943632125854, + "num_tokens": 454015560.0, + "step": 17797 + }, + { + "epoch": 1.9545354711179441, + "grad_norm": 2.4494011402130127, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.739446759223938, + "num_tokens": 454038172.0, + "step": 17798 + }, + { + "epoch": 1.9546452888205579, + "grad_norm": 2.279757022857666, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7265968322753906, + "num_tokens": 454062651.0, + "step": 17799 + }, + { + "epoch": 1.9547551065231716, + "grad_norm": 2.0804972648620605, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.6917977333068848, + "num_tokens": 454092561.0, + "step": 17800 + }, + { + "epoch": 1.9548649242257852, + "grad_norm": 2.2681987285614014, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7127243280410767, + "num_tokens": 454119550.0, + "step": 17801 + }, + { + "epoch": 1.9549747419283987, + "grad_norm": 2.2621943950653076, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7059611082077026, + "num_tokens": 454146932.0, + "step": 17802 + }, + { + "epoch": 1.9550845596310125, + "grad_norm": 2.348468780517578, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7402198910713196, + "num_tokens": 454170564.0, + "step": 17803 + }, + { + "epoch": 1.9551943773336262, + "grad_norm": 2.2217869758605957, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7102255821228027, + "num_tokens": 454196919.0, + "step": 17804 + }, + { + "epoch": 1.95530419503624, + "grad_norm": 2.2787764072418213, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7153807878494263, + "num_tokens": 454224089.0, + "step": 17805 + }, + { + "epoch": 1.9554140127388535, + "grad_norm": 2.2715134620666504, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.6998232007026672, + "num_tokens": 454250898.0, + "step": 17806 + }, + { + "epoch": 1.955523830441467, + "grad_norm": 2.1378962993621826, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.6982458829879761, + "num_tokens": 454281175.0, + "step": 17807 + }, + { + "epoch": 1.9556336481440808, + "grad_norm": 2.4020566940307617, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7321563363075256, + "num_tokens": 454304512.0, + "step": 17808 + }, + { + "epoch": 1.9557434658466946, + "grad_norm": 2.2154033184051514, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7334924936294556, + "num_tokens": 454329911.0, + "step": 17809 + }, + { + "epoch": 1.955853283549308, + "grad_norm": 2.206841230392456, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7065297365188599, + "num_tokens": 454357450.0, + "step": 17810 + }, + { + "epoch": 1.9559631012519219, + "grad_norm": 2.631021499633789, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7250531911849976, + "num_tokens": 454376589.0, + "step": 17811 + }, + { + "epoch": 1.9560729189545354, + "grad_norm": 2.1023454666137695, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7100120782852173, + "num_tokens": 454404895.0, + "step": 17812 + }, + { + "epoch": 1.9561827366571491, + "grad_norm": 2.2322921752929688, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.6972833871841431, + "num_tokens": 454432955.0, + "step": 17813 + }, + { + "epoch": 1.956292554359763, + "grad_norm": 2.1471495628356934, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7176405191421509, + "num_tokens": 454459227.0, + "step": 17814 + }, + { + "epoch": 1.9564023720623764, + "grad_norm": 2.373650312423706, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7232527136802673, + "num_tokens": 454481954.0, + "step": 17815 + }, + { + "epoch": 1.95651218976499, + "grad_norm": 2.2687230110168457, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.6974649429321289, + "num_tokens": 454507736.0, + "step": 17816 + }, + { + "epoch": 1.9566220074676037, + "grad_norm": 2.192399024963379, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7235002517700195, + "num_tokens": 454533063.0, + "step": 17817 + }, + { + "epoch": 1.9567318251702175, + "grad_norm": 2.476367712020874, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7307003736495972, + "num_tokens": 454556778.0, + "step": 17818 + }, + { + "epoch": 1.9568416428728312, + "grad_norm": 2.247335433959961, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7266359329223633, + "num_tokens": 454581467.0, + "step": 17819 + }, + { + "epoch": 1.9569514605754448, + "grad_norm": 2.0362985134124756, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.703711986541748, + "num_tokens": 454615365.0, + "step": 17820 + }, + { + "epoch": 1.9570612782780583, + "grad_norm": 2.568437337875366, + "learning_rate": 1e-06, + "loss": 0.7709, + "mean_token_accuracy": 0.7520992755889893, + "num_tokens": 454634877.0, + "step": 17821 + }, + { + "epoch": 1.957171095980672, + "grad_norm": 2.350929021835327, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7233045101165771, + "num_tokens": 454657503.0, + "step": 17822 + }, + { + "epoch": 1.9572809136832858, + "grad_norm": 2.5432307720184326, + "learning_rate": 1e-06, + "loss": 0.7617, + "mean_token_accuracy": 0.7486363649368286, + "num_tokens": 454676798.0, + "step": 17823 + }, + { + "epoch": 1.9573907313858994, + "grad_norm": 2.423975944519043, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7440003752708435, + "num_tokens": 454698092.0, + "step": 17824 + }, + { + "epoch": 1.9575005490885131, + "grad_norm": 2.463242530822754, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7200656533241272, + "num_tokens": 454719000.0, + "step": 17825 + }, + { + "epoch": 1.9576103667911267, + "grad_norm": 2.514008045196533, + "learning_rate": 1e-06, + "loss": 0.8213, + "mean_token_accuracy": 0.7433059215545654, + "num_tokens": 454739379.0, + "step": 17826 + }, + { + "epoch": 1.9577201844937404, + "grad_norm": 2.147373676300049, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7362325191497803, + "num_tokens": 454765381.0, + "step": 17827 + }, + { + "epoch": 1.9578300021963542, + "grad_norm": 2.4865849018096924, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.6998088359832764, + "num_tokens": 454788032.0, + "step": 17828 + }, + { + "epoch": 1.9579398198989677, + "grad_norm": 2.0836234092712402, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7056481838226318, + "num_tokens": 454816509.0, + "step": 17829 + }, + { + "epoch": 1.9580496376015812, + "grad_norm": 2.3822343349456787, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7353473901748657, + "num_tokens": 454839098.0, + "step": 17830 + }, + { + "epoch": 1.958159455304195, + "grad_norm": 2.3014087677001953, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.703656792640686, + "num_tokens": 454865205.0, + "step": 17831 + }, + { + "epoch": 1.9582692730068088, + "grad_norm": 2.2190496921539307, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7359958291053772, + "num_tokens": 454888728.0, + "step": 17832 + }, + { + "epoch": 1.9583790907094225, + "grad_norm": 1.9220681190490723, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7227015495300293, + "num_tokens": 454922735.0, + "step": 17833 + }, + { + "epoch": 1.958488908412036, + "grad_norm": 2.145385265350342, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7348915934562683, + "num_tokens": 454949218.0, + "step": 17834 + }, + { + "epoch": 1.9585987261146496, + "grad_norm": 2.1180965900421143, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7170518636703491, + "num_tokens": 454976481.0, + "step": 17835 + }, + { + "epoch": 1.9587085438172633, + "grad_norm": 2.338693380355835, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7257659435272217, + "num_tokens": 455000968.0, + "step": 17836 + }, + { + "epoch": 1.958818361519877, + "grad_norm": 2.0452613830566406, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7119909524917603, + "num_tokens": 455030925.0, + "step": 17837 + }, + { + "epoch": 1.9589281792224906, + "grad_norm": 2.0467400550842285, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7129371166229248, + "num_tokens": 455060484.0, + "step": 17838 + }, + { + "epoch": 1.9590379969251042, + "grad_norm": 2.198925495147705, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.727491557598114, + "num_tokens": 455086316.0, + "step": 17839 + }, + { + "epoch": 1.959147814627718, + "grad_norm": 2.0744171142578125, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7272799015045166, + "num_tokens": 455112163.0, + "step": 17840 + }, + { + "epoch": 1.9592576323303317, + "grad_norm": 2.0653066635131836, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7419784665107727, + "num_tokens": 455140505.0, + "step": 17841 + }, + { + "epoch": 1.9593674500329454, + "grad_norm": 2.127293109893799, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7110906839370728, + "num_tokens": 455169601.0, + "step": 17842 + }, + { + "epoch": 1.959477267735559, + "grad_norm": 2.2869203090667725, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7164708375930786, + "num_tokens": 455194873.0, + "step": 17843 + }, + { + "epoch": 1.9595870854381725, + "grad_norm": 2.2282018661499023, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7106310129165649, + "num_tokens": 455221266.0, + "step": 17844 + }, + { + "epoch": 1.9596969031407863, + "grad_norm": 2.0647220611572266, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.709175705909729, + "num_tokens": 455249344.0, + "step": 17845 + }, + { + "epoch": 1.9598067208434, + "grad_norm": 2.339956760406494, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7080180048942566, + "num_tokens": 455272924.0, + "step": 17846 + }, + { + "epoch": 1.9599165385460138, + "grad_norm": 2.629497766494751, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7258720993995667, + "num_tokens": 455292313.0, + "step": 17847 + }, + { + "epoch": 1.9600263562486273, + "grad_norm": 2.4377853870391846, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7125974893569946, + "num_tokens": 455315758.0, + "step": 17848 + }, + { + "epoch": 1.9601361739512408, + "grad_norm": 2.334336519241333, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7294582724571228, + "num_tokens": 455339544.0, + "step": 17849 + }, + { + "epoch": 1.9602459916538546, + "grad_norm": 2.555018424987793, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7401848435401917, + "num_tokens": 455359859.0, + "step": 17850 + }, + { + "epoch": 1.9603558093564684, + "grad_norm": 2.294579267501831, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7261258363723755, + "num_tokens": 455384155.0, + "step": 17851 + }, + { + "epoch": 1.960465627059082, + "grad_norm": 2.0637731552124023, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.6856223344802856, + "num_tokens": 455414680.0, + "step": 17852 + }, + { + "epoch": 1.9605754447616954, + "grad_norm": 2.412440061569214, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7163620591163635, + "num_tokens": 455436970.0, + "step": 17853 + }, + { + "epoch": 1.9606852624643092, + "grad_norm": 2.281111001968384, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7182859778404236, + "num_tokens": 455461454.0, + "step": 17854 + }, + { + "epoch": 1.960795080166923, + "grad_norm": 2.286424398422241, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7151965498924255, + "num_tokens": 455486661.0, + "step": 17855 + }, + { + "epoch": 1.9609048978695367, + "grad_norm": 2.2384188175201416, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7172049880027771, + "num_tokens": 455511922.0, + "step": 17856 + }, + { + "epoch": 1.9610147155721502, + "grad_norm": 2.0114660263061523, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7027566432952881, + "num_tokens": 455543401.0, + "step": 17857 + }, + { + "epoch": 1.9611245332747638, + "grad_norm": 2.217660665512085, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7406661510467529, + "num_tokens": 455569730.0, + "step": 17858 + }, + { + "epoch": 1.9612343509773775, + "grad_norm": 2.316667318344116, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7117912769317627, + "num_tokens": 455595883.0, + "step": 17859 + }, + { + "epoch": 1.9613441686799913, + "grad_norm": 2.563685178756714, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7204432487487793, + "num_tokens": 455618561.0, + "step": 17860 + }, + { + "epoch": 1.961453986382605, + "grad_norm": 2.1886558532714844, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.724098265171051, + "num_tokens": 455644864.0, + "step": 17861 + }, + { + "epoch": 1.9615638040852186, + "grad_norm": 2.0654373168945312, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.712450385093689, + "num_tokens": 455673181.0, + "step": 17862 + }, + { + "epoch": 1.9616736217878321, + "grad_norm": 2.269097328186035, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7179109454154968, + "num_tokens": 455699379.0, + "step": 17863 + }, + { + "epoch": 1.9617834394904459, + "grad_norm": 2.142178535461426, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7039886713027954, + "num_tokens": 455728427.0, + "step": 17864 + }, + { + "epoch": 1.9618932571930596, + "grad_norm": 2.4165546894073486, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.7065421342849731, + "num_tokens": 455752273.0, + "step": 17865 + }, + { + "epoch": 1.9620030748956732, + "grad_norm": 2.2749032974243164, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7040748596191406, + "num_tokens": 455778731.0, + "step": 17866 + }, + { + "epoch": 1.9621128925982867, + "grad_norm": 2.0924336910247803, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7205735445022583, + "num_tokens": 455804789.0, + "step": 17867 + }, + { + "epoch": 1.9622227103009005, + "grad_norm": 2.1355972290039062, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.702311635017395, + "num_tokens": 455831798.0, + "step": 17868 + }, + { + "epoch": 1.9623325280035142, + "grad_norm": 2.5878233909606934, + "learning_rate": 1e-06, + "loss": 0.7862, + "mean_token_accuracy": 0.7452282309532166, + "num_tokens": 455851755.0, + "step": 17869 + }, + { + "epoch": 1.962442345706128, + "grad_norm": 1.9146636724472046, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7172365188598633, + "num_tokens": 455884760.0, + "step": 17870 + }, + { + "epoch": 1.9625521634087415, + "grad_norm": 2.3404603004455566, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7230619192123413, + "num_tokens": 455907648.0, + "step": 17871 + }, + { + "epoch": 1.962661981111355, + "grad_norm": 2.13904070854187, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7302241325378418, + "num_tokens": 455934654.0, + "step": 17872 + }, + { + "epoch": 1.9627717988139688, + "grad_norm": 2.1308486461639404, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7276815176010132, + "num_tokens": 455963823.0, + "step": 17873 + }, + { + "epoch": 1.9628816165165826, + "grad_norm": 2.2495675086975098, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7062802910804749, + "num_tokens": 455990324.0, + "step": 17874 + }, + { + "epoch": 1.962991434219196, + "grad_norm": 2.336671829223633, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.71546471118927, + "num_tokens": 456016104.0, + "step": 17875 + }, + { + "epoch": 1.9631012519218098, + "grad_norm": 2.9291653633117676, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7319114804267883, + "num_tokens": 456034310.0, + "step": 17876 + }, + { + "epoch": 1.9632110696244234, + "grad_norm": 2.615671396255493, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.750141441822052, + "num_tokens": 456055154.0, + "step": 17877 + }, + { + "epoch": 1.9633208873270371, + "grad_norm": 2.465693950653076, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7105978727340698, + "num_tokens": 456079660.0, + "step": 17878 + }, + { + "epoch": 1.963430705029651, + "grad_norm": 2.662813901901245, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.730482816696167, + "num_tokens": 456098091.0, + "step": 17879 + }, + { + "epoch": 1.9635405227322644, + "grad_norm": 2.504664182662964, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7239158749580383, + "num_tokens": 456119907.0, + "step": 17880 + }, + { + "epoch": 1.963650340434878, + "grad_norm": 2.3530373573303223, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7360145449638367, + "num_tokens": 456144003.0, + "step": 17881 + }, + { + "epoch": 1.9637601581374917, + "grad_norm": 2.1429684162139893, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7231779098510742, + "num_tokens": 456172700.0, + "step": 17882 + }, + { + "epoch": 1.9638699758401055, + "grad_norm": 2.0912225246429443, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7266706228256226, + "num_tokens": 456200423.0, + "step": 17883 + }, + { + "epoch": 1.9639797935427192, + "grad_norm": 2.182358980178833, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7324486970901489, + "num_tokens": 456225336.0, + "step": 17884 + }, + { + "epoch": 1.9640896112453328, + "grad_norm": 2.145845413208008, + "learning_rate": 1e-06, + "loss": 1.012, + "mean_token_accuracy": 0.6927353143692017, + "num_tokens": 456253649.0, + "step": 17885 + }, + { + "epoch": 1.9641994289479463, + "grad_norm": 2.482600688934326, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7303733229637146, + "num_tokens": 456275312.0, + "step": 17886 + }, + { + "epoch": 1.96430924665056, + "grad_norm": 2.5597102642059326, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7118972539901733, + "num_tokens": 456296676.0, + "step": 17887 + }, + { + "epoch": 1.9644190643531738, + "grad_norm": 2.4988925457000732, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7500464916229248, + "num_tokens": 456316781.0, + "step": 17888 + }, + { + "epoch": 1.9645288820557874, + "grad_norm": 2.230687141418457, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7151110172271729, + "num_tokens": 456341449.0, + "step": 17889 + }, + { + "epoch": 1.964638699758401, + "grad_norm": 2.4788613319396973, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7262913584709167, + "num_tokens": 456362154.0, + "step": 17890 + }, + { + "epoch": 1.9647485174610146, + "grad_norm": 2.5364925861358643, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7314717769622803, + "num_tokens": 456384103.0, + "step": 17891 + }, + { + "epoch": 1.9648583351636284, + "grad_norm": 2.36309814453125, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7229037880897522, + "num_tokens": 456407319.0, + "step": 17892 + }, + { + "epoch": 1.9649681528662422, + "grad_norm": 2.346738576889038, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7092986106872559, + "num_tokens": 456431449.0, + "step": 17893 + }, + { + "epoch": 1.9650779705688557, + "grad_norm": 2.292996644973755, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7128230929374695, + "num_tokens": 456455263.0, + "step": 17894 + }, + { + "epoch": 1.9651877882714692, + "grad_norm": 2.2077484130859375, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7172322273254395, + "num_tokens": 456480275.0, + "step": 17895 + }, + { + "epoch": 1.965297605974083, + "grad_norm": 2.433978319168091, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7289060354232788, + "num_tokens": 456503339.0, + "step": 17896 + }, + { + "epoch": 1.9654074236766967, + "grad_norm": 2.3530776500701904, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7360351085662842, + "num_tokens": 456527780.0, + "step": 17897 + }, + { + "epoch": 1.9655172413793105, + "grad_norm": 2.2209274768829346, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7148601412773132, + "num_tokens": 456553671.0, + "step": 17898 + }, + { + "epoch": 1.965627059081924, + "grad_norm": 2.424038887023926, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.7475972771644592, + "num_tokens": 456575224.0, + "step": 17899 + }, + { + "epoch": 1.9657368767845376, + "grad_norm": 2.4708287715911865, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7070745825767517, + "num_tokens": 456597386.0, + "step": 17900 + }, + { + "epoch": 1.9658466944871513, + "grad_norm": 2.2011475563049316, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7246838212013245, + "num_tokens": 456624498.0, + "step": 17901 + }, + { + "epoch": 1.965956512189765, + "grad_norm": 2.2042887210845947, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7346925139427185, + "num_tokens": 456650066.0, + "step": 17902 + }, + { + "epoch": 1.9660663298923786, + "grad_norm": 2.270085334777832, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7234244346618652, + "num_tokens": 456675141.0, + "step": 17903 + }, + { + "epoch": 1.9661761475949922, + "grad_norm": 2.581629991531372, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7131432294845581, + "num_tokens": 456696542.0, + "step": 17904 + }, + { + "epoch": 1.966285965297606, + "grad_norm": 1.8615193367004395, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6900360584259033, + "num_tokens": 456733805.0, + "step": 17905 + }, + { + "epoch": 1.9663957830002197, + "grad_norm": 2.2162091732025146, + "learning_rate": 1e-06, + "loss": 0.7671, + "mean_token_accuracy": 0.7579464912414551, + "num_tokens": 456757991.0, + "step": 17906 + }, + { + "epoch": 1.9665056007028334, + "grad_norm": 2.3098678588867188, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7415142059326172, + "num_tokens": 456781025.0, + "step": 17907 + }, + { + "epoch": 1.966615418405447, + "grad_norm": 2.2772057056427, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.720403254032135, + "num_tokens": 456805740.0, + "step": 17908 + }, + { + "epoch": 1.9667252361080605, + "grad_norm": 2.068117380142212, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7145922780036926, + "num_tokens": 456833546.0, + "step": 17909 + }, + { + "epoch": 1.9668350538106742, + "grad_norm": 1.9451627731323242, + "learning_rate": 1e-06, + "loss": 1.0307, + "mean_token_accuracy": 0.6868940591812134, + "num_tokens": 456868618.0, + "step": 17910 + }, + { + "epoch": 1.966944871513288, + "grad_norm": 2.1771602630615234, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.741975724697113, + "num_tokens": 456894379.0, + "step": 17911 + }, + { + "epoch": 1.9670546892159018, + "grad_norm": 2.220947504043579, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7151095867156982, + "num_tokens": 456920548.0, + "step": 17912 + }, + { + "epoch": 1.9671645069185153, + "grad_norm": 2.034623146057129, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.763852596282959, + "num_tokens": 456945129.0, + "step": 17913 + }, + { + "epoch": 1.9672743246211288, + "grad_norm": 2.2466485500335693, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7359377145767212, + "num_tokens": 456970345.0, + "step": 17914 + }, + { + "epoch": 1.9673841423237426, + "grad_norm": 2.2680823802948, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7061537504196167, + "num_tokens": 456994923.0, + "step": 17915 + }, + { + "epoch": 1.9674939600263563, + "grad_norm": 2.214630126953125, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7214467525482178, + "num_tokens": 457020850.0, + "step": 17916 + }, + { + "epoch": 1.9676037777289699, + "grad_norm": 2.319483757019043, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7283497452735901, + "num_tokens": 457045307.0, + "step": 17917 + }, + { + "epoch": 1.9677135954315834, + "grad_norm": 2.1332013607025146, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7076617479324341, + "num_tokens": 457073325.0, + "step": 17918 + }, + { + "epoch": 1.9678234131341972, + "grad_norm": 2.245619773864746, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7382023334503174, + "num_tokens": 457096453.0, + "step": 17919 + }, + { + "epoch": 1.967933230836811, + "grad_norm": 2.4970247745513916, + "learning_rate": 1e-06, + "loss": 0.8057, + "mean_token_accuracy": 0.7432761788368225, + "num_tokens": 457117776.0, + "step": 17920 + }, + { + "epoch": 1.9680430485394247, + "grad_norm": 2.082747459411621, + "learning_rate": 1e-06, + "loss": 0.7384, + "mean_token_accuracy": 0.7618650197982788, + "num_tokens": 457143095.0, + "step": 17921 + }, + { + "epoch": 1.9681528662420382, + "grad_norm": 2.4769933223724365, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7341430187225342, + "num_tokens": 457163575.0, + "step": 17922 + }, + { + "epoch": 1.9682626839446518, + "grad_norm": 1.9884390830993652, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7175729274749756, + "num_tokens": 457193812.0, + "step": 17923 + }, + { + "epoch": 1.9683725016472655, + "grad_norm": 2.2192087173461914, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7250051498413086, + "num_tokens": 457221889.0, + "step": 17924 + }, + { + "epoch": 1.9684823193498793, + "grad_norm": 2.2597858905792236, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7363358736038208, + "num_tokens": 457243967.0, + "step": 17925 + }, + { + "epoch": 1.9685921370524928, + "grad_norm": 2.3023688793182373, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7449201345443726, + "num_tokens": 457267281.0, + "step": 17926 + }, + { + "epoch": 1.9687019547551066, + "grad_norm": 2.1948177814483643, + "learning_rate": 1e-06, + "loss": 1.0308, + "mean_token_accuracy": 0.6890918612480164, + "num_tokens": 457296348.0, + "step": 17927 + }, + { + "epoch": 1.96881177245772, + "grad_norm": 2.27439284324646, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7253303527832031, + "num_tokens": 457321303.0, + "step": 17928 + }, + { + "epoch": 1.9689215901603339, + "grad_norm": 1.9507216215133667, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7037110924720764, + "num_tokens": 457353358.0, + "step": 17929 + }, + { + "epoch": 1.9690314078629476, + "grad_norm": 2.522462844848633, + "learning_rate": 1e-06, + "loss": 0.7528, + "mean_token_accuracy": 0.7597633600234985, + "num_tokens": 457373542.0, + "step": 17930 + }, + { + "epoch": 1.9691412255655611, + "grad_norm": 2.259935140609741, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7205420136451721, + "num_tokens": 457398684.0, + "step": 17931 + }, + { + "epoch": 1.9692510432681747, + "grad_norm": 2.286292791366577, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.70307457447052, + "num_tokens": 457424720.0, + "step": 17932 + }, + { + "epoch": 1.9693608609707884, + "grad_norm": 2.740129232406616, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7428306341171265, + "num_tokens": 457444899.0, + "step": 17933 + }, + { + "epoch": 1.9694706786734022, + "grad_norm": 2.0066757202148438, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7065907716751099, + "num_tokens": 457474877.0, + "step": 17934 + }, + { + "epoch": 1.969580496376016, + "grad_norm": 2.253821849822998, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7106987833976746, + "num_tokens": 457500295.0, + "step": 17935 + }, + { + "epoch": 1.9696903140786295, + "grad_norm": 2.081292152404785, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7192476987838745, + "num_tokens": 457529092.0, + "step": 17936 + }, + { + "epoch": 1.969800131781243, + "grad_norm": 1.9972666501998901, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7352259159088135, + "num_tokens": 457558906.0, + "step": 17937 + }, + { + "epoch": 1.9699099494838568, + "grad_norm": 2.341385841369629, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7220826148986816, + "num_tokens": 457582531.0, + "step": 17938 + }, + { + "epoch": 1.9700197671864705, + "grad_norm": 2.6842803955078125, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7381687164306641, + "num_tokens": 457601567.0, + "step": 17939 + }, + { + "epoch": 1.970129584889084, + "grad_norm": 2.3516602516174316, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7216512560844421, + "num_tokens": 457624904.0, + "step": 17940 + }, + { + "epoch": 1.9702394025916978, + "grad_norm": 2.1351442337036133, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7268446683883667, + "num_tokens": 457651934.0, + "step": 17941 + }, + { + "epoch": 1.9703492202943114, + "grad_norm": 2.3296337127685547, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7491039633750916, + "num_tokens": 457673625.0, + "step": 17942 + }, + { + "epoch": 1.9704590379969251, + "grad_norm": 2.3296661376953125, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.742620587348938, + "num_tokens": 457696317.0, + "step": 17943 + }, + { + "epoch": 1.9705688556995389, + "grad_norm": 2.337181568145752, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7259063124656677, + "num_tokens": 457720652.0, + "step": 17944 + }, + { + "epoch": 1.9706786734021524, + "grad_norm": 2.088336706161499, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7200359106063843, + "num_tokens": 457750394.0, + "step": 17945 + }, + { + "epoch": 1.970788491104766, + "grad_norm": 2.521836042404175, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7178409099578857, + "num_tokens": 457773244.0, + "step": 17946 + }, + { + "epoch": 1.9708983088073797, + "grad_norm": 2.1343202590942383, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7078670263290405, + "num_tokens": 457801762.0, + "step": 17947 + }, + { + "epoch": 1.9710081265099935, + "grad_norm": 2.241708517074585, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7205321788787842, + "num_tokens": 457827495.0, + "step": 17948 + }, + { + "epoch": 1.9711179442126072, + "grad_norm": 1.9557216167449951, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7044844627380371, + "num_tokens": 457859169.0, + "step": 17949 + }, + { + "epoch": 1.9712277619152208, + "grad_norm": 2.140317678451538, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7103504538536072, + "num_tokens": 457885850.0, + "step": 17950 + }, + { + "epoch": 1.9713375796178343, + "grad_norm": 2.4400765895843506, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.738641619682312, + "num_tokens": 457907274.0, + "step": 17951 + }, + { + "epoch": 1.971447397320448, + "grad_norm": 2.1098766326904297, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.713646411895752, + "num_tokens": 457935607.0, + "step": 17952 + }, + { + "epoch": 1.9715572150230618, + "grad_norm": 2.1438355445861816, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7175958752632141, + "num_tokens": 457962896.0, + "step": 17953 + }, + { + "epoch": 1.9716670327256753, + "grad_norm": 2.0041117668151855, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7223837375640869, + "num_tokens": 457992635.0, + "step": 17954 + }, + { + "epoch": 1.9717768504282889, + "grad_norm": 2.219682216644287, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7247976064682007, + "num_tokens": 458018333.0, + "step": 17955 + }, + { + "epoch": 1.9718866681309026, + "grad_norm": 2.1197752952575684, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7155884504318237, + "num_tokens": 458047724.0, + "step": 17956 + }, + { + "epoch": 1.9719964858335164, + "grad_norm": 2.142340898513794, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7135348320007324, + "num_tokens": 458073589.0, + "step": 17957 + }, + { + "epoch": 1.9721063035361301, + "grad_norm": 2.1611742973327637, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7056317329406738, + "num_tokens": 458100459.0, + "step": 17958 + }, + { + "epoch": 1.9722161212387437, + "grad_norm": 2.229900598526001, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7167476415634155, + "num_tokens": 458126758.0, + "step": 17959 + }, + { + "epoch": 1.9723259389413572, + "grad_norm": 2.233628273010254, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.690642774105072, + "num_tokens": 458155350.0, + "step": 17960 + }, + { + "epoch": 1.972435756643971, + "grad_norm": 2.209974527359009, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7235654592514038, + "num_tokens": 458182330.0, + "step": 17961 + }, + { + "epoch": 1.9725455743465847, + "grad_norm": 2.048226833343506, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.699066162109375, + "num_tokens": 458214670.0, + "step": 17962 + }, + { + "epoch": 1.9726553920491985, + "grad_norm": 2.1452019214630127, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7270846366882324, + "num_tokens": 458241446.0, + "step": 17963 + }, + { + "epoch": 1.972765209751812, + "grad_norm": 2.196488857269287, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7146179676055908, + "num_tokens": 458265826.0, + "step": 17964 + }, + { + "epoch": 1.9728750274544256, + "grad_norm": 2.2066869735717773, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7147027850151062, + "num_tokens": 458292498.0, + "step": 17965 + }, + { + "epoch": 1.9729848451570393, + "grad_norm": 2.0698082447052, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7066102027893066, + "num_tokens": 458320664.0, + "step": 17966 + }, + { + "epoch": 1.973094662859653, + "grad_norm": 2.4609947204589844, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7188643217086792, + "num_tokens": 458342764.0, + "step": 17967 + }, + { + "epoch": 1.9732044805622666, + "grad_norm": 1.953026294708252, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7215088605880737, + "num_tokens": 458374017.0, + "step": 17968 + }, + { + "epoch": 1.9733142982648801, + "grad_norm": 2.3230555057525635, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7286677956581116, + "num_tokens": 458398156.0, + "step": 17969 + }, + { + "epoch": 1.973424115967494, + "grad_norm": 2.519960641860962, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7198755741119385, + "num_tokens": 458419311.0, + "step": 17970 + }, + { + "epoch": 1.9735339336701077, + "grad_norm": 2.2033138275146484, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7414054870605469, + "num_tokens": 458446403.0, + "step": 17971 + }, + { + "epoch": 1.9736437513727214, + "grad_norm": 2.325981855392456, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7213177680969238, + "num_tokens": 458470636.0, + "step": 17972 + }, + { + "epoch": 1.973753569075335, + "grad_norm": 2.5038001537323, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7184900641441345, + "num_tokens": 458490733.0, + "step": 17973 + }, + { + "epoch": 1.9738633867779485, + "grad_norm": 1.9549707174301147, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7389670014381409, + "num_tokens": 458519446.0, + "step": 17974 + }, + { + "epoch": 1.9739732044805622, + "grad_norm": 2.3562190532684326, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7465915679931641, + "num_tokens": 458542537.0, + "step": 17975 + }, + { + "epoch": 1.974083022183176, + "grad_norm": 2.2242510318756104, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7201499342918396, + "num_tokens": 458567707.0, + "step": 17976 + }, + { + "epoch": 1.9741928398857898, + "grad_norm": 2.47011399269104, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.710396409034729, + "num_tokens": 458590284.0, + "step": 17977 + }, + { + "epoch": 1.9743026575884033, + "grad_norm": 2.4482107162475586, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7397423982620239, + "num_tokens": 458612514.0, + "step": 17978 + }, + { + "epoch": 1.9744124752910168, + "grad_norm": 2.4631967544555664, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.750880241394043, + "num_tokens": 458632623.0, + "step": 17979 + }, + { + "epoch": 1.9745222929936306, + "grad_norm": 2.1889116764068604, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7135059833526611, + "num_tokens": 458659232.0, + "step": 17980 + }, + { + "epoch": 1.9746321106962443, + "grad_norm": 2.2361268997192383, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7103649377822876, + "num_tokens": 458685112.0, + "step": 17981 + }, + { + "epoch": 1.9747419283988579, + "grad_norm": 2.3976504802703857, + "learning_rate": 1e-06, + "loss": 0.8264, + "mean_token_accuracy": 0.7403357028961182, + "num_tokens": 458707434.0, + "step": 17982 + }, + { + "epoch": 1.9748517461014714, + "grad_norm": 2.382007360458374, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7413303256034851, + "num_tokens": 458729365.0, + "step": 17983 + }, + { + "epoch": 1.9749615638040852, + "grad_norm": 2.2134768962860107, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6936684250831604, + "num_tokens": 458755609.0, + "step": 17984 + }, + { + "epoch": 1.975071381506699, + "grad_norm": 2.141425609588623, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7270462512969971, + "num_tokens": 458782017.0, + "step": 17985 + }, + { + "epoch": 1.9751811992093127, + "grad_norm": 2.2859623432159424, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7144457101821899, + "num_tokens": 458807003.0, + "step": 17986 + }, + { + "epoch": 1.9752910169119262, + "grad_norm": 2.0833747386932373, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.6990480422973633, + "num_tokens": 458835607.0, + "step": 17987 + }, + { + "epoch": 1.9754008346145397, + "grad_norm": 2.116001605987549, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7253516912460327, + "num_tokens": 458861193.0, + "step": 17988 + }, + { + "epoch": 1.9755106523171535, + "grad_norm": 2.260514497756958, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7447382211685181, + "num_tokens": 458883953.0, + "step": 17989 + }, + { + "epoch": 1.9756204700197673, + "grad_norm": 2.549023151397705, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7383097410202026, + "num_tokens": 458904425.0, + "step": 17990 + }, + { + "epoch": 1.9757302877223808, + "grad_norm": 2.5009937286376953, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7194696664810181, + "num_tokens": 458925953.0, + "step": 17991 + }, + { + "epoch": 1.9758401054249946, + "grad_norm": 2.437971353530884, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7276210784912109, + "num_tokens": 458947894.0, + "step": 17992 + }, + { + "epoch": 1.975949923127608, + "grad_norm": 2.610530138015747, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7349891662597656, + "num_tokens": 458967920.0, + "step": 17993 + }, + { + "epoch": 1.9760597408302218, + "grad_norm": 2.2664120197296143, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7048883438110352, + "num_tokens": 458994362.0, + "step": 17994 + }, + { + "epoch": 1.9761695585328356, + "grad_norm": 2.4163124561309814, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7130599021911621, + "num_tokens": 459017290.0, + "step": 17995 + }, + { + "epoch": 1.9762793762354491, + "grad_norm": 2.095430850982666, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.734542191028595, + "num_tokens": 459045316.0, + "step": 17996 + }, + { + "epoch": 1.9763891939380627, + "grad_norm": 2.1558423042297363, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7195422649383545, + "num_tokens": 459071325.0, + "step": 17997 + }, + { + "epoch": 1.9764990116406764, + "grad_norm": 2.081024169921875, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7051205635070801, + "num_tokens": 459100759.0, + "step": 17998 + }, + { + "epoch": 1.9766088293432902, + "grad_norm": 2.1417977809906006, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7323952913284302, + "num_tokens": 459127010.0, + "step": 17999 + }, + { + "epoch": 1.976718647045904, + "grad_norm": 2.473471164703369, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7215098142623901, + "num_tokens": 459147986.0, + "step": 18000 + }, + { + "epoch": 1.9768284647485175, + "grad_norm": 2.4131171703338623, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7060422301292419, + "num_tokens": 459172106.0, + "step": 18001 + }, + { + "epoch": 1.976938282451131, + "grad_norm": 2.2591700553894043, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7300558090209961, + "num_tokens": 459195692.0, + "step": 18002 + }, + { + "epoch": 1.9770481001537448, + "grad_norm": 2.2819225788116455, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7137326002120972, + "num_tokens": 459218608.0, + "step": 18003 + }, + { + "epoch": 1.9771579178563585, + "grad_norm": 2.24149751663208, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7029761075973511, + "num_tokens": 459245393.0, + "step": 18004 + }, + { + "epoch": 1.977267735558972, + "grad_norm": 2.2757675647735596, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7333348989486694, + "num_tokens": 459271944.0, + "step": 18005 + }, + { + "epoch": 1.9773775532615858, + "grad_norm": 2.510721445083618, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7312172651290894, + "num_tokens": 459294410.0, + "step": 18006 + }, + { + "epoch": 1.9774873709641994, + "grad_norm": 2.2129390239715576, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7181946635246277, + "num_tokens": 459321288.0, + "step": 18007 + }, + { + "epoch": 1.977597188666813, + "grad_norm": 2.016399621963501, + "learning_rate": 1e-06, + "loss": 1.0608, + "mean_token_accuracy": 0.6852931380271912, + "num_tokens": 459351642.0, + "step": 18008 + }, + { + "epoch": 1.9777070063694269, + "grad_norm": 2.144313335418701, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6889455914497375, + "num_tokens": 459381540.0, + "step": 18009 + }, + { + "epoch": 1.9778168240720404, + "grad_norm": 2.378767251968384, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.72120600938797, + "num_tokens": 459404746.0, + "step": 18010 + }, + { + "epoch": 1.977926641774654, + "grad_norm": 2.239335060119629, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7142503261566162, + "num_tokens": 459430385.0, + "step": 18011 + }, + { + "epoch": 1.9780364594772677, + "grad_norm": 2.382493257522583, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7202267050743103, + "num_tokens": 459453352.0, + "step": 18012 + }, + { + "epoch": 1.9781462771798815, + "grad_norm": 2.244631767272949, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7303952574729919, + "num_tokens": 459476423.0, + "step": 18013 + }, + { + "epoch": 1.9782560948824952, + "grad_norm": 2.373506546020508, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.712550699710846, + "num_tokens": 459497969.0, + "step": 18014 + }, + { + "epoch": 1.9783659125851087, + "grad_norm": 2.275672674179077, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7103444337844849, + "num_tokens": 459524003.0, + "step": 18015 + }, + { + "epoch": 1.9784757302877223, + "grad_norm": 2.4865641593933105, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7137230634689331, + "num_tokens": 459544735.0, + "step": 18016 + }, + { + "epoch": 1.978585547990336, + "grad_norm": 2.314208507537842, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7074477076530457, + "num_tokens": 459571247.0, + "step": 18017 + }, + { + "epoch": 1.9786953656929498, + "grad_norm": 2.011136293411255, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.726498007774353, + "num_tokens": 459599216.0, + "step": 18018 + }, + { + "epoch": 1.9788051833955633, + "grad_norm": 2.1486096382141113, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7081136107444763, + "num_tokens": 459627258.0, + "step": 18019 + }, + { + "epoch": 1.9789150010981769, + "grad_norm": 2.275193214416504, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7465092539787292, + "num_tokens": 459650881.0, + "step": 18020 + }, + { + "epoch": 1.9790248188007906, + "grad_norm": 2.0935659408569336, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7359347343444824, + "num_tokens": 459678185.0, + "step": 18021 + }, + { + "epoch": 1.9791346365034044, + "grad_norm": 2.1683781147003174, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7134369015693665, + "num_tokens": 459705390.0, + "step": 18022 + }, + { + "epoch": 1.9792444542060181, + "grad_norm": 2.2143466472625732, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7259479761123657, + "num_tokens": 459729839.0, + "step": 18023 + }, + { + "epoch": 1.9793542719086317, + "grad_norm": 2.573760747909546, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7202880382537842, + "num_tokens": 459750598.0, + "step": 18024 + }, + { + "epoch": 1.9794640896112452, + "grad_norm": 2.2099132537841797, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7211602926254272, + "num_tokens": 459776190.0, + "step": 18025 + }, + { + "epoch": 1.979573907313859, + "grad_norm": 2.2595772743225098, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7006319761276245, + "num_tokens": 459802257.0, + "step": 18026 + }, + { + "epoch": 1.9796837250164727, + "grad_norm": 2.424119234085083, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7296468019485474, + "num_tokens": 459824919.0, + "step": 18027 + }, + { + "epoch": 1.9797935427190865, + "grad_norm": 2.112072229385376, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7212901711463928, + "num_tokens": 459853017.0, + "step": 18028 + }, + { + "epoch": 1.9799033604217, + "grad_norm": 2.390097141265869, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.7580262422561646, + "num_tokens": 459875002.0, + "step": 18029 + }, + { + "epoch": 1.9800131781243135, + "grad_norm": 2.143019437789917, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.75071120262146, + "num_tokens": 459901014.0, + "step": 18030 + }, + { + "epoch": 1.9801229958269273, + "grad_norm": 2.60329270362854, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.736552357673645, + "num_tokens": 459922385.0, + "step": 18031 + }, + { + "epoch": 1.980232813529541, + "grad_norm": 2.2121355533599854, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7199897766113281, + "num_tokens": 459947687.0, + "step": 18032 + }, + { + "epoch": 1.9803426312321546, + "grad_norm": 2.3219046592712402, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7261143326759338, + "num_tokens": 459970608.0, + "step": 18033 + }, + { + "epoch": 1.9804524489347681, + "grad_norm": 2.3263049125671387, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7378607988357544, + "num_tokens": 459993708.0, + "step": 18034 + }, + { + "epoch": 1.9805622666373819, + "grad_norm": 2.453587770462036, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7094199657440186, + "num_tokens": 460015653.0, + "step": 18035 + }, + { + "epoch": 1.9806720843399956, + "grad_norm": 2.3175675868988037, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7045831084251404, + "num_tokens": 460042408.0, + "step": 18036 + }, + { + "epoch": 1.9807819020426094, + "grad_norm": 2.37974214553833, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7178734540939331, + "num_tokens": 460066618.0, + "step": 18037 + }, + { + "epoch": 1.980891719745223, + "grad_norm": 2.2322261333465576, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.6936057209968567, + "num_tokens": 460094647.0, + "step": 18038 + }, + { + "epoch": 1.9810015374478365, + "grad_norm": 2.2948694229125977, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7232260704040527, + "num_tokens": 460119706.0, + "step": 18039 + }, + { + "epoch": 1.9811113551504502, + "grad_norm": 2.3193633556365967, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7201393246650696, + "num_tokens": 460143377.0, + "step": 18040 + }, + { + "epoch": 1.981221172853064, + "grad_norm": 2.130063056945801, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7144689559936523, + "num_tokens": 460172101.0, + "step": 18041 + }, + { + "epoch": 1.9813309905556777, + "grad_norm": 2.396388530731201, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7453887462615967, + "num_tokens": 460194279.0, + "step": 18042 + }, + { + "epoch": 1.9814408082582913, + "grad_norm": 2.3278582096099854, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7382925152778625, + "num_tokens": 460219038.0, + "step": 18043 + }, + { + "epoch": 1.9815506259609048, + "grad_norm": 2.572436571121216, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7134107351303101, + "num_tokens": 460240110.0, + "step": 18044 + }, + { + "epoch": 1.9816604436635186, + "grad_norm": 2.2544050216674805, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7142415046691895, + "num_tokens": 460265498.0, + "step": 18045 + }, + { + "epoch": 1.9817702613661323, + "grad_norm": 2.123331308364868, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7131171226501465, + "num_tokens": 460292455.0, + "step": 18046 + }, + { + "epoch": 1.9818800790687459, + "grad_norm": 2.144909143447876, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7303013205528259, + "num_tokens": 460319735.0, + "step": 18047 + }, + { + "epoch": 1.9819898967713594, + "grad_norm": 2.0807065963745117, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7299699187278748, + "num_tokens": 460346604.0, + "step": 18048 + }, + { + "epoch": 1.9820997144739732, + "grad_norm": 2.0468623638153076, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.6955205202102661, + "num_tokens": 460377497.0, + "step": 18049 + }, + { + "epoch": 1.982209532176587, + "grad_norm": 2.3529720306396484, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7377791404724121, + "num_tokens": 460400395.0, + "step": 18050 + }, + { + "epoch": 1.9823193498792007, + "grad_norm": 1.9893490076065063, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7110302448272705, + "num_tokens": 460431339.0, + "step": 18051 + }, + { + "epoch": 1.9824291675818142, + "grad_norm": 2.4196088314056396, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7149959206581116, + "num_tokens": 460454208.0, + "step": 18052 + }, + { + "epoch": 1.9825389852844277, + "grad_norm": 2.4019339084625244, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.730301022529602, + "num_tokens": 460476649.0, + "step": 18053 + }, + { + "epoch": 1.9826488029870415, + "grad_norm": 2.2360219955444336, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7362725734710693, + "num_tokens": 460500871.0, + "step": 18054 + }, + { + "epoch": 1.9827586206896552, + "grad_norm": 2.0531015396118164, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7067710161209106, + "num_tokens": 460529967.0, + "step": 18055 + }, + { + "epoch": 1.9828684383922688, + "grad_norm": 2.1913299560546875, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7197390198707581, + "num_tokens": 460556313.0, + "step": 18056 + }, + { + "epoch": 1.9829782560948825, + "grad_norm": 2.279839038848877, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7118855714797974, + "num_tokens": 460581587.0, + "step": 18057 + }, + { + "epoch": 1.983088073797496, + "grad_norm": 2.2036945819854736, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7363879680633545, + "num_tokens": 460607281.0, + "step": 18058 + }, + { + "epoch": 1.9831978915001098, + "grad_norm": 2.184115171432495, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7060250043869019, + "num_tokens": 460635224.0, + "step": 18059 + }, + { + "epoch": 1.9833077092027236, + "grad_norm": 2.421192169189453, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7175006866455078, + "num_tokens": 460657284.0, + "step": 18060 + }, + { + "epoch": 1.9834175269053371, + "grad_norm": 1.9595801830291748, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7140433192253113, + "num_tokens": 460690539.0, + "step": 18061 + }, + { + "epoch": 1.9835273446079507, + "grad_norm": 2.0143063068389893, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7218403816223145, + "num_tokens": 460722456.0, + "step": 18062 + }, + { + "epoch": 1.9836371623105644, + "grad_norm": 2.553985357284546, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7086143493652344, + "num_tokens": 460745042.0, + "step": 18063 + }, + { + "epoch": 1.9837469800131782, + "grad_norm": 2.1703145503997803, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.7431002259254456, + "num_tokens": 460770460.0, + "step": 18064 + }, + { + "epoch": 1.983856797715792, + "grad_norm": 2.133075475692749, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7208921313285828, + "num_tokens": 460797065.0, + "step": 18065 + }, + { + "epoch": 1.9839666154184055, + "grad_norm": 2.212948799133301, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7273029685020447, + "num_tokens": 460821423.0, + "step": 18066 + }, + { + "epoch": 1.984076433121019, + "grad_norm": 2.2230165004730225, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.723362147808075, + "num_tokens": 460845825.0, + "step": 18067 + }, + { + "epoch": 1.9841862508236328, + "grad_norm": 2.1663737297058105, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.7634082436561584, + "num_tokens": 460868917.0, + "step": 18068 + }, + { + "epoch": 1.9842960685262465, + "grad_norm": 2.6234419345855713, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7267133593559265, + "num_tokens": 460889975.0, + "step": 18069 + }, + { + "epoch": 1.98440588622886, + "grad_norm": 2.1950392723083496, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7439066171646118, + "num_tokens": 460915244.0, + "step": 18070 + }, + { + "epoch": 1.9845157039314738, + "grad_norm": 2.34281849861145, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.7059546113014221, + "num_tokens": 460940819.0, + "step": 18071 + }, + { + "epoch": 1.9846255216340873, + "grad_norm": 1.8567532300949097, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7030439972877502, + "num_tokens": 460976004.0, + "step": 18072 + }, + { + "epoch": 1.984735339336701, + "grad_norm": 2.457061290740967, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7213754653930664, + "num_tokens": 460999236.0, + "step": 18073 + }, + { + "epoch": 1.9848451570393149, + "grad_norm": 2.130307197570801, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7214361429214478, + "num_tokens": 461026568.0, + "step": 18074 + }, + { + "epoch": 1.9849549747419284, + "grad_norm": 2.5749385356903076, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7318643927574158, + "num_tokens": 461046201.0, + "step": 18075 + }, + { + "epoch": 1.985064792444542, + "grad_norm": 2.8521194458007812, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7609311938285828, + "num_tokens": 461062506.0, + "step": 18076 + }, + { + "epoch": 1.9851746101471557, + "grad_norm": 2.237478494644165, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7328321933746338, + "num_tokens": 461087640.0, + "step": 18077 + }, + { + "epoch": 1.9852844278497694, + "grad_norm": 2.329275131225586, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7201751470565796, + "num_tokens": 461112175.0, + "step": 18078 + }, + { + "epoch": 1.9853942455523832, + "grad_norm": 2.1807286739349365, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7207127809524536, + "num_tokens": 461139979.0, + "step": 18079 + }, + { + "epoch": 1.9855040632549967, + "grad_norm": 2.3045942783355713, + "learning_rate": 1e-06, + "loss": 0.8118, + "mean_token_accuracy": 0.7407172918319702, + "num_tokens": 461163656.0, + "step": 18080 + }, + { + "epoch": 1.9856138809576103, + "grad_norm": 2.3308753967285156, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7150574922561646, + "num_tokens": 461191175.0, + "step": 18081 + }, + { + "epoch": 1.985723698660224, + "grad_norm": 2.5019969940185547, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7094504833221436, + "num_tokens": 461214456.0, + "step": 18082 + }, + { + "epoch": 1.9858335163628378, + "grad_norm": 2.0884807109832764, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7121589183807373, + "num_tokens": 461244717.0, + "step": 18083 + }, + { + "epoch": 1.9859433340654513, + "grad_norm": 2.2356021404266357, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7378779053688049, + "num_tokens": 461268676.0, + "step": 18084 + }, + { + "epoch": 1.9860531517680649, + "grad_norm": 2.075028896331787, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7083591222763062, + "num_tokens": 461297889.0, + "step": 18085 + }, + { + "epoch": 1.9861629694706786, + "grad_norm": 2.2498912811279297, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7359148859977722, + "num_tokens": 461321986.0, + "step": 18086 + }, + { + "epoch": 1.9862727871732924, + "grad_norm": 2.3751332759857178, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7470794916152954, + "num_tokens": 461345605.0, + "step": 18087 + }, + { + "epoch": 1.9863826048759061, + "grad_norm": 2.1580007076263428, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6882046461105347, + "num_tokens": 461372344.0, + "step": 18088 + }, + { + "epoch": 1.9864924225785197, + "grad_norm": 2.3971383571624756, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7271270155906677, + "num_tokens": 461395089.0, + "step": 18089 + }, + { + "epoch": 1.9866022402811332, + "grad_norm": 2.458566665649414, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7250487804412842, + "num_tokens": 461417248.0, + "step": 18090 + }, + { + "epoch": 1.986712057983747, + "grad_norm": 1.922792673110962, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6934514045715332, + "num_tokens": 461449423.0, + "step": 18091 + }, + { + "epoch": 1.9868218756863607, + "grad_norm": 2.257220506668091, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7411494255065918, + "num_tokens": 461473859.0, + "step": 18092 + }, + { + "epoch": 1.9869316933889745, + "grad_norm": 2.2230799198150635, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.718698263168335, + "num_tokens": 461500365.0, + "step": 18093 + }, + { + "epoch": 1.987041511091588, + "grad_norm": 2.001384735107422, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7135895490646362, + "num_tokens": 461531744.0, + "step": 18094 + }, + { + "epoch": 1.9871513287942015, + "grad_norm": 2.305246114730835, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7187108993530273, + "num_tokens": 461556748.0, + "step": 18095 + }, + { + "epoch": 1.9872611464968153, + "grad_norm": 2.360987424850464, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7078652381896973, + "num_tokens": 461579723.0, + "step": 18096 + }, + { + "epoch": 1.987370964199429, + "grad_norm": 2.4313430786132812, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7265533208847046, + "num_tokens": 461602961.0, + "step": 18097 + }, + { + "epoch": 1.9874807819020426, + "grad_norm": 2.3379569053649902, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7062010169029236, + "num_tokens": 461626193.0, + "step": 18098 + }, + { + "epoch": 1.9875905996046561, + "grad_norm": 2.146695137023926, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7200589776039124, + "num_tokens": 461651625.0, + "step": 18099 + }, + { + "epoch": 1.9877004173072699, + "grad_norm": 2.1217329502105713, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7139887809753418, + "num_tokens": 461682023.0, + "step": 18100 + }, + { + "epoch": 1.9878102350098836, + "grad_norm": 2.1079578399658203, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7195017337799072, + "num_tokens": 461710729.0, + "step": 18101 + }, + { + "epoch": 1.9879200527124974, + "grad_norm": 2.3309056758880615, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7618831396102905, + "num_tokens": 461732452.0, + "step": 18102 + }, + { + "epoch": 1.988029870415111, + "grad_norm": 2.070592164993286, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7171016335487366, + "num_tokens": 461762810.0, + "step": 18103 + }, + { + "epoch": 1.9881396881177245, + "grad_norm": 2.130392074584961, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7272341847419739, + "num_tokens": 461790017.0, + "step": 18104 + }, + { + "epoch": 1.9882495058203382, + "grad_norm": 2.2370474338531494, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.7577821612358093, + "num_tokens": 461814656.0, + "step": 18105 + }, + { + "epoch": 1.988359323522952, + "grad_norm": 2.253183126449585, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7388694882392883, + "num_tokens": 461839072.0, + "step": 18106 + }, + { + "epoch": 1.9884691412255655, + "grad_norm": 2.5595545768737793, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7073956727981567, + "num_tokens": 461860450.0, + "step": 18107 + }, + { + "epoch": 1.9885789589281793, + "grad_norm": 2.1747727394104004, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.734018862247467, + "num_tokens": 461885369.0, + "step": 18108 + }, + { + "epoch": 1.9886887766307928, + "grad_norm": 2.4952454566955566, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7366594076156616, + "num_tokens": 461904353.0, + "step": 18109 + }, + { + "epoch": 1.9887985943334066, + "grad_norm": 2.242476224899292, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7249302268028259, + "num_tokens": 461935559.0, + "step": 18110 + }, + { + "epoch": 1.9889084120360203, + "grad_norm": 2.4401967525482178, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.7538837790489197, + "num_tokens": 461955442.0, + "step": 18111 + }, + { + "epoch": 1.9890182297386338, + "grad_norm": 2.1589269638061523, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7068199515342712, + "num_tokens": 461981664.0, + "step": 18112 + }, + { + "epoch": 1.9891280474412474, + "grad_norm": 1.9026358127593994, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7231074571609497, + "num_tokens": 462014362.0, + "step": 18113 + }, + { + "epoch": 1.9892378651438611, + "grad_norm": 2.2414984703063965, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7101105451583862, + "num_tokens": 462037936.0, + "step": 18114 + }, + { + "epoch": 1.989347682846475, + "grad_norm": 2.5584051609039307, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7219727635383606, + "num_tokens": 462058575.0, + "step": 18115 + }, + { + "epoch": 1.9894575005490887, + "grad_norm": 2.466832160949707, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7264996767044067, + "num_tokens": 462080727.0, + "step": 18116 + }, + { + "epoch": 1.9895673182517022, + "grad_norm": 1.882602572441101, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.703436553478241, + "num_tokens": 462115583.0, + "step": 18117 + }, + { + "epoch": 1.9896771359543157, + "grad_norm": 2.7155721187591553, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7347066402435303, + "num_tokens": 462134082.0, + "step": 18118 + }, + { + "epoch": 1.9897869536569295, + "grad_norm": 2.213085651397705, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7356807589530945, + "num_tokens": 462159283.0, + "step": 18119 + }, + { + "epoch": 1.9898967713595432, + "grad_norm": 2.0231330394744873, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.7004801034927368, + "num_tokens": 462189049.0, + "step": 18120 + }, + { + "epoch": 1.9900065890621568, + "grad_norm": 2.437533140182495, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7297862768173218, + "num_tokens": 462208873.0, + "step": 18121 + }, + { + "epoch": 1.9901164067647705, + "grad_norm": 2.1680963039398193, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.71628737449646, + "num_tokens": 462235720.0, + "step": 18122 + }, + { + "epoch": 1.990226224467384, + "grad_norm": 2.2203869819641113, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7352193593978882, + "num_tokens": 462261823.0, + "step": 18123 + }, + { + "epoch": 1.9903360421699978, + "grad_norm": 1.957722544670105, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.723334789276123, + "num_tokens": 462292726.0, + "step": 18124 + }, + { + "epoch": 1.9904458598726116, + "grad_norm": 2.3974409103393555, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7365946769714355, + "num_tokens": 462314728.0, + "step": 18125 + }, + { + "epoch": 1.9905556775752251, + "grad_norm": 2.264230966567993, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7439358234405518, + "num_tokens": 462338109.0, + "step": 18126 + }, + { + "epoch": 1.9906654952778386, + "grad_norm": 2.206444501876831, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7140367031097412, + "num_tokens": 462364224.0, + "step": 18127 + }, + { + "epoch": 1.9907753129804524, + "grad_norm": 2.310515880584717, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7153486013412476, + "num_tokens": 462388059.0, + "step": 18128 + }, + { + "epoch": 1.9908851306830662, + "grad_norm": 2.1219241619110107, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7147384285926819, + "num_tokens": 462418155.0, + "step": 18129 + }, + { + "epoch": 1.99099494838568, + "grad_norm": 2.332108736038208, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7127293944358826, + "num_tokens": 462442918.0, + "step": 18130 + }, + { + "epoch": 1.9911047660882935, + "grad_norm": 2.364318609237671, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7252321243286133, + "num_tokens": 462465659.0, + "step": 18131 + }, + { + "epoch": 1.991214583790907, + "grad_norm": 2.2292072772979736, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7122021913528442, + "num_tokens": 462491313.0, + "step": 18132 + }, + { + "epoch": 1.9913244014935207, + "grad_norm": 2.514111280441284, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7595850229263306, + "num_tokens": 462510185.0, + "step": 18133 + }, + { + "epoch": 1.9914342191961345, + "grad_norm": 2.1927952766418457, + "learning_rate": 1e-06, + "loss": 0.8069, + "mean_token_accuracy": 0.739739179611206, + "num_tokens": 462535217.0, + "step": 18134 + }, + { + "epoch": 1.991544036898748, + "grad_norm": 2.2501983642578125, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7097395658493042, + "num_tokens": 462560205.0, + "step": 18135 + }, + { + "epoch": 1.9916538546013618, + "grad_norm": 2.534926176071167, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7522300481796265, + "num_tokens": 462582027.0, + "step": 18136 + }, + { + "epoch": 1.9917636723039753, + "grad_norm": 2.3264358043670654, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.691288948059082, + "num_tokens": 462607357.0, + "step": 18137 + }, + { + "epoch": 1.991873490006589, + "grad_norm": 2.297625780105591, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7381366491317749, + "num_tokens": 462630066.0, + "step": 18138 + }, + { + "epoch": 1.9919833077092028, + "grad_norm": 2.106166362762451, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7096407413482666, + "num_tokens": 462658957.0, + "step": 18139 + }, + { + "epoch": 1.9920931254118164, + "grad_norm": 2.1021485328674316, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.708334743976593, + "num_tokens": 462688948.0, + "step": 18140 + }, + { + "epoch": 1.99220294311443, + "grad_norm": 1.935058832168579, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6983624696731567, + "num_tokens": 462722077.0, + "step": 18141 + }, + { + "epoch": 1.9923127608170437, + "grad_norm": 2.071382761001587, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.6925394535064697, + "num_tokens": 462751169.0, + "step": 18142 + }, + { + "epoch": 1.9924225785196574, + "grad_norm": 2.314180850982666, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7377673387527466, + "num_tokens": 462774139.0, + "step": 18143 + }, + { + "epoch": 1.9925323962222712, + "grad_norm": 2.206389904022217, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7333123683929443, + "num_tokens": 462799226.0, + "step": 18144 + }, + { + "epoch": 1.9926422139248847, + "grad_norm": 2.169887065887451, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7290310263633728, + "num_tokens": 462825185.0, + "step": 18145 + }, + { + "epoch": 1.9927520316274983, + "grad_norm": 2.0601773262023926, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.7016404271125793, + "num_tokens": 462857228.0, + "step": 18146 + }, + { + "epoch": 1.992861849330112, + "grad_norm": 2.5495822429656982, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7192080020904541, + "num_tokens": 462878454.0, + "step": 18147 + }, + { + "epoch": 1.9929716670327258, + "grad_norm": 2.2905547618865967, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7071512341499329, + "num_tokens": 462903109.0, + "step": 18148 + }, + { + "epoch": 1.9930814847353393, + "grad_norm": 2.106266975402832, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7050553560256958, + "num_tokens": 462932798.0, + "step": 18149 + }, + { + "epoch": 1.9931913024379528, + "grad_norm": 2.0365405082702637, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7094894647598267, + "num_tokens": 462963048.0, + "step": 18150 + }, + { + "epoch": 1.9933011201405666, + "grad_norm": 2.1489884853363037, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7387329339981079, + "num_tokens": 462987980.0, + "step": 18151 + }, + { + "epoch": 1.9934109378431804, + "grad_norm": 1.984578013420105, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7281929850578308, + "num_tokens": 463015831.0, + "step": 18152 + }, + { + "epoch": 1.993520755545794, + "grad_norm": 1.931199550628662, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.709822416305542, + "num_tokens": 463047545.0, + "step": 18153 + }, + { + "epoch": 1.9936305732484076, + "grad_norm": 2.0287764072418213, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.697097897529602, + "num_tokens": 463077868.0, + "step": 18154 + }, + { + "epoch": 1.9937403909510212, + "grad_norm": 2.4149651527404785, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.722474217414856, + "num_tokens": 463101686.0, + "step": 18155 + }, + { + "epoch": 1.993850208653635, + "grad_norm": 2.2916135787963867, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.7504990100860596, + "num_tokens": 463124383.0, + "step": 18156 + }, + { + "epoch": 1.9939600263562487, + "grad_norm": 2.19419002532959, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7117776870727539, + "num_tokens": 463153293.0, + "step": 18157 + }, + { + "epoch": 1.9940698440588625, + "grad_norm": 2.522779703140259, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7309221029281616, + "num_tokens": 463174100.0, + "step": 18158 + }, + { + "epoch": 1.994179661761476, + "grad_norm": 2.1705374717712402, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.741892397403717, + "num_tokens": 463199207.0, + "step": 18159 + }, + { + "epoch": 1.9942894794640895, + "grad_norm": 2.1515917778015137, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7229589223861694, + "num_tokens": 463225647.0, + "step": 18160 + }, + { + "epoch": 1.9943992971667033, + "grad_norm": 2.460242509841919, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.7068727612495422, + "num_tokens": 463248591.0, + "step": 18161 + }, + { + "epoch": 1.994509114869317, + "grad_norm": 2.4438984394073486, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7347379326820374, + "num_tokens": 463270647.0, + "step": 18162 + }, + { + "epoch": 1.9946189325719306, + "grad_norm": 2.2555716037750244, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7224246263504028, + "num_tokens": 463295056.0, + "step": 18163 + }, + { + "epoch": 1.994728750274544, + "grad_norm": 2.44404673576355, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.730999231338501, + "num_tokens": 463317738.0, + "step": 18164 + }, + { + "epoch": 1.9948385679771579, + "grad_norm": 2.524005174636841, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7424827814102173, + "num_tokens": 463338767.0, + "step": 18165 + }, + { + "epoch": 1.9949483856797716, + "grad_norm": 2.462280511856079, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7166335582733154, + "num_tokens": 463360839.0, + "step": 18166 + }, + { + "epoch": 1.9950582033823854, + "grad_norm": 2.138857126235962, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7168803811073303, + "num_tokens": 463387702.0, + "step": 18167 + }, + { + "epoch": 1.995168021084999, + "grad_norm": 2.1573948860168457, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7169797420501709, + "num_tokens": 463414719.0, + "step": 18168 + }, + { + "epoch": 1.9952778387876124, + "grad_norm": 2.2723634243011475, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7053663730621338, + "num_tokens": 463441482.0, + "step": 18169 + }, + { + "epoch": 1.9953876564902262, + "grad_norm": 1.9773811101913452, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7016023397445679, + "num_tokens": 463473259.0, + "step": 18170 + }, + { + "epoch": 1.99549747419284, + "grad_norm": 2.2064504623413086, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.737398087978363, + "num_tokens": 463499884.0, + "step": 18171 + }, + { + "epoch": 1.9956072918954535, + "grad_norm": 2.4225211143493652, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7346823811531067, + "num_tokens": 463520961.0, + "step": 18172 + }, + { + "epoch": 1.9957171095980673, + "grad_norm": 2.5514299869537354, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7361892461776733, + "num_tokens": 463542278.0, + "step": 18173 + }, + { + "epoch": 1.9958269273006808, + "grad_norm": 2.111232042312622, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7179455161094666, + "num_tokens": 463569755.0, + "step": 18174 + }, + { + "epoch": 1.9959367450032945, + "grad_norm": 2.731532096862793, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7170052528381348, + "num_tokens": 463595669.0, + "step": 18175 + }, + { + "epoch": 1.9960465627059083, + "grad_norm": 2.0954463481903076, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7148138284683228, + "num_tokens": 463627083.0, + "step": 18176 + }, + { + "epoch": 1.9961563804085218, + "grad_norm": 2.360142230987549, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7344731092453003, + "num_tokens": 463650767.0, + "step": 18177 + }, + { + "epoch": 1.9962661981111354, + "grad_norm": 2.22995924949646, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7239352464675903, + "num_tokens": 463674940.0, + "step": 18178 + }, + { + "epoch": 1.9963760158137491, + "grad_norm": 2.060253143310547, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6914510726928711, + "num_tokens": 463704304.0, + "step": 18179 + }, + { + "epoch": 1.9964858335163629, + "grad_norm": 2.1889381408691406, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7136930823326111, + "num_tokens": 463730197.0, + "step": 18180 + }, + { + "epoch": 1.9965956512189766, + "grad_norm": 2.120612859725952, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.6997660398483276, + "num_tokens": 463758078.0, + "step": 18181 + }, + { + "epoch": 1.9967054689215902, + "grad_norm": 2.039128303527832, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7097117900848389, + "num_tokens": 463786838.0, + "step": 18182 + }, + { + "epoch": 1.9968152866242037, + "grad_norm": 1.9821776151657104, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7306845188140869, + "num_tokens": 463816751.0, + "step": 18183 + }, + { + "epoch": 1.9969251043268175, + "grad_norm": 2.401190757751465, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7219390869140625, + "num_tokens": 463838738.0, + "step": 18184 + }, + { + "epoch": 1.9970349220294312, + "grad_norm": 2.478757619857788, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7131286859512329, + "num_tokens": 463860403.0, + "step": 18185 + }, + { + "epoch": 1.9971447397320448, + "grad_norm": 2.095926284790039, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7117470502853394, + "num_tokens": 463888313.0, + "step": 18186 + }, + { + "epoch": 1.9972545574346585, + "grad_norm": 2.0160555839538574, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7226675748825073, + "num_tokens": 463920015.0, + "step": 18187 + }, + { + "epoch": 1.997364375137272, + "grad_norm": 2.645399570465088, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7333117127418518, + "num_tokens": 463940593.0, + "step": 18188 + }, + { + "epoch": 1.9974741928398858, + "grad_norm": 2.4010801315307617, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7337814569473267, + "num_tokens": 463964365.0, + "step": 18189 + }, + { + "epoch": 1.9975840105424996, + "grad_norm": 2.2442610263824463, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7272544503211975, + "num_tokens": 463990574.0, + "step": 18190 + }, + { + "epoch": 1.997693828245113, + "grad_norm": 2.103174924850464, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.6981201767921448, + "num_tokens": 464019810.0, + "step": 18191 + }, + { + "epoch": 1.9978036459477266, + "grad_norm": 2.4901421070098877, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7231727838516235, + "num_tokens": 464040979.0, + "step": 18192 + }, + { + "epoch": 1.9979134636503404, + "grad_norm": 2.138777494430542, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7039773464202881, + "num_tokens": 464069530.0, + "step": 18193 + }, + { + "epoch": 1.9980232813529542, + "grad_norm": 2.1096408367156982, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.6920592784881592, + "num_tokens": 464099143.0, + "step": 18194 + }, + { + "epoch": 1.998133099055568, + "grad_norm": 2.2281479835510254, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7373606562614441, + "num_tokens": 464123933.0, + "step": 18195 + }, + { + "epoch": 1.9982429167581814, + "grad_norm": 2.027662515640259, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7098946571350098, + "num_tokens": 464153995.0, + "step": 18196 + }, + { + "epoch": 1.998352734460795, + "grad_norm": 2.43349027633667, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7150027751922607, + "num_tokens": 464177708.0, + "step": 18197 + }, + { + "epoch": 1.9984625521634087, + "grad_norm": 2.3983452320098877, + "learning_rate": 1e-06, + "loss": 0.7794, + "mean_token_accuracy": 0.74601149559021, + "num_tokens": 464200271.0, + "step": 18198 + }, + { + "epoch": 1.9985723698660225, + "grad_norm": 2.3161392211914062, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7328349351882935, + "num_tokens": 464223681.0, + "step": 18199 + }, + { + "epoch": 1.998682187568636, + "grad_norm": 2.541292667388916, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7140886783599854, + "num_tokens": 464244840.0, + "step": 18200 + }, + { + "epoch": 1.9987920052712496, + "grad_norm": 2.2650206089019775, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7407348155975342, + "num_tokens": 464268242.0, + "step": 18201 + }, + { + "epoch": 1.9989018229738633, + "grad_norm": 2.200789451599121, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.7011599540710449, + "num_tokens": 464297924.0, + "step": 18202 + }, + { + "epoch": 1.999011640676477, + "grad_norm": 2.399899482727051, + "learning_rate": 1e-06, + "loss": 0.7582, + "mean_token_accuracy": 0.7550626397132874, + "num_tokens": 464319485.0, + "step": 18203 + }, + { + "epoch": 1.9991214583790908, + "grad_norm": 2.4816906452178955, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.7473132610321045, + "num_tokens": 464340204.0, + "step": 18204 + }, + { + "epoch": 1.9992312760817044, + "grad_norm": 2.266871213912964, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.724032998085022, + "num_tokens": 464364539.0, + "step": 18205 + }, + { + "epoch": 1.999341093784318, + "grad_norm": 2.3171286582946777, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.7027261853218079, + "num_tokens": 464391464.0, + "step": 18206 + }, + { + "epoch": 1.9994509114869317, + "grad_norm": 2.4048819541931152, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7336130738258362, + "num_tokens": 464414705.0, + "step": 18207 + }, + { + "epoch": 1.9995607291895454, + "grad_norm": 2.2105541229248047, + "learning_rate": 1e-06, + "loss": 0.8, + "mean_token_accuracy": 0.7493746280670166, + "num_tokens": 464439378.0, + "step": 18208 + }, + { + "epoch": 1.9996705468921592, + "grad_norm": 1.8789410591125488, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.715895414352417, + "num_tokens": 464473367.0, + "step": 18209 + }, + { + "epoch": 1.9997803645947727, + "grad_norm": 2.4452617168426514, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7410985231399536, + "num_tokens": 464495349.0, + "step": 18210 + }, + { + "epoch": 1.9998901822973862, + "grad_norm": 2.2885584831237793, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7221116423606873, + "num_tokens": 464521560.0, + "step": 18211 + }, + { + "epoch": 2.0, + "grad_norm": 2.194023370742798, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7477123737335205, + "num_tokens": 464545414.0, + "step": 18212 + }, + { + "epoch": 2.0, + "step": 18212, + "total_flos": 2.091827002923077e+19, + "train_loss": 0.9512690712848968, + "train_runtime": 21456.929, + "train_samples_per_second": 13.58, + "train_steps_per_second": 0.849 + } + ], + "logging_steps": 1, + "max_steps": 18212, + "num_input_tokens_seen": 0, + "num_train_epochs": 2, + "save_steps": 9106, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 2.091827002923077e+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..aadaf27 --- /dev/null +++ b/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db6e1e768846a27b8766eaa47f5bf974da84c0a5c181e2a069e5a87f2423e5f9 +size 13329